WFManagement

Overcoming WFM Professional Defeat & Disengagement

noreply@blogger.com (Doug Casterton) — Sun, 12 Sep 2021 12:48:00 +0000

Reading Time: 3 mins 22 secs

Let me set the scene; The truth is on your side, yet no one will trust you. Your data is all there. You have spent a great deal of time preparing the right charts. You have even calculated standard deviations and correlations to back your findings... every angle has been considered. Yet, still, there is that one stakeholder who just says, "Nope, I don't think that's correct."

So, you stomp back to your desk in righteous indignation with the knowledge that yet again, you will be proven correct and then maybe next time they will listen. But, instead, the question you should be asking yourself is, "what did I do wrong?"

It is the responsibility of a WFM professional to ensure that plans reflect data-driven realities at all times, to be aware of the operation, and to deliver exceptional results for customers, employees, and organisations alike, even if that means calling out things that aren't going well.

However, many workforce management teams assume they are in a battle with internal partners over adopting offered suggestions. However, this defeatism stems from their inability to understand why anyone would want to work in a reactive environment. Any WFM professional knows this leads to high employee turnover, low employee satisfaction, low customer satisfaction, lack of first contact resolution, and higher than necessary business costs.

So, how do you become an essential part of the solution while resisting the need to withdraw because you believe your efforts to run a proactive operation are being disregarded by those in charge?

The bottom line is that if you fail to get others to believe in your workforce management output, then you must do a better job at selling. Yes, it is true; when you first started in workforce management, you likely had no idea how much a creative, clearly defined sales skill set you would need. Still, the ability to tell stories with your findings and to "sell" a plan and its collection of recommendations are what distinguishes the best Workforce Manage experts from the rest.

So here are some tips to selling your plan & recommendations that I have picked up over the years.

No Surprise Culture - Let folks know in advance what you have found. It may feel good to use statistics to strike fear into the hearts of others, but it rarely results in the desired effect of obtaining buy-in. Rather, pre-distribute your findings to the group, beginning with those you feel likely will be the most resistant to the results. Then, providing them with the opportunity to come up with solutions can very much help with buy-in.

Leave the inner geek at home - Using the most fundamental visuals and terminology will go you much further than using a statistically significant scatter plot with correlations and R-squares. Use the 4eye principle on your work before presenting it; if that person can't follow and understand what your selected visualisations are telling you, it's pretty likely that your stakeholders won't either.

Tone and delivery - Another area where workforce management teams frequently make mistakes is tone and delivery. Starting a conversation with "and here's what I discovered about after call work!" is likely to be received less well versus "After I was asked to look at handle time trends, I dug a little deeper and discovered some interesting trends with after call work."

Engage rather than accuse - After you've provided the facts, be cautious about drawing conclusions about the root cause (unless you have unequivocal evidence) or giving off-the-cuff ideas about how they should conduct their work unless you have substantial knowledge in this area. Nothing annoys a contact centre manager more than when their WFM team informs them that they don't coach correctly and then tells them how to solve it. Simply describe the facts and then engage them in a discussion about what they believe is the core reason and what actions they think will bring the results back into line.

Make it less risky When someone decides to buy into what you're selling, their mind immediately starts seeking reasons not to. So don't sit back and wait for the inevitable defences: Instead, anticipate challenges and objections ahead of time and be prepared to respond convincingly. For example, if the stakeholder's criticism is "we tried this before, and it didn't work," be ready to explain how your plan/idea/recommendation is significantly different. Take note of the specific variables and circumstances that make it more likely to succeed.

Create Momentum and trial close - Throughout your presentation of your findings, solicit feedback often. The best thing is that if you keep checking for agreement, your stakeholders will often want to choose or at least change the talk to action-oriented at the conclusion. If you aren't getting those buying signals, that's fine; just ask for the following stages. Do this by summarising your presentation and asking a few closing questions to confirm that your stakeholder agrees it's feasible. "Does this make sense to you?" for example. If you gain agreement, you'll have the go-ahead to proceed further.

5 Reasons why WFM delivers CX

noreply@blogger.com (Doug Casterton) — Sun, 05 Sep 2021 16:07:00 +0000

Reading Time: 4 Minutes

For too long, Workforce Management's primary outcome has only been seen as the driver of efficiencies at scale. While this is undoubtedly true, organization, planning, and well-trained personnel are also required to provide a memorable customer experience, and this is where Workforce Management methodology can also shine.

Workforce Management has a significant impact on Customer Experience, and in my opinion, there are six broad categories.

The classic: maintain adequate staffing levels, so customers never have to wait.
Customers today expect fast responses without going through multiple agents, channels, or departments.
Your customers can never be any happier than your employees.
Real-Time action drives increases in training and coaching opportunities, particularly when paired with Real-Time automation software.
Incorporating big data into WFM strategy can improve human interactions and customer experiences.

Let's Break it down.

The classic: maintain adequate staffing levels, so customers never have to wait.

Most operations will see spikes in contact volume throughout the year. So whether you've just launched a new product, are running a special promotion, or anything else, workforce management methodology ensures that adequate staffing levels are in place so that customers never have to wait to speak with a representative.

Customers today expect fast responses without going through multiple agents, channels, or departments.

Workforce management will frequently divide the workforce into broad skill groups, grouping agents who do not all have distinct skills together. A complex skill with a low contact volume, for example, is often not worth training everyone - which I refer to as a skill cylinder. It's critical to train a sufficient number of agents in this skill cycle to ensure that these contacts don't just abandon

Workforce Management also plays a vital role in reducing transfer rates. While a skill-based routing engine can methodically match a customer query with the best-suited agent, this is only effective if an agent is available, hence why workforce management is critical.

Your customers can never be any happier than your employees.

Many studies have discovered a link between employee and customer satisfaction. For example, according to Gallup's "State of the American Workplace" survey, happy employees lead to satisfied customers.

Agents are prepared for success and happiness long before they sit down to handle customer contacts. This occurs when they leave for work. It includes whether or not the agent had a say in when they began working, which days they could take off, and when they could take breaks. Workforce Management has a significant influence on all of these elements.

Employees who have a say in the fundamentals of their work environment are more likely to feel empowered than those who simply show up at a predetermined time because they are afraid of losing their job and income. They can also be more focused at work because they can schedule their time around family or social obligations. As a result, they will have a good workflow and will not see work interruption in their lives. For some tips on how to provide greater engagement via your scheduling practice, here are two articles worth reading SelfScheduling & EquityScheduling.

Real-Time action drives increases in training and coaching opportunities, particularly when paired with Real-Time automation software.

On one side, poor Grade of Service and High Abandon Rates lead to poor customer experience. However, unless your customer operation requires a highly high Grade of Service, such as an emergency service, you will generate a log of idle time for little customer experience gains. On the other hand, an excellent real-time team will proactively take this opportunity to increase coaching and training time, improving customer experience over time as Agents become more proficient at their jobs. Couple this with a real-time automation solution, and you can this to whole new levels.

Incorporating big data into WFM strategy can improve human interactions and customer experiences.

How do organizations identify what is truly important and then tailor it to empower employees and inspire customer loyalty? While digitalization has made life easier by automating essential business processes, it has also resulted in a constantly flooded world with data. Controlling data – a critical source of Customer Experience intelligence – effectively and dynamically is the key to success. The beauty of WFM methodology is its proven ability to collect essential data in one place, process it, and analyze it in real-time to drive the customer experience.

Workforce management can collect and share data with the HR department to identify the skills needed to meet corporate objectives and develop effective recruitment and succession plans. Then, use quality evaluations and analytics to detect coaching needs and schedule training at less busy times. WFM can also provide easy access to a wealth of historical data to quickly predict, create, and search future headcount and specific skill requirements. This is especially useful when downsizing during organizational change or to accommodate future growth plans.

The Next Big Thing to improve a BPO Bill to Pay Ratio

noreply@blogger.com (Doug Casterton) — Sun, 22 Aug 2021 12:03:00 +0000

Reading Time: 5 minutes.

The Business Process Outsourcing (BPO) industry aims to deliver outstanding customer service while also making a profit. This is no simple undertaking, and it demands careful cost control and continuous refinement to do more with less.

Any BPO wants to enhance their bill-to-pay ratio (the difference between the amount they bill a customer and the amount they pay their staff per hour), most BPOs tend to have razor-thin margins - it's just the nature of the business. Therefore, any improvement can be a competitive advantage.

To do this, BPOs strive to optimise the number of contacts or productive minutes split by their hourly cost, whether the contract is a cost-per-contact or cost-per-productive minute, to ensure that the majority of the time it pays its Agent is billable.

This means that each time an agent is off or waiting for the next contact is considered as an impediment to the bill-to-pay ratio's efficiency.

So what is the standard practice, and what are the challenges faced?

When billable volume is unavailable, the most competitive BPOs provide employees unpaid lunch breaks or discretionary time off, which is unproductive and wastes time for real-time analysts who must make scheduling exceptions.

Investing in training and coaching is another way to improve the bill-to-pay ratio. However,

while idle time in a contact centre might account for a large percentage of an agent's day, it is frequently divided up into small "chunks" - 20 seconds here, 40 seconds there - that are insufficiently useful for training or coaching purposes. So the usual method for a contact centre is to invest in employee development is to add additional time (often referred to as shrinkage) to the workforce management plan staffing required to allow time for training and to coach their people.

Finally, another method to improvise billable hours in a cost-per-contact contract set-up is to redistribute agents across multiple channels to complete billable chat and email work during natural idle times. In this way, the BPO can ultimately increase its billing-to-pay ratio by adding billable time to an idle hour or minute for which it is already paying. The challenge is usually the time it takes to redeploy an agent to a different channel, and again, idle time is often 20 seconds here, 40 seconds there. Additionally, act too slow, and you might start to miss service level on top of wasted productivity.

Enter Real-Time Intra-Day Automation

With Real-Time intraday automation technology, idle time can be aggregated and broken into larger "chunks," allowing for overtime reductions or greater time allocated to discretionary unpaid time-off or training/coaching, thus resulting in a decrease in Shrinkage over-staffing.

This is accomplished by continually re-forecasting agent demand in real-time based on actual customer arrival and how long each task takes and tracking which agents have shown up for work in real-time—then using this constantly revised accent requirement automatically adapting the WFM shift plan to find the idle time necessary to give employees their entire working hours in every schedule created. This idle time is then put to good use by automatically scheduling additional training, reducing backlogs, or work that would be done with overtime. Shrinkage in the Workforce Management plan can then be decreased without sacrificing customer service performance.

Even Better

Every major contact centre has a significant but often unquantified administrative overhead - the cost of managing and approving schedule changes due to delays, illness, shift swaps, and training requirements. Real-Time intraday automation technology can automate this overhead without sacrificing customer service.

In addition, you can create real-time agility at speed not possible with a human only real-time team. For example, the delay caused by a backlog in processing and approving change requests cause a lot of frustration for team leaders and agents alike, sometimes even doing more harm than good to service level or productivity. Real-Time intraday automation technology enables self-service change requests, increasing satisfaction through improved work flexibility and instant clarity.

Back in 2018, I dipped my toe into Real-Time Automation technology as an early adopter. At the time, the technology still had a way to go to be truly effective, but I was seriously impressed with the potential, and I am convinced this is the future. Whilst this article was written with a BPO in mind, the potential value of a client-side operation is equally compelling. Here is a video of me speaking about some of the successes I had with this technology - please don't judge my early attempt at speaking... it was back in a time when I still had a full head of hair.

Final Thought

A word of warning to those thinking of deploying real-time automation technology. Before relying on a toolset like this, I would highly encourage the organizations to ensure they have done as much as possible with human-based processes. Not every decision point is mathematical in the contact centre world, so retaining that human element is also very important. That said, WFM automation can generate significant benefits with its ability to analyze and act on operating conditions much faster than a human can. So the potential to extract just-in-time training or coaching opportunities from seemingly immeasurable pockets of low occupancy in real-time across large agents is quite compelling. The critical thing to remember is that the business rules used to drive this prescribed benefit need to be very clearly thought through, tested, and socialized with all staff. It's one thing to grab those low occupancy pockets, but it demands the operations team to have prebuilt the coaching or training modules you want the automation to kick off. All of that takes a lot of preparation, care, and feeding (nothing worse than an outdated business rule driving activity, thus resulting in it being more disruptive than value add).

Workforce Management (WFM) Best Practices

noreply@blogger.com (Doug Casterton) — Sun, 08 Aug 2021 15:33:00 +0000

Reading Time: 3 minutes.

One of the most exciting but challenging things about having a career in workforce management was that there was very little out there that could help me learn the job when I first started out. As a result, most people fall into the position, not knowing much about it and are forced to learn on the job by experiment. More recently, professional bodies like The Forum, GWFM and the SWPP have been excellent in providing content, seminars, & accreditation that have provided workforce management professionals with a foundation that has made learning the job much easier today.

That said, there will be multiple times during our lifetime career where we will find ourselves in a new role with a new company. This is the perfect opportunity to ask 'stupid' questions, listen with a purpose, understand the environment from a place of "where they have been, where they are, and where they want to go", and the #1 rule ~ assume nothing. Another significant part about starting a new job is that you get the chance to apply all those quick wins you have learnt elsewhere. I have found myself switching companies several times, and I have yet to find one company that has not highly customised parts of their workforce management process. Some even had ways of working that have been in place since the beginning, despite not making sense for the modern contact centre.

