PPC Change Forecasting Using Expected Value

Amanda wrote a brain tingling post on bookending PPC changes back in April. This gave a comprehensive overview of how to forecast the impact changes/optimizations you intend to make to a PPC account. The first half of this no-duh tip is to read Amanda’s post and always bookend your changes. The second half is to use Expected Value (EV) as part of that forecasting model.

While to some this may seem complicated, it’s really pretty simple. EV is simply the percentage chance you think a change has at realizing its fully forecasted value multiplied by the forecasted value.

For example, if you are going to make a change that over a month is forecasted to increase spend by $1,000 and leads by 10 but think that there is only about a 75% chance that those numbers will be fully realized, your adjusted forecast is for $750 increase in spend and 7.5 new leads.

Applying Expected Value to your forecasting, while simply an estimate, help make your numbers more accurate. They can also help you focus on tasks that have the highest Expected Value versus the highest gross forecast value.

Imagine you have two tasks. Task A is projected to bring in 10 leads, and B is projected to bring in 7 leads. Using raw data only you will stare your day focused on task A. But A is a more complicated task, such as a QS breakout, and has only a 50% chance of realizing its full value. On the other hand, task B is keyword additions where you have done an analysis and found some keywords that you feel are likely to convert. You place a 75% EV on those keyword. In this case A is expected to generate 5 conversions and B is expected to generate 5.25 conversions.

Report on Forecasting in Monthly Reports

Most monthly reports include a great analysis of the past month with some pretty graphs, charts, and comparisons. They will likely include some important dates (such as when the account was launched in a new engine, or started remarketing) and even some action items for the next month. One thing that is often missing from these reports is a forecast for the next month.

This isn’t just budget and lead goals, but a further analysis of what the account is actually expected to do based on seasonality, conversion rate optimizations (CRO), concluded tests, etc. The goals are the benchmarks, but the forecasting gives you, your boss, and/or the client an idea of how achievable those goals are based on real insight.

Last week a post (okay, I wrote the post) was written on using Google Insights for Search. This combined with some internally created projections should be added to your next monthly reports. This will help you understand where you have work to do, and will prevent those awkward mid month conversations where you are explaining that impressions are down for the month and how that’s a seasonal trend.

Determine (and stick to) Your Lag Conversion Attribution Model

Some accounts have near 0 conversion lag. This means that most, if not all, of their conversions come on the same day as the first click on a PPC ad. Other accounts have time lag so significant that it can make a major impact on strategy and optimization. If you know this, and adjust accordingly, it’s just one more step in your reporting and account management process. But if you ignore this you could be hurting your accounts, and making yourself look bad.

A quick example of this is an ecommerce client of mine. When doing the April monthly report for this client there were 250 conversions for the month. Within 2-days there were 6 more conversions. That’s a 2.6% increase in conversions. Luckily, we make sure to attribute these conversions to May (because that’s when they actually came in and our client made revenue off of them).

It’s an easy process. When doing this month’s monthly report you look at last month’s numbers as (keep track in a simple excel spreadsheet), then pull up the interface and see how many more you’ve had over that number, and add those conversions to how many you see this month in the interface.

You can do your numbers differently; the important thing is simply that you are consistent, and that you don’t ignore lag conversions.

Those are my 3 no-duh tips that most people forget to keep in mind when forecasting, optimizing or creating monthly reports.  What simple things do you do that many people forget?

Check out this Guide to Pizza for some other fun facts!

According to a close read of the language of the blog post, this is a great thing.  “Many of the actions” that people took shortly after receiving custom alerts “can now be done by creating automated rules.”   Also, they’re not canceling or murdering alerts, Google has “decided to sunset” them.  How nice.  Who doesn’t love a good sunset?

Another great thing about this change – even though your e-mail notifications are soon to be a thing of the past, Google is “actively working” toward creating an e-mail only option in their automated rules.  Eventually, you’ll be able to receive an e-mail notification when your rule requirements are met, but there won’t be any action taken.  Which, when you really think about it, is the thing that they’re getting rid of right now.  How amazing!

According to any sort of sarcasm meter you may possess, this change is lame.  I’m sure they’ll get around to figuring out that tricky e-mail notification thingy (you know, the thing that predates automated rules) once an acceptable level of users have switched their custom alerts over to automated rules, which means more immediate action on Google’s part.  Which inevitably leads down the road to higher bids, more competition and more money for Google.