Time pressures are often the most significant barrier. Very few workforce management teams are not flat out busy. In fact, I would challenge any workforce management team that is not busy to ask themselves whether they are innovative enough. Culture and mindset also has a lot of say, with the words we use to describe others' practices reflecting whether this is a deep-rooted belief or just an occasional one-off. For example, when referring to best practices, are terms such as transfer, replicate, emulate, duplicate, translate, benchmark, roll out, propagate, or disseminate used? Or are words like employ, imitate, evangelise, clone, copy, and shamelessly steal instead used?

Whilst starting out in a new job is a perfect opportunity to apply the best practice you have learnt elsewhere, it does not have to stop there.

Capturing and reusing best practice knowledge is already widely identified as one of the most effective ways of taking any process or way of working to the next level, and you don't have to start out as a knowledgeable expert in that practice to benefit from it.

So what are the best ways of discovering and managing best practices in the context of workforce management .....

Identify Requirements - start by considering where one can really add value, looking at what areas need attention. Who will benefit the most from better knowledge and understanding of good practices? How will they access and use these?
Discover Best Practices - There are several ways to identify best practices. One way is to examine individuals and groups that deliver excellent results and thus are likely to use good practices. Having discovered these, next discern what parts of their overall approach or methodology represent good practice. This is best achieved by speaking to people within an organisation who have knowledge of the relevant procedure. This is where having a solid network comes in handy.
Document The Practices Good practice descriptions should be kept in a database and include the following:

A short title that can be accompanied by a brief abstract about the practice.
A number of short sections outlining what processes, functions, were applied.
Where was this practice applicable? What problems does did it solve?
What resources and skills were needed to carry out the practice?
What are the processes and steps involved? Are performance measures associated with the practice?
What proved difficult? What would the person you spoke to do differently if they were to do it again?
Any Tools and Techniques used in developing the practice.

How to Calculate FTE for Offline Queues?

noreply@blogger.com (Unknown) — Wed, 24 Feb 2021 09:31:00 +0000

In our Telegram Group, the most often asked question is "How to Calculate FTE for Offline Queue?" with certain TAT. Join our Telegram Group today by clicking here WFManagement Telegram

Today, in this article, I'll try my best to solve the problem with some practical illustration on how we can calculate the FTE requirement.

Let's start to explore!!

While we are trying to calculate the FTE for Offline Queue, the only thing which confuses us the most is the TAT which stands for Turn Around Time.

So, What is Turn Around Time and how is it decided?

Turn Around Time is the minimum or the Target time required to complete a task. The most simplest way to calculate the Actual TAT is Turn Around Time = Exit Time - Arrival Time.

During the initial stages, the Turn Around Time is decided using the Time & Motion Study.

Time & Motion Study is a systematic observation, analysis, and measurement of the separate steps in the performance of a specific task for the purpose of establishing Turn Around Time and the Productivity(explained at the later part of this article).

The Study also aims at improving the TAT and the Productivity.

Once a proper Time & Motion Study is established, the Turn Around Time is decided and it becomes a target to achieve.

Some of the examples of Turn Around Time for Offline Queues are as below

100% in 4 Hours
80% in 2 Hours
100% in 24 Hours
100% in 48 Hours etc.,

Now we have the Target Turn Around Time. The nest step is to Calculate how many FTEs are required to do the job.

I have attached a excel file at the end of this article which shows all the calculation methods explained below.

Now we have the Target Turn Around Time. The next step is to Calculate how many FTEs are required to do the job. Before that, let's look at some parameters required for the calculation.

Productivity

While doing the Time and Motion Study, the firm will also decided how many tasks can be processed by an agent at a specific time. For e.g. 20 tasks per day. This is called as Productivity.

Although, there are many other terms for this, we will refer to Productivity for this article.

Two major examples of Productivity are shown below

20 emails per Day
15 emails per hour

Now, the above Productivity can be converted to AHT with just a simple calculation.

20 emails per Day = (3600*8)/20 = 1440 Seconds
15 emails per hour = 3600/15 = 240 Seconds

At the same time, the AHT can also be converted to Productivity with the below calculation

1152 Seconds = (3600*8)/1152 = 25 emails per day
300 Seconds = 3600/300 = 12 emails per hour

The 8 is the shift length and if you want to change it, you can. Also if you wan to calculate AHT in Minutes, please use 60 instead of 3600.

Occupancy

Occupancy is the parameter where things get tricky. In Immediate Service Queues such as Inbound, Chat etc., the occupancy is an output derived from Erlang C. It is calculated as a weighted average of all the interval wise (15 mins or 30 mins or 60 mins) occupancies derived using the Volume, AHT and the Scheduled FTE.

But will it be the same for Offline Queues? I don't think so

If you look at the Excel Attachment below, I have considered Erlang to calculate what would be the ideal Occupancy for 100% in 2 Hours TAT. Please refer to "Erlang Method - ASA" tab.

I have used the ASA formula to find the agents required for 15000 transactions spread across a week in an interval level with an AHT of 300 Secs. The Macros is edited a little to accommodate for the Interval Length.

As you can see, from the Excel file, the occupancy is coming at 94.86%. Is this a good occupancy to run at? Not at all. The agents would end up burning out which leads to higher attrition.

If your TAT is around 80% or 70% in 2 hours or 4 hours, you can use the SLA formula to find the agents required. Please refer to "Erlang Method - SLA" tab.

As you can see, from the Excel file, the occupancy is coming at 96.47% even for SLA scenario which is definitely very high.

Therefore for Offline Queues, it is always suggested to go with standard occupancy of 85% which will be ideal.

You can change the TAT, Volume or AHT in the file to see the change in Occupancy for yourself.

The Final Step!!

FTE Calculation

There are multiple ways to calculate FTE for Offline queue and I've tried my best to shown as many ways as possible. Refer to the Tab "Workload Calculation" in the excel file.

Method 1 - Using AHT

This is a simple Workload method where the Volume, AHT in Secs and Occupancy of 85% is used to calculate the FTE Required without Shrinkage.

Method 2 - Using Hourly Productivity

If an agent has a target of 12 emails per hour, that means we would know the total email an agent can complete in a week just by multiplying 12 times 40 hours times occupancy of 85%. By this calculation, we found that an agent can complete 408 transactions a week. Therefore we would need 37 agents to do 15000 transactions (i.e. 15000/408).

Method 3 - Using Daily Productivity

If an agent has a target of 96 emails per day, that means we would know the total email an agent can complete in a week just by multiplying 96 times 5 days times occupancy of 85%. By this calculation, we found that an agent can complete 408 transactions a week. Therefore we would need 37 agents to do 15000 transactions (i.e. 15000/408).

Now what if the productivity which was calculated using the Time and Motion Study already had Occupancy baked into it?

As told earlier, the Time and Motion study is a systematic analysis of a particular task. While studying the productivity, the firm may have already accounted for all the idle time in between, Therefore we don't have to additionally add the occupancy again.

However this has to confirmed by analysing the study if we have the source or by asking the right person how the productivity target was determined.

Let's see how to calculate FTE when the occupancy is already a part of AHT

Method 4 - Using AHT

This is a simple Workload method where the Volume and AHT in Secs is used to calculate the FTE Required without Shrinkage.

Method 5 - Using Hourly Productivity

If an agent has a target of 12 emails per hour, that means we would know the total email an agent can complete in a week just by multiplying 12 times 40 hours. By this calculation, we found that an agent can complete 480 transactions a week. Therefore we would need 31 agents to do 15000 transactions (i.e. 15000/480).

Method 6 - Using Daily Productivity

If an agent has a target of 96 emails per day, that means we would know the total email an agent can complete in a week just by multiplying 96 times 5 days. By this calculation, we found that an agent can complete 480 transactions a week. Therefore we would need 31 agents to do 15000 transactions (i.e. 15000/480).

Conclusion

To conclude, while we are calculating the FTE for Offline Queues, I would not take Erlang into consideration unless the TAT is 1 hour or less. I would always consider a standard occupancy of 85% provided if the same is being accepted by the Operations team too.

Click below to access the Excel file.

Offline FTE Calculation Sheet

Thank you for reading and stay tuned😊

Disclaimer: This article is purely my personal view and understanding, this doesn't depict any organisation's data.

Vinay Vasudevan

Schedule Inflexibility - Basic Introduction

noreply@blogger.com (Unknown) — Tue, 01 Dec 2020 19:04:00 +0000

In Scheduling, the term Schedule Inflexibility or Schedule Inflex is used as a measure of efficiency of the scheduling process and it seems to be always tricky to understand.

In this article, I will try my best to explain what is Schedule Inflex and how to calculate it.

So, what is Schedule Inflex?

Schedule Inflex sometimes also known as Shift Inflex or Schedule Wastage is a measure of additional staffing scheduled above the requirements due to certain constraints.

Note: The term Inflex stands for inflexibility and it should not be replaced with the word influx.

Let's say, the FTE requirement at 9:00am is 25 and the requirement at 9:30am is 30. We have a batch of agents whose shift starts at 9:00am.

To meet the requirement at 9:30 interval, we will schedule 30 people at 9:00am. This will result in having 5 excess FTE at 9:00am interval.

The excess FTE is technically wastage and unfortunately it can't be avoided due to the above constraint. This wastage is known as "Schedule Inflex".

There are various aspects or constraints which contributes to the increase in Schedule Inflex and some of them are mentioned below

Call Arrival Pattern

Call Arrival Pattern is the major reason for Schedule inflex due to uneven arrival of the call in the intervals. Let's say we are dealing with two different arrival patterns i.e. Normal Bell Curve and a M-Curve.

Though the volume and AHT remains the same for the day, we will have more inflex in the M-Curve comparative to the bell curve. This is due to the excess staff which will be in those intervals between the two curves. This is better shown using the image below

Labour law

Though this category may not contribute much to the schedule inflex, it is still considered as a constraint.

The best example would be the part timers. In the above M-Curve, if we had options to include part timers, the inflex may have been reduced. However in some countries, the concept of part timers is still not implemented which will create wastage.

Team based scheduling

In some companies, the scheduling are built team wise instead of agent wise, this may create excess staffing in some intervals which will lead to wastage.

Consecutive Week offs

This is yet another approach of certain companies to have a good employee satisfaction. While improving the employee satisfaction is good, it may result in having excess headcount in some intervals

Transport Constraints

In those companies, where the transport is provided by them, the shifts are predefined due to the transport. This leads in having excess headcount.

Now lets come to the major question. How to measure Schedule Inflex?

There are various methods to calculate the schedule inflex, however in this article we are going to look at most simplest yet very effective method to calculate the schedule inflex.

I have attached a excel file at the end which contains the calculation, please make sure to see that

The process of calculation is broken into steps for better understanding

Step 1 - The weekly call volume should be divided into Day wise and Interval wise using the past interval wise pattern.
Step 2 - Once we have the interval wise calls, we should use the Erlang Function to calculate the Interval wise requirements.
Step 3 - Against the interval wise requirement, the FTEs have to be scheduled and further divided into interval wise to compare the over/under.
Step 4 - Now, just by dividing the total FTEs scheduled and the Required FTE, the final Schedule Inflexibility is calculated.

From the excel file attached, the Schedule Inflex is computed as 118.32%(i.e. 22/19 = 118.32%). This Inflex can be used for future forecasted volume to find out the FTEs to be scheduled without actually doing a Scheduling simulation.

Here is how it can be done. Let's say the weekly call volume is now 8000 instead of 7000(volume considered in the excel file). With the same AHT, the new Required FTE would be 21.

For us to calculate the FTE to be scheduled, simply multiply the previously calculated Schedule Inflex with the new requirement i.e. 21*118.32% = 25(rounded). This way schedule inflex can be used in the planning process.

Hopefully, with the above explanation, the Schedule Inflexibility now seems to be not so tricky😝

Thank you for reading and stay tuned😊

Schedule Inflexibility Calculation Sheet

Disclaimer: This article is purely my personal view and understanding, this doesn't depict any organisation's data

Vinay Vasudevan

To Social Media or Not Social Media that is the question?

noreply@blogger.com (Doug Casterton) — Sun, 18 Oct 2020 17:59:00 +0000

Synchronous (CHAT) vs. asynchronous messaging (SMS, Social Media, WhatsApp)

What is synchronous messaging?

Synchronous messaging is suitable for simple conversations that can be solved in one sitting or with one agent.

What is Asynchronous messaging?

Asynchronous messaging is more convenient for customers and is also better for complex issues that require more than one sitting or one agent to fix. Good for (IT issues, Retention issues, Account problems, Adjustments, Complex Cx Issues)

Benefits to Social Media & Tips

• From my experience, Social Media has some of the highest FCR and NPS scores. Generally, upon 1st post by the customer, the agent has whatever TAR (Time to Average Response) the company sets to make an initial response within a set target (TAR) and then once the customer responds, the clock resets and the same (TAR) window now can be used to solve the customer’s issue. By the next response, the issue is generally resolved and the case can be closed. This requires minimal effort from the customer and generally is a much more positive experience for them

• Social Media requires less FTE to handles large volumes as (TAR) can be set based on customer’s habits, interaction type, service standards, SLA, or within an industry-standard target of about 15 mins. Since the customer can go about their day the (TAR) can be set to 20 or 30 mins to reduce FTE requirements in line of business that may be of less priority or that by its nature would require more time like a complex IT issue

• Stop and Go (asynchronous messaging) - Less Pressure on Agent to finish transaction real-time without customer holding (No AHT pressure) live. Customers can go about their day while their problem is being solved vs. Waiting on hold

• Natural and less stressful for agents (Agents are encouraged to be more themselves) when interacting with customer * Quality guidelines apply *

• Asynchronous messaging is also better for complex issues that require more than one sitting or one agent to fix

• Can easily integrate social media with self-serve options

• Asynchronous messaging is the new email: Email can be missed or go unopened within the companies SLA. Social media (Asynchronous messaging) has a target set and has to be responded to within a targeted TAR (time to average response)

• Social Media is best staffed with Top Tier or most experienced agents. Agents should have extensive access and authority to make changes and bring a customer account to a resolution. This will increase FCR and Boost NPS.

How to plan for Social Media (FTE and Forecasting) - High Level

In a nutshell... social media would work the same as SMS or WhatsApp. The initial contact from cx would start a conversation ... with multiple stop and go responses. SMS and WhatsApp would work the same way as social media... it’s a stop and go interaction. One a post/or message is received from a customer; it is used as a starting point ... and then generates an average amount of responses aka conversation rate. An average handle time can be summed up from all the responses in each unique conversation to forecast the same way as chat or voice

One then needs to then use conversation rate from the initial 1st contact to come up with (average conversation rate) and then multiply * conversations by * conversation rate to get (responses offered) and then sum up AHT

Using the response AHT you can plan at the interval / monthly / daily level on responses and avg AHT per response totals 

Also, the same cx may not respond in the same interval. At the interval level, you would then need to plan to whatever responses and AHT that happen in a 900/1800/3600 interval to get a FTE plan at hourly/30mins/15min level

Excel Formulas (Social Media) Planning

Responses = (Unique Conversations * Responses Per Conversation)

Responses Per Conversation = IFERROR (responses/conversations,0)

Social Media ASA = IFERROR (answertime/handled/60,0)

Social Media AHT = IFERROR (handle time/handled,0)

Social Media SL Calc.% = IFERROR (# of responses handled within (Target TAR)/handled,0)

Introduction to Call Routing Techniques

noreply@blogger.com (Unknown) — Sun, 04 Oct 2020 14:17:00 +0000

Connecting customers to the agents as quickly as possible is the first most important requirements of a contact center. Connecting them to the right agent is also equally important.

This can be done only through a better Call Routing Technology. Today, in this article we are going to look at different call routing techniques available for a contact center mainly for inbound queue.

The Call Routing System which is otherwise also known as Automatic Call Distributor (ACD) is a tool or a software which routes the customer call from IVR (Interactive Voice Response) to the right agent according to the pre-defined algorithm. The pre-defined algorithm defines the type of call routing system which is what we are going to discuss in this article.

The main benefit of using the right method of call routing is optimisation of human resource which can save a lot of money to an organisation. The identification of which call routing strategy will benefit your organisation depends upon the desired service level type, technical capabilities, cost to build the strategy etc.

How does Call Routing Works?

Usually, in contact centers, a Voice over Internet Protocol (VoIP) phone system is used along with Computer Telephony Integration (CTI) to create a IVR and Call Routing system.

In simple words, the customer calls the contact center, reaches the IVR, selects appropriate options in IVR and either the query is solved in IVR or the call gets connected to an agent.

Some of the information used to determine the routing techniques are:

Caller ID - Unique ID or the phone number of the customer
Time of the day – The time zone of both caller and the contact center
Location – The location of the caller
IVR – Pre-defined options which will help the caller to reach appropriate department
Self Service – Most of the basis queries are answered in IVR using the Artificial Intelligence
Agent Skills – Depending upon the options selected in IVR and depending on the customer profile, the call will be routed to appropriate agent.

Types of Call Routing Techniques

Most Idle/Least Occupied Routing

This is the most common type of Call Routing method where an agent with most idle time gets the call first at the time of call arrival. This type of call routing is used by the call center where the agents are universally trained on the process without any special callers. This method is also called Least Occupied technique since the agent with most idle time will have the least occupancy at the time of call arrival.

Skill Based Routing

The second most common call routing techniques is Skill Based Routing. In this method, agents are separated according the special skills. The special skills could be process based (e.g. Tier 1, Tier 2 etc.), language based (e.g. English, French, Spanish etc.) and many other skills depending upon the business requirements. When the IVR identifies the type of customer calling, the call will be routed accordingly to the right agent. For e.g. Customer may be Spanish speaker whose call will be directed to Spanish speaking agent.

Department Based Routing

In this routing method, the calls are directed to the respective department according to the customer’s selection in IVR. The departments could be Sales, Service, Billing, Technical support etc. When a customer calls the contact center, he/she has to go through the IVR to select some pre-defined options. According to the department selected, the calls will be routed to the right agent of that department.

Value Based Routing

Value based routing, also known as Smart Routing, is the technique of redirecting the call basis the customer value. Customer value can be details such as Demographics, previous action on the call, potential frustration or lower customer satisfaction. Data is analysed by the system according to the value of the customer and the call is directed or routed to the most appropriate agent. The Value Based Routing sometimes can also be referred as Business Rules Routing.

For e.g. A customer’s call who had given low score in the survey during the previous call may be directed to the escalation team directly instead of the front-line agent.

Geography Based Routing

This is a type of routing where the system identifies the caller location through GPS and the phone number of the customer. The call is then routed to the respective location. This type of routing works when a contact center has multiple branches to support all the location of the customer base.

Data Directed Routing

Data directed routing is an intelligent routing technique where the call might be routed to appropriate department depending upon the customer’s data. For e.g. a customer may call for a credit card query, but his bill may be overdue from a long time. This is identified by the system and the call is automatically routed to the collections department. This may also happen in case of frustrated customer whose call can be directly routed to escalation team, though the customer selected a different option in IVR

Round Robin Routing

Round Robin call routing is a technique where the call is distributed evenly across the agent one by one. When a call arrives from the IVR, the system identifies the agent who took the last call and then directs the call to the next agent. The next call then goes to the next agent. The order of agents who would take the call is predefined in the system. The major disadvantage of this method is uneven distribution of wait time as the caller might wait for that particular agent to answer.

Multimedia Routing

In today’s world, a customer may contact the call center using multiple media such as e-mail, chat etc. The Multimedia routing is a technique where the other media is converted to a call. This type of routing would be mainly Outbound. For e.g. A customer will send an email to the contact center and as soon as the email is received, an outbound call will be placed by the system and the agent will speak to customer.

Conclusion

The above-mentioned Call Routing techniques are some basic types which are still widely used in many contact centers.

Though we have seen different call routing methods above, sometimes the call may get misdirected to a wrong agent. The agent then would politely assist the customer by transferring the call to the right department. This may lead to customer dissatisfaction. This is the only major drawback of the call routing techniques which is inevitable.

However, with the recent development in Machine Learning and Artificial Intelligence, many advanced Call Routing Techniques are built which are proving to be really useful in optimising the operations of a contact center.

Thank you for Reading and Stay Tuned😊

Disclaimer: This article is purely my personal view and understanding, this doesn't depict any organisation data

Vinay Vasudevan

The impact on Email FTE Requirements following a change in Service Level Targets

noreply@blogger.com (Doug Casterton) — Tue, 29 Sep 2020 12:39:00 +0000

Below is an interesting extract taken from the WFManagement Telegram community group - if you have an interest in Workforce Management and are interested in joining here is link with further background and joining information WFManagement Telegram Group

I leave my thoughts at the end.....

---------------------------

Question: Is there an effect on required FTE number for Email channel, if the SLA target get changed? For example, suppose current SLA target (for Email only) is 90% within 4 hours. Provided all other matrices such as volume, AHT, delay, occupancy, arrival pattern, shrinkage remain unchanged, will the FTE requirement go down if SLA target revised at 70%? How much will be the change? Can anybody assist to explain the relationship between SLA target and FTE headcount for email and if there is any standard formula?

Reply 1: Whenever my guys have planned emails we’ve based it on a carry forward calculation rather than service level, e.g. less than x percentage of work brought forward + input carried forward into the next day / week. By handling volumes that way I’ve always found the service level as such looks after itself.

-------------------------------My Views--------------------------------

Changing your Email Service Level should not by itself in the long run change FTE requirements to meet the new service level target.

In its simplified form Productive FTE required for the email channel is based on "New Incoming + keeping backlog level stable".

There is no need to load workload for any form of Erlang given this is not a live channel unless you your new Service Level target is near live i.e. less than say 1 hour. Therefore changing the Email Service Level will mean you will have a short term requirement increase or decrease (depending on whether you are increasing/decreasing the SLA target) whilst you get the backlog level to the point that is balanced against the new Service Level Target, but otherwise required productive FTE requirements will return back to the same levels once backlogs are again balanced.

Live Contacts after the lines close - what to do....

noreply@blogger.com (Doug Casterton) — Sat, 26 Sep 2020 19:09:00 +0000

Question: Our call centre lines closes at 8pm (a recent change from 10pm) and every evening we have some calls in queue until 8.15. We have implemented scheduling advisors to work until 8.15 but this hasn’t gone down very well with our workforce who wants to finish at 8 as this creates issues with public transport. What do other call centres do with calls in queue or arriving close to when their lines close? Really interested to find out

Reply1 : We have cross-trained resources that are working normal hours at that precise time. Also, this could be done in the normal first come first served where whomever is waiting the longest on the queue will get the calls whether thats one of the dedicated resources or one of the cross-trained resources. Alternatively you can have it set up as an overflow strategy where the call will only get to the cross-trained resources if certain parameters happen (like only if there are agents available on the cross-trained team and if nobody in your dedicated team is available.

Reply 2: We utilised Supervisors to help ensure 1) queues were cleared and/or answer calls that were queued to clear faster and 2) all agents were signed off for the evening. Transportation is a unique challenge but maybe this solution could help.

---------------------My Views---------------------

Transport needs is an an often forgotten dynamic when it comes to Scheduling considerations, especially important for an out of town or poorly supported public transport site. Not only should the safety and welfare of your staff be a manor consideration when leaving the site at night, but knowing that if you don't leave the centre before you miss your only transport home is bound to impact customer experience as the agent rushes through the conversation to get out on time. Transport issues aside, having to work past your shift can be a real pain for agents, especially when you have Real-Time times constantly reminding them to not be late for your shift and ensure they achieve their schedule adherence targets.

I don't think there is a wrong or right way to handle this issue apart from perhaps ignoring it. If you have the option to utilise other sites with different opening hours then this is an obvious solution. The solution mentioned in the question of having people finish 10-15min (depending on how long the AHT is) after the lines close works well, but like all things setting expectations up front and gaining their buy-in by explaining WHY being there for the customer really add value to engagement. You can also identify those that have the biggest life-outside-of-work reason for needing to get away on time and through a routing solution have lines close 5-10mins before their shift finishes. For those that have to stay over to clear those lines, the least you can do is time bank the extra hours they work and then either give them that time back at some point in the future or pay it out as single time overtime (OT). I find accumulating minutes owned over a monthly period and then either paying it out as OT or providing them a separate PTO allowance works quite effectively.

Whatever you decide to do, like with all scheduling, engagement with Agents is key - set the expectation up front and explain the why and how they contribute to the customer experience and the brand by support this activity. Even better give the agent the problem statement and ask them to brainstorm solutions, this way they feel part of the solution and not done to.

The Forecaster’s Luck by Ger Koole

noreply@blogger.com (Unknown) — Wed, 26 Aug 2020 08:04:00 +0000

I am lucky not to be a forecaster… because they really have a difficult job. They have to differentiate between the effects of many factors, such as day-of-the-week, week-of-the-year, long-term trend, marketing events - without being able to tell exactly what the effect of each factor is. The reason for this is twofold: factors change over time, depending on the time of the year, changes in the call center or the company’s business, etc. But also because everything is obscured by a layer of noise. Think about it: if you forecast 30 calls in half an hour, then 30 customers won’t agree together to call one after another with exactly 1-minute intervals. Instead, there will be longer and shorter inter-arrival times and periods looking like clusters, like in the graph below. Mathematicians call this a Poisson process.

This noise has three important consequences. A forecaster can never do it completely right: a perfect forecast is purely luck, the noise will most of the time prevent the actuals from being exactly equal to the forecasts. To measure forecasting errors, we use the WAPE, the weighted absolute percentage error. It is the sum of the absolute interval errors divided by the total volume. For example: you forecast a volume of 100 for 4 consecutive quarters. The actuals are 90, 105, 100, 110. Then the WAPE is:

Some advanced mathematics gives a beautiful simple formula for the minimal WAPE:

In the figure below we plotted the minimal WAPE for different volumes. To give an example: for a forecast of 250 the minimal WAPE is √ 2 / (π 250) = 5%. Thus, as the minimal WAPE decreases with the volume, for a volume below 250 the accuracy of the forecast can never be below 5%!

Noise also makes forecasting more difficult. It obscures the signal hidden behind the noise. To achieve an accuracy as high as possible given the noise requires lots of data and a deep understanding of on the one hand the nature of call centers and on the other hand advanced statistics and machine learning.

But the noise can also work in the advantage of the forecaster: the forecasting errors are, to some extent at least, obscured by the noise. When the volume is low you won’t see a small forecasting error because the noise is much bigger. Of course, you cannot hide behind the noise when it is smaller than the error you made in the other components.

This raises the question: when do you see the error through the noise? And can we find the error simply by deducting the minimal WAPE from the observed WAPE? If the FC was 250 and the WAPE 12% was the forecasting error then 12 - 5 = 7%? Unfortunately reality is not that simple. To obtain some insight look at the figure below. It shows volume on the horizontal axis and WAPE on the vertical one, for different forecasting errors plotted at the right axis. The bottom red line is the minimal WAPE: the WAPE with a 0% forecasting error. For example, when the volume is 20, then the error by the Poisson noise is also around 20%.