Just like their mandatory update to ad rotation a few weeks ago, this is a change from Google that in effect is a forced step toward automation.  So, I hope you like Google and the things it does for your account.  It can be great.  Really great.  But it can also be terrible.

I’ve had accounts that had spotty conversion tracking, which makes custom alerts great and automated rules terrible.  That doesn’t much matter now.  We can all look forward to a new day when Google releases that great new functionality where an automated rule only generates an e-mail alert, but for now, prepare yourself to bid a tear-filled adieu to that thing that already exists: a custom e-mail alert.

Isn’t it a beautiful sunset, folks?

To use the tool, simply log into your analytics account, expand the content topic section, and then select in-page analytics, which will be your last option. Once you select it, a browser-like window within analytics will appear, that shows your home page. A top line report for your selected date range will appear above this window, which will show you the data set that is being analyzed and some key stats like pageviews, bounce rate, exit %, etc. A cool trick you can use is to open a new tab within the browser and load your website. Do this and wait a few seconds for the data to load and you’ll have full screen action with your click percentages included, pretty awesome. Unfortunately, using a full browser does, at times, limit your functionality by only showing you top-level traffic and not giving you the option for advanced segments, which you’ll need to do the analysis I’m going to walk through. If for some reason that segmentation option doesn’t show up, just jump back into analytics. The image below is a screen grab of my full browser view (minus the client data for confidentiality reason), which after a few refreshes did populate the segments I loaded in analytics.

In-Page Analytics View From Unique Browser

Now that you’re ready to go, what I wanted to understand is how mobile visitors navigate the site as compared to desktop users. In order to do this, I selected mobile traffic and compared it to all visits. To get more accuracy, you could setup a segment that includes all visits minus mobile but for a quick evaluation, comparing all traffic versus mobile should do the trick. Once I had my segments setup, I selected Mobile Traffic from the drop down box and first looked at clicks. As an FYI, this particular client is a SaaS platform and their site is mobile optimized but not mobile specific. Two things that were immediately apparent to me:

1. 69% of mobile users who hit the home page immediately clicked through to the login page. What I can draw from this is that there are a large amount of users who are actually using the platform via their mobile device. My immediate thought is that they should be taking a hard look at adapting their software into a mobile app for android and iphone. This will help in their campaigns on a couple fronts. First, it’s a huge selling point for users. “Buy our service, get our app for free and use it anywhere.” Second, it will help with direct conversions via mobile PPC searches. Click to download app extensions are made for this. Streamline the process and give folks an option to get right into the software.
2. Out of the mobile folks who didn’t click the login button, roughly 30% of them clicked through to the pricing page. This makes sense to me as that’s typically my plan of attack when sourcing SaaS. For this particular client, it’s also the page with the strongest call to action and easiest path to a conversion. This also led me to think about and compare these statistics to their overall traffic (which if you have that as a segment option will automatically populate when you hover over a percentage box) and also dig a little deeper into the benefits of sending traffic to the pricing page.

Here’s what I found on the additional dig. For all traffic, a smaller percentage of people click through to the pricing page and the clicks to other sections of the website are more equal rather than skewed heavily to that particular page. What I really wanted to know was, where are the people who actually convert going? Well, this is where things get interesting. After selecting completed purchases as my metric, rather than clicks, it was clear that the behaviors of mobile and search are very different. The two most popular pages are “how it works” and “pricing.” My intuition told me that mobile wouldn’t convert on a more content heavy page like “how it works,” especially knowing that a user needs to click through to the pricing page to actually convert. This was exactly the opposite. Mobile users who clicked through the how it works page converted nearly 4x more frequently than going direct to the pricing page. Unbelievable! Adding a heavy content driven page and step to the conversion funnel actually improves conversion rates on this mobile page.

The desktop statistics were drastically the opposite. Those that clicked through the pricing model converted 6x more frequently than those going first to the how it works page. This got me thinking about what I thought I knew about mobile users. Best practices told me less information, more streamlined, easy conversion funnel. However statistics are telling me that I need to deliver the information before I ask for the sale.

The conclusion I drew from this, is that for this particular client, where over 20% of their traffic is currently coming from mobile, is that they need a customized mobile experience. They need to control the experience their mobile users participate in, give them fewer options to click, and make the path succinct while also providing relevant information up front. While this does go against a more tradition approach of “less is more,” the data is telling me that in this case, more is actually more.