Now let’s look at the other lines. They represent the WAPE for different levels of FC errors - the percentages are given on the right axis. When, again, the volume is 20 but you made a 20% FC error, i.e., you predicted 16 or 24, then the WAPE is around 25%. Thus you cannot simply add error and minimal WAPE: in this case 15% of the error is absorbed by the noise. On the other hand, when volumes and/or errors are high the WAPE is equal to the error: you do not see the noise, it disappears in the error.

You can also reverse this reasoning: when the volume was 100 and the WAPE around 10%, then the error can be anything up to say the minimal WAPE thus 10%, the lines close together. But when the error is significantly higher than the minimal WAPE, then the error is close to the WAPE, you cannot blame the noise.

To conclude: despite the noise, enough data plus advanced methods and understanding can give reliable forecasting - but up to a certain level, given by the minimal WAPE. For low volumes, forecasting errors are obscured by the noise, but for higher volumes the WAPE has to be completely contributed to the forecasting errors, the noise is too small to hide behind.

Ger is full professor at Vrije Universiteit Amsterdam and founder and chief scientist at CCmath.
To find out more about CCmath and its products, see www.CCmath.com.
To learn more about the background and practice of WFM, see WFMacademy.CCmath.com
The author thanks Giuseppe Catanese for his feedback.

The Maths of Adherence

noreply@blogger.com (Unknown) — Mon, 03 Aug 2020 11:41:00 +0000

--------------------------------------------------------------------------------------------------------------------------

Ger has been a wonderful contributor in the field of mathematics applied to contact centres, you can check out is other articles posted here in WFManagement via this link: Ger Koole other articles.

Ger provides with this article sound theoretical knowledge on the elements you should consider when devising a new schedule adherence target. Additionally, at the end of Ger's post I have included the schedule adherence target calculator I use when devising new schedule targets.

Happy Reading. Doug

--------------------------------------------------------------------------------------------------------------------------

Agent scheduling is neither the first nor the last step to make sure your agents do the right thing at the right moment. Before you need budget and capacity planning; after it you need real-time performance management. An important part of that is ‘schedule adherence’: making sure your agents do what they are supposed to do. However, 100% adherence to the schedule is impossible and undesirable. Evidently, when an agent is busy with a call the moment she is supposed to move to another activity there should be time to finish the call without having a bad adherence score. For this reason agents are not required to have a 100% adherence, but what is reasonable? How much time do you expect that an ongoing call will last when an agent is supposed to switch activity? Half the average handling seems like the obvious answer, but the mathematics show something else.

To understand the mathematics, let us discuss first the so-called ‘waiting-time paradox’. Consider a bus that arrives on average every 15 minutes at a bus stop. A passenger regularly takes the bus without looking at the schedule. What is her average waiting time? One would expect 7.5 minutes, but that is not the case. Let us consider the situation below. The inter-arrival times between buses are 10, 20, 25 and 5 minutes. On average 15 minutes, but different because of the traffic density, traffic lights, etc. A random arrival in an interval waits on average half the length of the interval. But a passenger is twice as likely to arrive in the 20-minute interval than in the 10-minute interval. The average waiting time is: 10/60 * 5 + 20/60 * 10 + 25/60 * 12.5 + 5/60 * 2.5 = 9.6 minutes, much longer than 7.5 minutes! And the more variable the inter-arrival times, the longer the average waiting time.

Back to call centers. An activity is more likely to end during a long call, just as a commuter has a higher chance to arrive in a long interval. How much time does an agent work on a call when she is supposed to move to another activity? We did some computations using representative data. In the figure below we see the histogram of the handling times and, next to it, the histogram of the residual time at the end of an activity.

Both have a shape typical for the so-called lognormal distribution, a long tail to the right. The residual times are more centred and thus less variable. The most important figures however are the means: 255 and 194 seconds. Thus the AHT is 4m15s, and the average residual time is 3m14s, considerably more than half the AHT! To conclude the mathematics: the common wisdom that an agent spends half the AHT on finishing a call at the end of an activity is false, it is, on average, considerably longer.

What are the learnings from this? First, that we should rethink what an acceptable level of adherence is. Or, better, that we should differentiate between forms of adherence: finishing a call should probably have no influence on an agent’s adherence score, but starting late after a break does.

But perhaps the most important learning is that randomness often has surprising consequences which can even be counter-intuitive. Deciding based on “common sense” can lead to adverse consequences for customer service, such as agents having to rush finishing a call. This feels unfair and gives the impression of management taking arbitrary measures. Understanding the maths behind the call center can lead to happier customers and employees!

Ger is full professor at Vrije Universiteit Amsterdam and founder and chief scientist at CCmath.
To find out more about CCmath and its products, see www.CCmath.com.
To know more about the waiting-time paradox, see en.wikipedia.org/wiki/Residual_time#Waiting_time_paradox.

The author thanks Wout Bakker and Bas Hoogink for their feedback.

--------------------------------------------------------------------------------------------------------------------------

Ger's articles provides hints for a formula that is more sophisticated, something I will be pondering over in the future. For a simple method for calculating schedule adherence targets, that takes into account the no of planned I suggest the following formula:

Here is a link to the calculator I use for devising new schedule adherence targets: Schedule Adherence Target Calculator

All the best. Doug

--------------------------------------------------------------------------------------------------------------------------

Introduction to Simple Forecasting Methods in Excel

noreply@blogger.com (Unknown) — Sun, 02 Aug 2020 11:44:00 +0000

Forecasting is one of the most important aspects in business. We use the techniques to anticipate the future.

In business, the forecasting is used to allocate the manpower, budget etc. Forecasting is so powerful that all the business planning relies on forecasting.

Fortunately, the contact center forecasting comes under WFM bucket and due to this, WFM is looked as a very important and pivotal department in an organisation.

There are numerous methods of forecasting available and each has its own calculations and usage.

Today, in this we are going to look at some simple forecasting methods along with their calculations. We will also look at how to select a best fit model depending upon the Forecast Accuracy/Error.

As always, I have attached a excel sheet at the end which contains detailed calculations of the Simple Forecasting Methods.

So, let's begin our quest!!

Naive Forecasting Method or Random Walk Method

The logic of Naive Forecasting Method is that the forecasted values will be equal to the previous period value. The Naive Method is also called as Random Walk Method.

For e.g., if we are forecasting for the month of January, the forecasted value will be equal to December. This is illustrated by a formula as shown below.

where, Yt is the period to be forecasted and Yt-1 is the previous period.

Seasonal Naive Forecasting Method or Seasonal Random Walk Method

This is a slightly improved method over the Naive to factor the seasonality. While the forecasters were using the Naive Method, they had a problem of the previous seasonality not getting factored in the forecast.

Therefore the Seasonal Naive Method was formulated.

Let us consider that we have to forecast for the month of January and we have the previous year forecast as historical. Now, instead of considering December as the forecasted value for January as we did in Naive method, we will consider the previous year January value for the current year January.

This is illustrated by a formula as shown below.

where, Yt is the period to be forecasted and Yt-m is the previous seasonal period.

Random Walk with Drift

We know that the Naive and Seasonal Naive are also called as Random Walk. In this method we use the Seasonal Random Walk, but with an additional component called drift.

For any historical data, we have certain forecast components such as Level, Trend and Seasonality. These are called the "Building blocks of Forecasting". I would urge the reader to clink on this link to read more on the Building blocks of Forecasting. Contact Center Basic Forecasting Building Blocks

Anyways, coming back to the drift. Since we are using Seasonal Random Walk, the seasonality part of the forecast is taken care of, the actual historical value acts as a Level.

Therefore the only pending component is the Trend. To factor this in our forecast, we use the additional drift component to our forecasting.

Let's take a look at the below formula for Random walk with Drift.

The formula may look at little overwhelming, but it is very easy when you do it in excel. Please refer to the attachment.

Moving Average Method

This is perhaps the most common method used in forecasting. As the name indicates, this is a technique where we consider the average of subset of previous data as the forecasted value.

As we move towards the forecast period, the subset also moves, therefore the name Moving Average.

The Moving Average is denoted by MA(n) and n stands for the period of the subset. We read it as Moving Average of Order n.

For e.g. if we have monthly call volume data for a year and we are trying to forecast the value for the next 4 months, we can use the average of the first four months of previous data for Month 1. When we move to Month 2, the average also moves to the next four month excluding the 1st month and including the 5th month. Thus the average move as we progress to our forecast.

In the excel file attached below, i have considered the subset to be 4, therefore the moving average is denoted as MA(4).

The key to get the best forecast accuracy is by changing the size of subset and finding the best size by doing a trail and error method.

Below is the formula for Moving Average.

Weighted Moving Average Method

In the moving average, we gave equal importance to all the values in the subset. However there are instances where the most recent historical value has some external factor to it such as Marketing, New Product etc.

In this scenario, we cannot give equal weights to all the value in a subset. Therefore we use the Weighted Moving Average Method.

The Weighted Moving Average is denoted by WMA(n) and n stands for the period of the subset. We read it as Weighted Moving Average of Order n.

The most recent historical value gets more weightage and the weightage reduces as we go further to the past. However all the weights added together should be equal to 1 or 100%.

In the excel file attached below, i have considered the subset to be 4 and the weights as 10%, 20%, 30% and 40%, therefore the weighted moving average is denoted as WMA(4).

The key to get the best forecast accuracy is by changing the size of subset and changing the weights and finding the best size and weights by doing a trail and error method.

How to find the best fit model?

We have looked at different simple forecasting models above, but how do we decide which model is the best fit for our data?

To understand this, we have to look at the different ways of looking at the forecasting error. Let us first look below at few terms which are required to calculate the forecast accuracy.

Error: The difference between the forecasted value and the actual value.

Absolute Error: The absolute difference between the forecasted value and the actual value.

Square of Error: The square of the Absolute Error.

Once we have understanding of the above terms, we will proceed to calculate three different types of forecasting error.

MAE (Mean Absolute Error): Average of the Absolute error for all the time period.
RMSE (Root Mean Square Error): Square Root of the Average of the Squared Error for all time period.
MAPE (Mean Absolute Percentage Error): Average of Absolute Error divided by the average of actual value for all the time period

With the example in the attached excel file, below are the values of MAE, RMSE and MAPE.

Just by looking at the above numbers, it is very obvious that for the given data set, the Seasonal Naive has the lowest error. Therefore the Seasonal Naive Method or Random Seasonal Naive is the best fit model for forecasting.

Note: These methods are tried and tested in R and Python

Attachment: Simple Forecasting Model

Thank you for Reading😊😊

Stay Tuned!!

Disclaimer: This article is purely my personal view and understanding, this doesn't depict any organisation data

Vinay Vasudevan

Contact Center Basic Forecasting Building Blocks

noreply@blogger.com (Doug Casterton) — Sun, 26 Jul 2020 16:36:00 +0000

The basic building blocks of a systematic forecasting approach (Systematic: Components of the time series with consistency or recurrence and can be described and modelled) can be based on relatively simple mathematical methods.

Daily, Weekly, Monthly Volume forecasting involves thinking of a series as a combination of level, trend, seasonality, and noise components in addition to the Poisson fluctuations that you can expect. These components for a customer operational forecast have been depicted above.

Simply put, to be able to predict the future, we need to first understand what has occurred previously and extrapolate that into the future. Thus, analysing the historical volume and explaining what has happened using these factors above are perhaps the most critical part of any forecasting process. Analysing the influence of these factors one by one is known as a decomposition method.

Making a graphical plot of the available data is often not that insightful due to the weekly cycle and special events that are likely to mask the identity of a slight but steady increase in contact volume. For example, we usually can quite clearly see the weekly cycle, with less traffic on Saturday and Sunday vs a weekday - often with Monday as the highest volume day. However, it is often difficult to identify all outliers to derive a trend. For this reason, we have to identify the special events and holidays and then deseasonalize the data. This should be done for the weekly cycle but can also be completed for the yearly cycle if enough data is available.

In a perfect world following deserializing, we would get a straight line representing the long-term trend, with special days and the seasons filtered out. Of course, if the line represented a flat profile after this effort, then there is no trend. However, we rarely get such a flat profile, and there will still be unexplained variations. Some of these will be caused by normal Poisson fluctuations, but often they are more significant than we would expect just from a Poisson distribution. Once you have identified the trend, you can start to assemble the forecasting.

Once the trend is determined, we can now include the seasons (both intra-week and intra-year) and then the special days to produce the forecast. While deseasonalizing consisted of dividing by the average per day or week, seasonalizing consists of multiplying the deseasonalized forecast by the seasonal factors. To correct the forecast for special days and events, we have to quantify their effect. This can be achieved via various methods, but even simple averaging can effective as a method if past events are likely to be similar to future ones.

So this concludes with a brief overview of a straightforward but effective forecasting method. It should be noted that many elements need to factor in, and they will vary from operation to operation. For example, things that are likely to differ from operation to operation are; the amount of available data, the volatility of that data, and the extent to which special days influence.

Thus, the only way to decide on what to include and what not to include is to test continually using a systematic approach of continuous improvement.

For further reading on the Contact Center Forecasting topic, check the following out:

Volume Demand Forecasting – introducing some more slightly more advanced topics

Time-Series Forecasting - some common methods used.

Forecasting Accuracy Targets - key considerations when looking to measure forecast accuracy

Methods to Measure Forecast Error - The first rule of forecasting is that all forecasts are either wrong or lucky.

Smoothing Spline - It is really possible to make intra-day patterns better and easier.

WFM considerations when launching new customer support channels

noreply@blogger.com (Doug Casterton) — Sat, 18 Jul 2020 16:32:00 +0000

The contact center has already evolved from the humble call center, as more channels of communication became commonplace. With the meteoric rise of Facebook Messenger, Whatsapp, Telegram, Viber, WeChat, Line, and IMO (you can read more about this here) things are only likely to continue getting more complicated.

There are many things to consider when thinking about implementing a new channel into a contact centre. This article aims to provide a view from the Workforce Management perspective. Additionally, with permission from Assembled (https://www.assembled.com/) I have included one of their videos which shares stories from their customers on how they've navigated the rollout of new customer channels and what they've learned from past experiences.

Current and Future Demand Size – economies of scale apply when planning for customer channels. For example, in a live channel you have to allow for random contact arrival (queuing theory) in order to meet service level. The larger your volume and workforce, the higher the probability is that an agent will be available in time to answer the customer contact within SLA. As a result, you can increase occupancy and thus also cost efficiency. On very small volume channels you might be facing the opposite problem, resulting in you applying minimum staffing rules not because of pure workload need but just because you need to cover the opening hours of the channel.

Channel Hopping – yes for every new channel you introduce you are likely to also introduce incremental volume as a result of customers switching between channels depending on how emotive or complex a conversation has become. This needs to be incorporated into any volume demand forecasting.

Technology Capability. What is your current technology’s ability to provide a seamless multi-channel customer experience. How will agent and customer switch between channel and what efficiencies or inefficiencies will this produce?

Data Needs - How will data be captured and how accessible will this be to meet reporting/insight needs. So often data needs are the last consideration and yet Data is the life blood of any workforce management plan.

Workforce Interchangeability - How interchangeable is the new channel with existing channels i.e. can resources be merge or will two separate teams be required for each channel.

Channel Assumptions – each channel type will require different supporting planning assumptions. Sometimes this might just be a different productivity level assumption, but a new channel type can also drive a requirement for a completely new assumption. For example, Concurrency for Live Chat versus Occupancy for Phone channels.

Highlights From Video

(0:00) Defining channel strategy principles

(0:42) Lessons from a channel launch that was rolled back

(2:36) Stripe adds chat support to email support

(5:05) Hacks to improve case routing for realtime channels

(7:15) Daily Harvest launches SMS and inbound phone support

(10:09) Managing social media in support

WFM for Outbound Planning

noreply@blogger.com (Unknown) — Sun, 12 Jul 2020 16:00:00 +0000

There are many contact centers which have the majority of their business that depends on the Outbound Campaign.

Especially when the contact center is involved in sales or collections, outbound calls are critical to success. There are many other reasons too for why outbound calls are required and it is totally dependant on the process.

Outbound calls can give us insights about the customer's need. It helps us to balance the sales and improve the brand awareness.

But, from a WFM planning standpoint, how to manage outbound?

Well, there are many ways to calculate the agents required for outbound campaigns.

In this blog, we will see few methods to calculate the No. of agents requires for Outbound calling using some KPIs which are generally used in the Outbound contact centers.

First, let's gather a few prerequisites for the calculations as shown below

No. of Contacts to be Dialled

This is mainly the database in which data will be fed into the dialer for the agents to dial.

This data should be considered as Volume. We need to understand that every customer in this database is equivalent to one volume, therefore if we have 1000 customers in the outbound database, that means the volume is 1000.

Successful and Non-Successful Average Handle Time

In an Outbound Planning two types of AHT comes into picture. i.e. Successful and Non-Successful AHT.

Now, when an agent makes an outbound call, not all the customers receive the call. Some may disconnect the call, some may not receive, some may have directed to voicemail etc. In short whenever an agent can't talk to the customer about the product, the call is considered to be Non-Successful.

Though the call was Non-Successful, an agent would have spent certain time on that call either listening to the ring or voicemail etc. That time spent is called Non-Successful AHT.

The inverse of this is the agent successfully speaking to customer, which is called as Successful AHT. As a industry standards, the Non-Successful AHT cannot be more than 60 secs.

We will look at how both the AHTs will affect the planning in sometime.

Type of Dialer which will be used

The Dialer plays a major role in the Outbound Contact Center, it is like a ACD which routes the call. There are different types of dialer available for a outbound contact center.

I had written a blog on different outbound dialer few weeks back. If you have not read it, i would strongly recommend to read so that you can understand this article completely.

Overview of Outbound Dailer

Turn Around Time

All the Outbound calls comes with certain Turn Around Time. This totally depends upon the type of outbound call, the cost factored and the Voice of Customer. Some examples of Outbound Turn Around Time is given below

- 100% calls to be made in 24 hours within the lead generation.

- 80% of the calls to be made within 4 Hours.

Connect Ratio

Now it is very common that all the calls made to the customer will not be successful. We ourselves would have disconnected the call from many marketing campaigns. Also some customers may have activated DND.

Therefore, the measurement of successfully placed calls against the total calls is called Connect Ratio.

For E.g. Out of 1000 Calls, only 700 calls were connected to customer and was tagged as successful, hence connect ratio is 700/1000 = 70%.

Number of Attempts

It is obvious that the contact center will not stop at a single attempts if the customer doesn't receive the call. There will be multiple attempts made to a single customer.

Earlier I had mentioned that each customer in the database is equal to one volume, and if we try multiple attempts to a customer, that means the volume is also increasing. This will have an impact on the FTE Required.

The total no. of attempts to be made is deciding by historical analysis, telecom laws and the type of outbound calls. Also there will be a specified time gap between each attempts.

Productivity

Productivity is a key components used in many deferred media planning such as Back Office, Email, Outbound etc.

The industrial standard is to consider around 85% and we will also do the same.

Now that we have most of the inputs required to calculate the FTE, let's go ahead and do it.

I have attached the Excel sheet at the end which has the calculations embedded. Please go through to get complete understanding.

In the attachment, there are two types of FTE Calculations with 4 models for each types. Both the types shows different way of Outbound calculation which is completely dependant on how the process is designed.

As mentioned earlier, there are many types of calculations for Outbound and i'm only showing two types for the illustration purposes.

The Models mentioned in the attachment represents the type of dialer as mentioned below.

Model 1 - Manual Dialer
Model 2 - Preview Dialer
Model 3 - Progressive Dialer
Model 4 - Predictive Dialer

Let's look at the calculations below.

Type 1 Calculations

Here we have used all the KPIs to calculate the FTE required. Note that the Non-Successful AHT is decreasing depending upon the type of dialer used.

The workload is divided into Successful and Non-Successful for better understanding. The Successful Workload is simple and straight forward where the overall volume is multiplied with the connect ratio and the successful AHT.

When we say connect ratio, it is no. of calls connected as explained earlier. This essentially means that the remaining 25% was not connected at all. Therefore they turn into Non-Successful workload.

However, if you observe the Non-Successful workload calculation, i have used the no.of attempts too. Theoretically if the total no. of attempts is 4, we assume that the 75% of connected customers take only the 1st attempt and the remaining 25% customer take the 3 attempts. therefor making the total attempts as 4.

So, to have the Non-Successful Workload, i have taken 25% of the volume with Non-Successful AHT and 3 remaining attempts.

Once we have both the Workload, we just sum it up and calculate the FTE required. In the FTE Required Calculations, i have divided the workload by 8 considering it to be shift length for a day.

Now based on the type of dialer used, the Non-Successful AHT reduces decreasing the overall FTE required.

Note: The shift length and the shrinkage are subjected to change.

Type 2 Calculations

What we saw in the Type 1 can be theoretically implemented for planning, but it majorly relies on the assumption that all the successful contacts happens in first attempt. But is it practically possible?

I don't think so, therefore I have also attached the Type 2 Calculations. Here, as you see in the table besides the calculation, there are attempts and the % of calls connected for each attempts.

Now we use the Daily volume and calculate the Successful Call and Non-Successful Calls. In the first attempt, when 25% of the calls are connected, the remaining 75% are dialled out. However in the second attempts, the connected calls are taken after removing the 25% which was already connected in the 1st attempt.

This bring a major difference in the workload. Since all the calls are dialled for every attempt, the Non-Successful workload is higher in the Type 2 than the Type 1, but removing the Successful calls of previous attempts reduces the Successful Workload.

Once we have both the Workload, we just sum it up and calculate the FTE required. In the FTE Required Calculations, i have divided the workload by 8 considering it to be shift length for a day.

Now based on the type of dialer used, the Non-Successful AHT reduces decreasing the overall FTE required.

If you observe both Type 1 and Type 2 calculations, the FTE required in Type 2 calculations seems to be higher.

Outbound FTE Calculations

Thank you for Reading:)

Stay Tuned!!

Disclaimer: This article is purely my personal view and understanding, this doesn't depict any organisation data

Vinay Vasudevan

The Challenges surrounding Flexibility by Prof. Dr. Ger Koole

noreply@blogger.com (Unknown) — Fri, 03 Jul 2020 12:00:00 +0000

Today, as I’m writing this blog, is the first really warm summer day in Holland. Every WFM professional knows that this means less calls which arrive earlier in the day (unless you’re in airco sales). This results in overstaffing, especially in the afternoon. To avoid this we need flexibility. Perhaps some agents want to leave early, but the contract should allow it. Or you have some back-office work to do or trainings that can be scheduled, or even calls from a country where the weather is different.

Flexibility has many uses and faces. It has to do with when agents work and what they do. And both long and short term… At the same time flexible contracts that allow you to schedule more hours for the retail peak of the holiday season, and multi-channel agents who can deal with short-term fluctuations that would otherwise make their occupancy too low.

But flexibility also has its downsides. Agents often need a guaranteed number of working hours to earn a decent living and have other obligations and preferences limiting their availability. Having too many skills leads to low proficiencies on some of the skills, leading to bad service and unhappy agents. And flexibility comes at a price: usually flexible contracts cost more per productive hour and training agents costs time and money.

Fortunately, we do not need full flexibility. For all forms of flexibility it holds that the first agents having this form of flexibility help a lot, but then the returns decrease. And because the costs for flexibility increase linearly, there is a sweet spot with the optimal amount of flexibility.

How much flexibility do you need? And what type of flexibility? A bit of everything, but how much is really hard to answer. It depends on many factors, such as the error of your forecasts, your routing, the labor rules, etc. To compute it you cannot simply use the Erlang formula, but you need long-term simulations taking all aspects of your call center into account. There are some rules of thumbs however. If your SL and your occupancy are good and consistent over the day, then you probably have enough flexibility built into your system, you might even be able to do with a little bit less. If your SL is good or even higher than required, and your occupancy is too low, then you’re overstaffed. If your SL and occupancy fluctuate a lot, around the required levels, then you probably lack flexibility, or you are not able to use the flexibility you already have. Finally, better forecasting makes for better capacity plans and schedules reducing the need for flexibility, and if you consistently need to scale up or down then you probably make a systematic error in your forecast or schedule, you are always too low or too high.

Flexibility is crucial to obtain your SL and keep costs under control, but it also comes at a cost… it makes the control of your call center more complicated. Not all scheduling tools take advantage of all possibilities, and also multi-skill multi-channel intra-day management is hard to do optimally. Highly skilled WFM professionals might be the answer, but they are very hard to find and costly again. And even when a smart mathematician was watching all metrics in real time she would not be able to react quickly enough to changes in load or oversee all consequences of removing one agent on all skills and channels. The answer? Software, but advanced enough that it makes good automatic decisions. Reaction is instantaneous and you only need a smart mathematician once. The foundation is good routing software which does most of intra-management automatically, no more need for intra-day changes in skill settings. But also smart scheduling software can deal automatically with requests for a day off. This frees valuable resources from day-to-day processes allowing them to work on what really matters: improving the design of all aspects of the call center.

Ger is full professor at Vrije Universiteit Amsterdam and founder and chief scientist at CCmath. He has been a guest author for WFManagement on a number of occasions, you can find his other posts here: Ger Koole Articles. To found out more about CCmath and its products, click here: CCmath.com. The author thanks Wout Bakker and Bas Hoogink for their feedback.

The Dreaded Camel Hump 🐫

noreply@blogger.com (Doug Casterton) — Mon, 29 Jun 2020 08:56:00 +0000

This is Re-Poste from Marlon Martinez - original post can be found here “How to plan for Web Chat in a call center 101"

Hello workforce world! We all have experienced or know about that dreaded call center camel hump lol… we still live in a world where the majority of the corporate world works and would like to work the standard wish list 9am-5pm Monday to Friday shifts J.. however, as businesses evolve as well as technology ... your hours of operations are expanding, and businesses need to be more flexible to meet their customer needs across multiple age generations like the baby boomers/generation X and Millennials.

I have worked in a few call centers and the arrival patterns of voices calls tend to be fairly static or typical and this is what causes that dreaded camel hump ...were you have to staff up your mornings heavily in order to meet that big morning spike in volume ... however too cover your ever expanding operating hours... you have to ensure a balance in your workforce to meet that same spike in call volume that occurs at night. Also, we all tend to like a continuous shift... we run into that “Dreaded camel hump problem” were you have a huge overlap is staffing and your early afternoon to late afternoon is so overstaffed and you are burning money??!!! Well what can you do about it????

Below are some good suggestions:

1. Split shifts

2. Voluntary early leave (VEL)

3. Workforce Optimization Software Suite

4. Alternative platforms like Social Media and Web Chat

5. Work from home agents ** 5G and easy to get jump on’s

Topics in detail

#1- Split shifts

Split shifts are a great way to reduce the overlap in a call center camel hump situation, however from my experience they are not highly desired and have very low uptake. So most organizations... like this idea... but are generally disappointed on how ineffective it is ... when only a few agents tend to pick or want them.

#2 - Voluntary early leave (VEL) / Leave without pay (LWP)

VEL and LWP is a very effective way to reduce costs and can be done by line of sight (queue monitoring) or a pricier but very effective option is to buy or add on within your existing (WF) software a (WFO) workforce optimization suite. A WFO suite is something wonderful and works amazing if the quality of your data and arrivals patterns is super accurate and updated consistently. If you do not have a budget for this ... you can accomplish the same via excel but for both solutions. Keep in mind “Garbage in and you will get garbage out” so updating your call arrival data, weekly call distribution and all other metrics and variables is !!Super!! important!

#3 - Workforce Optimization Software Suite! Pricey but they are very effective and offer amazing interactive options that increase employee satisfaction and engagement.

(WFO) suites offer many great features like apps that your agents can use to apply automatically for early or unpaid leave. These requests can either be manually approved by your intraday/real-time analysts and or rules you can customize that can be triggered automatically by the software suite based on your ACD data like service levels and predicted staffing surpluses or shortages. These suites can also offer targeted training during those slow times when agents are not applying for early leave to keep your occupancy high by reducing idle time. Again, with this solution it’s also only as effective as the data that feeds to it. It can be highly effective and reduce your needs for extra workforce manpower. This solution only works well if its setup correctly and if the data that feeds into it is super accurate and constantly re-trended.

# 3 (B) To expand on my earlier point ... you do not need an expensive WFO suit to optimize your call centre and reduce costs... but I still highly recommend one. If you can’t afford it ... you can accomplish the same so long as you constantly update call distribution patters, Call arrival patterns, and all other workforce metrics. If your input data is solid... then you can make excel sheets that can calculate the FTE Staffing you will need by interval and you can predict where you can offer early leave in advance. The best solution would be to create scheduled export files from your ACD (Automated call distributor) and have that data feed into a excel sheet via SQL scripts that can constantly update your weekly call distribution, Call arrival patterns, Shrink, etc. and with simple FTE requirements formulas .. you can calculate staffing needed per interval in excel and you can have your real-time intraday analyst’s use that to offer things like early leave or overtime or even a full shift off to reduce your staffing costs. Also if you do a morning or evening reforecast consistently and update your input variables your forecasting and FTE deltas will be more reliable and can be done accurately same day.

*** This is very important “Garbage data in and you get Garbage data out” !! trending should be consistent and done every 6 to 8 weeks at minimum or even tighter during peak season or major launches or email or bill cycles like 2-3 weeks and even daily. ****

Topic #4 – (Social Media and Web Chat Platforms) for your young more tech savvy millennial customer base

Social Media and Chat is a great way and a new alternative to target your millennial customer base. My experience shows that Social media is an evening heavy arrival pattern platform and it is a great way to redirect you voice call volume. It requires less of a workforce to handle these interactions .. as they can have a much longer or wider answer threshold. In the social media world, they call this (TAR) time to average response. You can set this to whatever you want... 10/15/30 minutes for someone to respond to a post within your public and private social media platforms like Facebook and Twitter and other emerging platforms. Social Media is a stop and go interaction experience…your millennial customer base has a narrow attention span J lol... they just want to send a message and are more than willing to wait for a response ... so long as they accomplish the same results as calling in. It also requires minimal effort on their end so they have more time to watch game of thrones or catch Pokémon’s J.

Web chat tends to follow the same arrivals patterns as voice calls but can gain you amazing efficiency if your customer database and transaction systems can support multiples sessions. ** This is very important ** before you invest into a chat platform … be sure to check that your legacy systems can handle multiple records searches and transactions. Web chat is also a very friendly Millennial and Generation X alternative that can dramatically reduce your call centre cost per call/interaction.

Topic #5 - Work from home agents.

This one is a great one folks…while it may be very expensive to setup, technology is evolving constantly ** Like 5G** and not only will it save you money on real-estate space ... but most agents will already have their own WIFI connections and or it can be provided for them as part of the role or at a discount or subsidy. The reason why work from home options are so great! You can get jumps on easily and engage your agents and offer overtime without the need for them to drive 1-2 hours just to get to the center to assist in a major spike or unforeseen demand in volume. The uptake rate on work from home agents is HUGE and has great potential to save you money over the long term (ROI) and can help kill the camel hump problem ..very effectively.

Chat Forecasting: Concurrent AHT

noreply@blogger.com (Doug Casterton) — Sun, 21 Jun 2020 13:09:00 +0000

This is Re-Poste from Marlon Martinez - original post can be found here “How to plan for Web Chat in a call center 101"

Let’s start you off with the correct way to calculate concurrent AHT from your single session times per dept at agent level. It’s not much different from voice planning ... other than coming up with concurrent AHT. It also important to understand your chat/dept mix as this is important for the distribution pattern at the interval level and your chat push setup! Below is the most important part of chat planning! The Concurrency Formula :

Important notes

You can not divide or multiply your volumes by a subjective concurrent target or estimated concurrent chat rate

You should not be guessing your concurrency rate! Or allow your TM’s or OP’s to guess how many concurrent chats are being handled by your agents! Below is the actual way to get concurrent rate ( cc rate) which will make your chat planning a breeze!

And YES you can still use Erlang C or X and WFM platforms that use Erlang to plan for chat.. as long as you properly use concurrent AHT and not single session AHT or do any of the above

Single session times must be divided by total logged in “ productive time engaged in chats” and all offline/load must be removed from equation

With voice lines you have lines or trunks, chats does not have this... as you can open as many sessions aka lines/trunks as your system can handle. It is important to set a chat dept. limit as this can destabilize your platform server. In most chat platforms you can set an open session limit for each chat department to simulate (trunk/lines) required in erlang C or X.

Formulas

“ (Total chat time + total Wrap up time) aka "Agent_WorkTime" /(total engaged in chats time aka"Agent_ChatTime")

"Agent_WorkTime" (HandleTime)

Chat time = Time spent on chat per session

Wrap up time = time spent offline wrapping up session

"Agent_ChatTime"

Engaged time = total hours agent is engaged in chats with customers

Notes

Also Concurrency is calc. by workload/WorkTime vs chat per interval .. as some chat may only take seconds to complete

Actual Formulas

.Platform_Concur =IFERROR(Agent_WorkTime/Agent_ChatTime,0) = Concurrent Rate

.Platform_ConcAHT =IFERROR('.Platform_AHT'/'.Platform_Concur',0) = Concurrent AHT

Basic example of concurrency

20 chats XXX complex product @ 1000 seconds = 20000 ** 1 push

15 chats simple product @ 500 seconds = 7500 * 2 push

6.5 hours = 23400 seconds login in chAts engaged productive

27500 /23400 = 1.17 concurrent

80 % chats 1 push

20% are 2 push

Even more simple

If I was logged in for 1 hour taking chats ... so 3600 seconds ... but I spent 5000 seconds worth (talking/wrap) in several chats during that 1 hour of engaged / productive time ( those 5000 seconds could be multiple chat and durations but they all total up to 5000 seconds in that 1 hour interval ). Then my concurrency rate or how many simultaneous chat I handled in that 1 hour is 1.38 ... then you can use that concurrency rate to figure out the concurrent handle time for all those chats handled within the interval by diving it by your single session times within the hour/day/week/month. Concurrent handle time should be used for daily/monthly/interval staff plans.

Example

10 chats at 360 seconds = 3600

5 chats at 280 seconds = 1400

15 chats = 5000 seconds / 3600 seconds (1hr)

Concurrency rate is (5000/3600) = 1.38

Then 5000/1.38 = 3623.1

3623/15 chats = 241.5 concurrent AHT

Keep in mind ... you may ask how can you get 5000 seconds of chat time in 1 hour ... when 1 hour is only 3600 seconds... well those chats start and end times are being done in parallel... aka concurrently.

Good article on Chat Mix/Pushes Vs. Customer Experience

https://fmoutsource.com/resources/best-webchat-concurrency

MORE TIPS

For WEB chat. Effective concurrency comes from ensuring your agents have the right systems to access information and complete ** multiple transactions **

For your chat platform to work effectively and to reduce the cost per contact... The systems your agents access have to be just a strong as the platform itself... The major hurdle to a good concurrency rate in a chat environment is multisession capable software.

(Concurrency rate) explained simply is how many multiple chats sessions an agent can handle simultaneously each interval.

So, if your agents cannot access more than 1 account at a time... then your chat platform will be useless and it will cost just as much as a voice call which is a 1:1 Ratio.

If you do have a multisession capable customer database … ensure it can also handle multiple customer transactions. It would serve no purpose in a chat environment if you can look up multiple customer accounts and then be able to only process once actual transaction at a time.

*** window time and dear air chats***

AST (Average session time)

AST should include the full session time from the moment a chat is accepted/active to the time it is ended. So if I don't respond for 5 minutes, that should be in there the session time.

As far as a solution! ‪ You can control that AST and things like dead air chats or customer who take too long to respond by setting a standard ... if you don’t respond within 2

minutes we automatically end the chat and close that session! It a good methods to curtail dead air periods and or customer who abandon the chat or take too long to respond. If we waited till a customer ended every chat....well, some sessions would last days and also you need to set a reasonable standard for your customer to respond to you based on your line of business.

Parallel Capacity Agents

noreply@blogger.com (Doug Casterton) — Sat, 13 Jun 2020 13:32:00 +0000

This is re-post from Marlon Martinez - original post can be found here “Parallel Capacity Agents"

Parallel capacity agents

In a nutshell Parallel capacity agents can be created by letting all your agents keep their work from home hardware/kits, even if let’s say the majority come back to at site desks. The basic concept is to move to UBER style scheduling (read more about this in this article: Gig Economy Scheduling) AKA flex scheduling, many orgs have had an issue doing this as most of us are used to that camel hump style scheduling. Typically, all agents want (7am-3pm,8am-4pm and 9am-5pm) and we know there is less of an appetite for EVE and weekends shifts. So, if they became “Parallel Capacity Agents” they could jump on during peak periods from home and get paid a set premium as supposed to things like 2X OT, 3X OT ect. The advantage to the agents of course is they can pick up hours, shifts and save on transportations costs, parking, daycare costs etc...

Pro’s of “Parallel Capacity Agents”

· Less office space rental/lease fee’s

· More employee satisfaction (Less cost to agents getting to work) to do 2-3 hours of OT

· Can literally get up to double your FTE amounts during peak periods and weekends if a premium if offered or doing a product launch / Media issue/ or peak sales periods like Black Friday or Cyber Monday

· Works well in a FLEX schedule model

· Work well in an OMNI Channel model

· Camel hump be gone! 🐫 you can flex in agents in those AM and EVE spikes or on weekends

· FTE OT requests for special events and launches can be made well in advance and planned via the FLEX uber style (We tell you what we need) you just pick it up type shifts

· If you had 1000 FTE as an example and 70% came back onsite (30% will be WFH) but the 100% will have WFH kits… making them “Parallel Capacity Agents” and instead of have only 30% flex staffing you have 100%

· Info-structure is already in place (as we can see with this WFH pandemic issue), and you can subsidize your agent’s internet costs as well

· No need to contract 3rd party/outsourcers to meet EVE or weekends demands

· No need or a reduced amount of seasonal hiring (Less training cost/ Less seasonal attrition)You can handle volumes easily in excess of your (YOY) year over year base. YOY contact volume would be based on contact rate % and your customer base and scrubbed out anomalies and events non-reoccurring , ect.

· Less load issues, less absenteeism, and also can help agents stay at home when sick (Reduce virus/cold spread)

· WFH kits investment has already been made. You can assign each unit to each employee ID and make it part of the IT intake process/Have mobile techs come to agent’s home to fix any issue that cannot be fixed remotely

2020 What is WFMs role in the Contact Centers Future?

noreply@blogger.com (Doug Casterton) — Sun, 07 Jun 2020 17:13:00 +0000

Workforce Management practice in its present guise has really only been around for the last 20-25 years, although you could theoretically date its roots back to when employees were entering factories punching their entry/exit onto to cards in order to record what hours were worked.

It has however in the last couple of decades seen huge transformation, becoming ever more connected with various functions within a business to help inform the forecasting process or provide insight and solutions into ever more complex customer scenarios as businesses digitise.

Over the last few years, the shift toward a better employee experience has been one of the biggest trends in workforce management, involved in creating better experience for employees through more aligned scheduling practice that accounts for an employee’s life outside of work, and leading to greater efficiency by increasing attendance and reducing employee turn-over.

We have also seen Workforce management process and solutions develop to create a more communicative relationship between management and employees, helping employees to communicate and share information with one another in real-time, no matter if they’re working from home or at a separate location. Both of these trends are set to continue but improving your workforce management process is a never-ending goal for all businesses, so what are the 'bigger' trends I see emerging as a focus for WFM over the next 2-3yrs?

WFM to WFO to WEM and back to fix the PAST

WFM system vendor for the last couple decade have been focussing product development expenditure on developing a more holistic system that integrates under one roof WFM, QA, & Analytics. More recently the latest buzz word has been Workforce Engagement Management (WEM) with system focus on questions like; How do I develop a homeworking strategy which provides greater flexibility and maintains service levels?, How can I still react to changes fast – and be proactive in anticipating demand remotely?, How do I empower staff at home? How can managers measure and adjust performance on a more agile basis? This product development focus has however left some glaringly big holes such as:

1. Customized staffing models for every skill and channel type do not exist

2. Capacity Planning is still on the whole completed in excel

3. The ability to reconcile SL variances has to be done manually

4. Integration with the financials and automated reporting does not exist.

I am however seeing some signs of new players focus their product development in this direction. Read more about this topic here

Forecasting & Planning for Facebook Messenger, Whatsapp, Telegram, Viber, WeChat, Line

Asynchronous messaging or fire-and-forget conversations via Facebook, Whatsapp, Telegram ect certainly provides customer convenience that matches today's hectic life's. The challenges to the contact centre lie in leveraging context, duration/timing, and business rules to determine if a message is part of an existing dialog, enabling seamless transitions between automation/bots (AI and directed dialog) or delivering the entire asynchronous messaging conversation to an agent. Once you have overcome these technology & process challenges, planning for these new channel types can be very similar to how you plan for Email (which in its self is a form of asynchronous messaging when you think about it) but there are some key differences that need to be considered.... read about it here: Forecasting & Planning for Facebook Messenger, Whatsapp, Telegram, Viber, WeChat, Line

Workforce Management to Workforce Intelligence

Due to technology and the use of data, the field of workforce management is shifting more and more towards workforce intelligence. Technology supports the entire cycle: from scheduling to day management and from reporting to forecast. We know already that the planning process itself can be monitored and improved through data. The next step is the smart combination of data, operational insight, scenario planning and operational decisions.

Gig Economy Growing

In the last few years, the gig economy has skyrocketed in terms of popularity and its becoming increasingly popular for businesses to utilise as part of the business operations. With this comes the need to develop WFM process and practices that increase workforce flexibility but also improve the work-life balance for their employees by giving them more freedom to choose when and where they work. Read More About Gig Economy Scheduling here

Greater System Integration

There seems now to be an application designed for nearly every business challenge encountered these days. With this being the case integration of workforce management solutions with various systems such as Payroll, Learning Management Systems (LMS), quality, marketing, CRM, customer surveys, and omni-channel systems (to name but a few obvious one) would advantageous to create automation opportunities. This could also include integration with external systems to gather weather influences (good weather, less volume and vice versa), traffic on the roads the opportunities are endless. Whether an internal or external source the aim will be to create two-way synchronization of data and more importantly automated action as an opportunity for greater benefits.

Forecasting, Scheduling and Real-Time Management With AI

Using artificial intelligence has started to become popular for workforce management software but is still fairly limited in ability. However, I see this trend continue to grow over the next decade as businesses strive to use AI to cut down and automate on time-consuming and improve the accuracy on complicated activities. Additionally, the opportunity to use WFM AI in the real-time space is particularly interesting especially as predicting customer interactions is becoming ever more volatile due to the digital revolution requiring workforce plans to also become 'agile' in response. It is already possible to predict “near real time” via predictive modeling . Traffic management then becomes automated scenario planning with predefined response strategies. A supermarket can adjust prices in real time based on current supply and demand figures. If you translate that into WFM, real-time supply and demand for work and people can also be coordinated. If you would like to learn more about what possibilities AI can bring to workforce management, then check this video out where I talk about how WFM can be applied to Real-Time.

Empathy & Communication skills becomes No1.

With advancing technology, and a changing relationship between agents and the business, it is clear that workforce management must adapt. It is already no longer enough to be the analytical type that likes to sit in a dark room with the spreadsheet. McKinsey predicts that by 2030, employees with technological skills and social-emotional skills will be the most in demand skillsets. As greater parts of the workforce management process are automated you will need WFMers who understand how technology works and people with skills that cannot be replaced by technology: things that involve emotion, empathy and feeling. Read more about why I believe Empathy becomes the No1 skil of the future

Reflections on demand management (Part 2)

noreply@blogger.com (Doug Casterton) — Mon, 01 Jun 2020 11:30:00 +0000

The following article was originally written by Gabriel Pizarro Garcia in Spanish on the WFManagement sister blog – CLCIK HERE TO READ THE ORGINAL

I could not sleep, so I started to ask myself questions and as dominoes, many more began to fall. If you want to read more, go to part 1 -> link

Today we will continue to see demand issues in the management of a contact center. We will see the questions that arise in multi-channel scenarios, the importance of the care model, the impacts on demand that emerge from the points seen above and a brief summary.

Disclaimer: I work in demand management, workforce, and customer service, know that everything I will write is not in absolute quality, it is a reflection. I welcome the opinions and comments.

Many channels, many demand fronts

We could think of the attention of each channel as something independent and isolated. Customers call us, write on social networks, go to stores, write emails, use an app. By handling channels as silos, each manages their numbers.

Customers and customer service do not generally work that way.

It began many years ago to talk about multichannel (and after omnichannel) and we realized that customers who went to the store called and wrote an email ... and why not a tweet and a message on WhatsApp, while consulting the app ... ALL AT THE SAME TIME. There will be no shortage of those who send a letter.

The question here is, what do we do with all that data?

I wish we had a platform that would unify everything, at the CLIENT level and allow us to understand that perhaps the need is one, and we have demand in different channels that could be simplified by providing the solution to the client in one of the channels. I wish the world was that simple, the need so clear and the solution available to everyone.

In practice, companies will offer different types of transaction depending on the channel, and the client will choose where to contact, or there may also be no global strategy and the channels “compete” with each other to keep the best numbers, each showing its own achievements.

Omnichannel solutions are not always accessible to all budgets and projects are complex to implement. Time, cost and clear strategy are resources that not all companies have in abundance.

It often happens that the channels manage as silos, for organizational, historical reasons, power struggle, different approaches, dependence on organizational charts, etc.

Or maybe nobody thought about it.

No one saw everything from afar, as a customer and the company simply grew and opened channels, or bought systems and never unsubscribed the previous ones, or merged with other companies and and ... life.

Impact on Demand

After all working on this demand, I have to focus.

Given the existence of different channels to serve the client, we must understand some numbers that will help us see "the complete picture":

The demand of each channel, the number of transactions in each one
The number of unique clients that contact us per channel
Reiteration by channel, by type of call, by customer characteristics, etc.
The bounce between channels, that is, if a client uses a channel and has no solution, which is the next one where it will communicate
The favorite channel of each type of client
The impact of channel closure on the others
That a channel grows, does it imply that another decreases?
Can we move people between channels? Orient your experience?
And let's not forget about the money: does each channel cost the same? Each transaction?

There must be a unique strategy that considers these points, and based on that, demand could be oriented.

There are several examples of this, just to name a few:

In banking, by putting free ATMs and attention at a window with cost

Emails with sender "not responding" but with additional contact information

The reduced hours in face-to-face stores, and extended by phone, with app's available 7x24

These are all examples that try to direct demand towards certain channels.

The numbers I see, are the total demand?

The design of the service model, and the queues we set out to be able to serve customers, will create new challenges for us.

Do we serve by customer value?
By necessity / reason for the interaction?
By systems that advisers use?
By customer technology?
Depending on the number of calls you make?
If it is repeated?
If care is regulated?
Do we specialize a BackOffice?
Can all advisors attend to everything?

We have so many types of clients, advisors and products that the possibilities could be endless. Let's make it more complicated, and if the call has several of those characteristics at the same time? How do we prioritize?

In this, each company is a world, but surely we will end up generating specialized queues or making a huge queue where advisors must be very specialized and use 10 systems with all the complexity that this entails.

It is not always possible: there are more specific or technical requests that specialists will require. Applications and permissions cannot always be given to everyone (fraud prevention or just license fees). A good salesperson is not always a good technician. We don't want to serve our clients with our newest advisers, at least not the best.

Understanding how the service process is, whether in a store or on telephone or written channels, the queue design will generate “transfers” between queues or even between channels, which are not a demand in itself (the call is one, same transaction), but it should be taken this way when planning.

Example:

A person calls the call center to request information on their vehicle insurance policy, and at the end requests that a technician come to their home to charge the worn out battery due to the lack of use of their car.

If our service model has a queue of information for policies and another for home technicians, this will generate two interactions that in practice must be dimensioned separately, but correspond to the same client.

The example seems simple, only that sometimes the client was left with a doubt, and now he wants to speak to the policy advisor again, or sometimes he gets confused about the option, and enters badly, so we must transfer him or ask him to return to call ... what a dilemma.

Imagine this could be complicated if multiple channels are involved in the interaction.

The film could be worse: we may have a Back Office in the process, and the management lasts 20 days. The client in that period, could communicate for the same reason or for others. We mix synchronous with asynchronous interactions, with promises of deadlines.

Usually clients do not want to be transferred, they do not want to explain the reason for their call again to another advisor. Apart from generating breaks in the care processes, the transfers involve chains in the dimensioning that are difficult to synchronize.

We need a lot of information, and understand the relationship between metrics and the flow of each interaction that customers have with the company.

We are closing

To summarize, I want to list some points to keep in mind when evaluating customer demand and attention:

Understand that we have different demands:

Demand for different channels
Demand contained in self-management and automation (bot's)
Customer demand that is necessarily referred to assisted care
Demand that the same care model “generates” by transfers
Demand for repetition

This will require that we have a lot of information and we must connect it:

· Complete care model

· Global analysis of demand measured in transactions and in unique clients

· Defined care strategy: where we want to serve and where we are better

· Measure technology, connectivity, measure telecommuting

· Measure the automatisms and bots, all the non-assisted management

· Global and cross-channel access metrics

· Flexible infrastructure for quick adjustments, adaptation, overflows

Finally, go through and test the entire care process with the client's eyes, with family, friends.
If when talking about your work there are complaints and bad comments, there is a lot to work for!

(I imagine there are more things to consider, if you know some please help me in the comments)

What a serious mistake to think that the calls received were enough to make forecasts ... there is so much to investigate ...

This got extensive… I hope I have helped someone! If you have a book to recommend that I talk about this, I will be grateful.

Gabriel

Gabriel Pizarro Garcia

Spoiler: when you feel comfortable and think you know him, the attention model could change and who knows if it will return to how it was two years before.
Don't get used to your numbers and metrics, develop an overview of the business and a positive attitude!

Everything written in these lines could be fiction and any resemblance to reality is pure coincidence. It could apply for different industries and in different countries at least in South America, but it does not obey any particular company, client or situation. In some companies it is much simpler and in others much more complex, research and learn according to your reality.

Reflections on demand management (Part 1)

noreply@blogger.com (Doug Casterton) — Mon, 01 Jun 2020 11:29:00 +0000

The following article was originally written by Gabriel Pizarro Garcia in Spanish on the WFManagement sister blog – CLCIK HERE TO READ THE ORGINAL

Sunday.

Today my day started by recapitulating and sharing a book: “Factfulness: Ten reasons why we are wrong about the world. And why things are better than you think. " Anna Rosling Rönnlund, Hans Rosling and Ola Rosling book

It said something like this: " Knowledge based on data should be a maxim for all aspects of your life. In any problematic situation where decision-making is required, data is essential.

This awareness involves, in principle, an attitude of humility at every crossroads, by virtue of the certainty that no one can know all the data related to a situation. For this reason, it is always possible to recognize ignorance and wait until you have more data to express an opinion. Like children, adults must continually foster their curiosity to gradually increase their understanding of things. "

The book and the previous idea serve us all.

They will read demand management topics, the importance of knowing what place we occupy in the service process, the importance of technological resources, and some questions about automation.

Below, I'm going to write something that may be helpful to those just starting out or curious about the complexity of serving customers whether it's in the Contact center, in stores, or anywhere there are lines / queues.

The challenge is to understand demand so that we can later adjust capacity and provide the best possible service (in terms of speed and consistency). For this, we need information and business vision.

Today more than ever, we are in a crisis and everything is changing, we are accelerating processes.

Quarantine and what is already our “new reality” means that all the data we have are under discussion, from those related to forecasts (because nobody knows how this will continue), through to operational data (because the new reality of the telecommuting changed / created several indicators) ... and generates even more data ... is it enough with what we were measuring?

This personal and global situation caused me many questions , and incidentally, I wanted to write it to share.

Disclaimer: I work in demand management, workforce, and customer service, know that everything I will write is not in absolute quality, it is a reflection. Welcome the opinions and comments.

Where we are?

To analyze the information we have, it is important to know where we are, what role we play in the customer service process, because that will surely limit our access and the scope of the analysis.

· We are a company that has its own contact / stores?
· We are a customer service provider?
· We are a service channel within a multinational? One of how many?
· Are we an area within the care process?
· We are the last part of a chain of processes where the client is served?
· Does the care go one way or does it go back and forth between areas?

Business vision, thinking from the customer's eyes, is not always possible for everyone, simply because they do not have enough information to put together that map.

Do we see the entire customer journey?

IT areas

If we are a "normal" company, the world of connectivity and in general the IVRs, all telephony and sometimes even reports, will be in government and control of IT areas. This is very positive, because they are the specialists with the training and the ability to make things work.

The only detail is that their data, their information is usually not available, and Operations (responsible for channels, or customer service) do not know or measure, nor are they used to seeing. The information is not always expressed in business terms. It also happens, they participate in the implementation of the technology, but not in the development of the business below.

There are many problems that originate in connectivity and its correct dimensioning, also in people who do "tricks" and modify the information we receive.

Examples:

Inbound: We set up an operation of 50 advisers, but due to quarantine, we grew to triple that number. The initial infrastructure is not enough no matter how much we have people, we do not have channels. How does the client feel it? It calls and makes it busy. Faced with such a critical situation for the customer experience, how does the operation find out? You do not know, those calls do not exist if they do not enter the flow, are not recorded, do not record customer information or reason for calling. Only a report of channels can see that they are saturated, and that they need to be expanded.

One outbound: We set up an outgoing call operation with a dialer. We started with 15 people and quickly grew to 50 people given the good initial results. To our surprise, the base sweep advance, which was initially very good, did not grow at the same speed as the operation. The occupation of the agents fell noticeably and the account was no longer profitable. If, you guessed it, there were not enough channels for the necessary calls of an operation that grew, simply people did not have calls because the limit was set by the available channels to dial.

One of stores: We have a ticket system in the store that measures demand. The venue cannot exceed 100 people. We also have a hostess, who depending on how she sees the entrance of clients, generates tickets and delivers them. Sometimes it spawns a lot together to have on hand to avoid long lines. There may be some left over and at the end of the day he throws them away, some clients can also go through the box without a ticket, things like that happen daily. These are the data we use to forecast our demand, those of the ticket machine, when the only person who knows the demand in detail is the hostess.

This seems like something out of a movie, it happens with many implementations: the initial resource planning changes over time, and the installed capacities do not adjust, generating a drop in performance, low occupancy, loss of profitability, etc.

The size of the "pipe" will be our maximum limit. We must never forget that the capacity of the channels, of the connectivity, of the centrals, is ahead of all the operational numbers that we see.

This information should not be used exclusively by IT areas, much less at times when demand fluctuates and can easily exceed installed capacity. Having telework consultants, connectivity is the first thing to measure, to ensure that we do not have missed calls due to the new topology.

At the same time, people modify what is designed, generating deviations in the process.
We have to know and ensure that what is written on paper, is what happens in reality, the experience that the client lives.

It is important then to have availability of the information of the calls, in the whole process of attention.

We should monitor or at least have alerts of:

· Link saturation
· Channel saturation
· Packet loss
· Dropped calls without registration
· Customer complaints about not being able to communicate

Artificial Intelligence (?) And Self-management

It is time to automate.

The digital transformation has advanced in these months, which was expected to advance in years.

Technology allows us to develop self-management solutions, which once implemented should provide consistent responses, with a low failure rate and 100% availability, at low cost.

There are several features and, in general, a good implementation can achieve it. There are many examples of bots and self-management systems, and they are part of our daily reality.

This “contained” demand must be monitored not only with service indicators, but also with experience indicators to ensure that they do not fail and that they add value.

Some questions to ask the team to implement:

What happens if it fails? How does the demand impact?
What detours will I get when it doesn't work? Does it fail often?
What happens if you give incorrect information? What will happen to our client, follow or hang / cut? Does the client perceive it as silence, incorrect information?
It applies to stores, if we put some self-management totems, what happens if it is unplugged, who updates the content it provides?

Client: I want to speak to a person, thanks

BOT: Thanks to you for communicating, come back soon.

Being today a very important percentage of the total number of services in the different channels, the self-management and bot's metrics should be known to everyone.

When was the last time your robot failed? How many customers went through there?

This is where my Sunday came to.

These are some lines with the idea of pushing us to review things that we did not see before.

A new reality requires asking different questions, expanding our vision and adapting.

Do I have all the information necessary to make decisions, just as Mr. Rosling's book said?

Mmm I declare myself ignorant of some.
I'll keep asking

Gabriel

Gabriel Pizarro Garcia

Forecasting & Planning for Facebook Messenger, Whatsapp, Telegram, Viber, WeChat, Line...

noreply@blogger.com (Doug Casterton) — Thu, 21 May 2020 15:16:00 +0000

Messaging is the most frequent digital activity that a person does, there is no contention. If you've ever messaged someone to ask if it's a good time to give them a call, you've already experienced this shift.

The meteoric rise of Facebook Messenger, Whatsapp, Telegram, Viber, WeChat, Line, and IMO (if you live in Turkmenistan) has been nothing short of stunning. Why has this adoption rise occurred, how are customer service areas responding and how do you forecast and plan for this channel type – all questions I hope to provide some insight into with this article.

Firstly, let’s start with defining this channel type. Facebook Messenger, Whatsapp, Telegram are all examples of asynchronous communication. This is opposed to Phone, LiveChat, or face to face conversations which would be classified as synchronous communication.

Synchronous communication is immediate with people responding to information at the same time that it is shared with them. Asynchronous communication is simply where you don’t respond to information as soon as you receive it. We’ve actually being using it for millennia by sending letters across the world.

In customer service the rise of the internet in the 1990s and 2000s brought the email and livechat channels to life and moved the call center to the contact center.

Enter the rise of the mobile messenger apps - according to a Hoosuite 2019 report WhatsApp, Facebook messenger, and WeChat (the top 3 most popular apps) have 4 billion monthly active users. The rise of the smartphone has of course been a driver of adoption, but the hectic nature of modern life has also had a major influence; people do not want to spend large quantities of their time on the phone, computer or mobile waiting for a livechat to conclude. For those who consider time their most precious resource, the asynchronous nature of messaging may be appealing; they can respond whenever it's convenient, send a message from their mobile device, and then continue with other tasks whilst waiting for a response. Because these conversations are not carried out in real time there is less expectation that you will get an immediate response.

Similar to Email (another form of asynchronous messaging) this channel type also reduces failure demand because the conversation history can be saved within the case history, something more difficult to achieve with a phone call or a live chat since the conversation is session-based. A solid CRM system and speech/text analytics can certainly help with synchronous communication challenges but it is still common that customers are required to repeat information the next time they contact customer service. Asynchronous messaging on the other hand eliminates this inefficient repetition, allowing agents to gather customer information faster, engage other colleagues and solve the customer issue that much sooner.

So, there is no doubt that asynchronous messaging or fire-and-forget conversations provide customers the convenience that matches today's hectic life's. It is therefore surprising that adoption so far in customer service has not been as meteoric as it has been in people’s everyday personal life. However, with launch of WhatsApp for Business, Apple Business Chat (which enables customers to talk to businesses via the Messages App) or the advantages of implementing in-app messaging strategies to both App adoption but also the ability to identify customers thereby offering a smoother experience, I predict this is about to change very rapidly.

The main challenge for a contact center to truly adopt asynchronous messaging lies in leveraging context, duration/timing, and business rules to determine if a message is part of an existing dialog. It is also important to enable seamless transitions between automation/bots (AI and directed dialog) and deliver the entire asynchronous messaging conversation to an agent.

Dynamics and Behaviours of Asynchronous Messaging

The first step in forecasting and planning for a customer channel is to understand its dynamics and behaviours. So, let’s break it down…

Like the email channel asynchronous messaging conversations are long-lived and can span days, meaning there is the possibility that more than one agent or self-service would be handling the same case. Defining assumptions around desired length of time an agent stays dedicated to a case is critical e.g. if a customer has not responded in the last 2-8 minutes should the conversation be “timed-out” to ensure agent capacity is efficiently managed?

Asynchronous messaging is not real-time, so no need to apply occupancy assumptions meaning it has significant potential to decrease staffing requirements vs say Phone or livechat channels.

The fact that messaging can be asynchronous means that the pressure of service peaks can be spread out.

Asynchronous messaging shares similarities to Livechat in that agents can deal with more than one conversation concurrently.

Asynchronous messaging when combined with intelligent routing has greater potential to apply case ownership principles.

When setting how quickly you want to respond to the last customer reply, it’s worth noting that with asynchronous conversations not every interaction is like for like i.e. like email customers expect a quicker response to their first inquiry but are willing to wait longer between messages once they are confident their inquiry is being dealt with. Unlike email however there is a greater expectation that you won’t be waiting days for a reply – so critical to have regular checkpoints to reassure the customer their query is still being dealt with. Mapping these checkpoints is critical for forecasting purpose to understand channel hopping behaviour.

As a minimum you need to break contacts into two broad classification – first contacts (that require a quick response) and the rate and frequency of the subsequent contacts related to that conversation end to end. You will find each type of contact reason will have a different pattern and number of subsequent contacts. Thus mapping out these contact reasons is critical to ensure you are not lost in an average. With this in mind forecasting at both the interaction and the end to end conversation level is important depending on the granularity of the forecast i.e. the closer you get to real-time the greater the importance to adopt interaction level forecasts.

Like email Handle time can be broken into three components: time spent actually writing to the customer, time spent researching answers or enacting a process, and wrap up time to complete case notes.

From a WFM perspective there is lot yet to be discovered and understood about how to forecast and plan for Asynchronous messaging – there is no best practice yet established. What is however certain is that this channel is set for big growth in customer service in the coming years and I for one look forward to embracing the challenge with open arms.

Overview of Outbound Dialer

noreply@blogger.com (Unknown) — Sat, 16 May 2020 14:42:00 +0000

In an outbound contact center, the most heard word is "Dialler". In simple words, the dialler is a device which we use to make outbound calls.

I'm sure that all of us would have got at least one call from the banks for credit card, loan etc. or calls from stores regarding their new products.

We would have also experienced some calls where no one would speak on the other sides even after we answered the call. Some calls where the agents are already speaking to someone else.

These are the results of the use of dialler in the contact centers.

Today, we are going to look at the different types of Outbound Diallers used in contact centers and how do they work.

1. Manual Dialler

This is the most common type of a dialler. The simplest form of Manual dialler is connecting a telephone line with a phone to the agent's desk.

In this method, the agent looks at the number, inputs the number manually and dials them.

Manual dialler's main disadvantage is the amount of time taken by the agent to look up the number and dial.

Once dialled, they have to listed to the ring and proceed, or if the calls isn't answered, they have to mark the number and remember to dial again.

So, as we see lot of time goes in data cleansing and this affects the agent's productivity.

This dialler type is used mostly in the complex call environment where the agent is required to be prepared prior to speaking to customer.

Being cost effective, this dialler can also be used in small contact centers.

2. Auto Dialler

As the name says, this is a software bases dialler where the calls as placed automatically. The main reason for having an auto dialler is to get a better productivity.

There are mainly three types of auto dialler which we will look further below.

Preview Dialler

In this method, the agent has to tell the dialler that he/she is available for the call post customer data being presented to the agent.

This dialler allows the agent time to preview the customer number and the case history and the agent decides when to place the call.

Just like the manual dialler, this dialler can also be used for complex call environments.

The major difference between manual and preview is the time saving of looking up the number and dialling as they are done automatically. If the call isn't placed, the agent don't need to remember it as the dialler does the same.

Also, the agent can manually disconnect the call of it isn't answered.

Progressive Dialler

In this method also, the agent has to let the dialler know that he is available post which the agent is presented with data.

But, the call is dialled immediately as soon as the data is presented.

Calls that are not ringing or reaches voicemail are disconnected immediately and the calls which aren't answered are disconnected after some prescribed time.

This type of Dialler reduces the time spent by agent for the call and thus increasing the productivity.

This dialler can be used in those environment which has less complex calls.

Predictive Dialler

This is the most advanced dialler for an outbound campaign.

In this type, the calls are dialled automatically and is connected to the live agent as soon as he/she finishes the previous call.

This dialler is designed also to identify the voicemail and live human interaction and if the call reaches voicemail, the call is disconnected automatically. If the call is answered by actual human, the call is immediately connected to the idle agent.

If no idle agents are available, the call is disconnected automatically.

In this dialler, sometimes customers may receive the call with no agents actually available to talk.

This dialler reduces the idle time to a great extent and keeps the agents busy as much as possible. Therefore maximum productivity is achieved when this dialler is used.

Predictive dialler is mostly suitable for those environment where a large number of calls are to be made in a shorter duration.

Apart from the above three types of dialler, there are many others which are developed recently and are used by the contact center.

Selecting which dialler is suitable depends upon various factors such as the type of calls, time to make the calls, geographical location, budget of the company etc.

Thank you for reading.

Stay tuned!!

Disclaimer: The above article is purely my personal view and doesn't represent any contact centers.

Vinay Vasudevan

WFManagement

Overcoming WFM Professional Defeat & Disengagement

5 Reasons why WFM delivers CX

The Next Big Thing to improve a BPO Bill to Pay Ratio

Workforce Management (WFM) Best Practices

How to Calculate FTE for Offline Queues?

I have attached a excel file at the end of this article which shows all the calculation methods explained below.

Therefore for Offline Queues, it is always suggested to go with standard occupancy of 85% which will be ideal.

FTE Calculation

Now what if the productivity which was calculated using the Time and Motion Study already had Occupancy baked into it?

Conclusion

Offline FTE Calculation Sheet

Schedule Inflexibility - Basic Introduction

So, what is Schedule Inflex?

Now lets come to the major question. How to measure Schedule Inflex?

I have attached a excel file at the end which contains the calculation, please make sure to see that

Schedule Inflexibility Calculation Sheet

To Social Media or Not Social Media that is the question?

Introduction to Call Routing Techniques

How does Call Routing Works?

Types of Call Routing Techniques

Conclusion

The impact on Email FTE Requirements following a change in Service Level Targets

Live Contacts after the lines close - what to do....

The Forecaster’s Luck by Ger Koole

The Maths of Adherence

Introduction to Simple Forecasting Methods in Excel

Naive Forecasting Method or Random Walk Method

Seasonal Naive Forecasting Method or Seasonal Random Walk Method

Random Walk with Drift

Moving Average Method

Attachment: Simple Forecasting Model

Contact Center Basic Forecasting Building Blocks

WFM considerations when launching new customer support channels

Highlights From Video

WFM for Outbound Planning

The Challenges surrounding Flexibility by Prof. Dr. Ger Koole

The Dreaded Camel Hump 🐫

Chat Forecasting: Concurrent AHT

Parallel Capacity Agents

2020 What is WFMs role in the Contact Centers Future?

Reflections on demand management (Part 2)

Reflections on demand management (Part 1)

Forecasting & Planning for Facebook Messenger, Whatsapp, Telegram, Viber, WeChat, Line...

Overview of Outbound Dialer

To continue reading and learning about outbound check out this post: Applying WFM to Outbound