For those of you who keep reasonably up to date with the goings on in Google AdWords, you’ll know that they’ve recently made some changes to the ad rotation options. The original 3 settings were:

- Optimize for Clicks

- Optimize for Conversions

- Rotate evenly

The most recent changes have been made to the Rotate evenly setting. Now, if none of your ads in an ad group have been updated in more than 30 days, Google will automatically start optimizing for clicks.

I could go into a nice long rant about how terrible I believe these changes are, but that’s for another day. If you dislike this change as much as I do, please sign the petition to get it changed.

Anyway, on to the Q&A. This is a summary of a instant message conversation between my Google rep and I. In it, I’m the one asking the questions, and he’s the one giving answers.

1.) What day will this change take effect?

- I’m not sure. They don’t tell us that information. We find out when everyone else does.

Answer: May 11th. This setting is already live in your accounts.

2.) Is there an opt-out option?

- No, there is no opt out. Everyone’s accounts will have this.

3.) How quickly do the changes take effect?

- If on launch date, your ads haven’t been updated in 30 days, the algo will immediately begin serving your ads on an optimize for clicks basis. If there isn’t enough data to optimize for clicks, it will continue to rotate until there is.

4.) Will there be an alert or a notification to tell me when an ad group is switching to or is in optimize mode even though others are not?

- No, there are no notifications.

5.) Will there be a way to tell if an ad group has gone into the optimize for clicks setting?

- I don’t believe you will be able to tell other than seeing impression distribution changes.

6.) Does this affect only search campaigns or does it happen in display campaigns as well?

- This will happen in both search and display.

7.) Is there a threshold for optimize for clicks like there is for conversions?

- Optimizing for clicks is a very different calculation than optimizing for conversions. It’s based more of a confidence interval rather than an actual threshold.

8.) Is there a setting to make it optimize for conversions after 30 days rather than optimizing for clicks?

- There’s no automatic setting for that. You would have to do it manually.

9.) How much will optimizing for clicks throw off my ad copy tests if I don’t get to it immediately after the 30 day limit?

- It will have some effect, but it shouldn’t immediately make your tests invalid. The serving percentages don’t automatically go from 33/33/33 to 98/1/1. Depending on the data, they will go to something more like 45/35/30 or 40/40/20.

10.) Since the ad rotation settings are at the campaign level, will updating an ad in one ad group keep the entire campaign from going to optimize for clicks?

- No. These decisions are going to be made at the ad group level, so each ad group would have to have at least one ad updated every 30 days.

11.) How major does the ad change have to be? Can it be something small or does it have to be a completely new ad?

- Any change at all will make the 30 clock reset.

12.) Does pausing or activating an ad count as a change and reset the 30 day clock?

- Yes, this will reset the ad test for each ad group you pause/enable campaigns in.

13.) If I change the rotation settings at the campaign level from Rotate Evenly, to Optimize for Conversions, and back to Rotate Evenly, will the 30 clock reset?

- No, because the clock changes only when the ad is changed, paused, or enabled.

If anyone has any other information given to them by their reps or has found contradictory answers anywhere, please share them in the comments!

Use Google Insights for problem solving and keyword identification

Google insights can be used to help troubleshoot when a past performing keyword starts to do poorly, or starts doing very well. Simply type your query into insights and you’ll see if volume is trending up or down.  For “Grocery Coupons” you can see a clear decline in volume.

This tells you that you better start adding some keywords, or find other expansion opportunities because your CPC is going to rise (in this example it probably already has) and inventory is limited. To determine if this is an industry trend or a keyword specific trend you can click the “Growth relative to “x” category” tab above the chart. For Coupon’s the X is replace with Shopping and looks like the below graph:

Based on this you can see that the shopping category spikes in the holiday season, but besides that is relatively stable. Grocery Coupons however, has relatively declined by 50%.

At the bottom of the screen you’ll also see new rising searchers and related top searches. This information can be used to find new keywords that you should be bidding on. It’s not an extensive list, but it shows some categories that have an upward trend so you at least know what direction to go in.

Use Google Insights “Foercast” to help determine and allocate budgets

This was a feature added in 2009 and works only for larger volume search queries but it can help you determine, based on seasonality, determine what your budgets should be in the coming months. Not all queries will show forecasting, but when available this tool can assist with setting budgets.

The below graph compares search queries for “Craft Beer” (blue), and “Green Beer” (Red). You can see the obvious and expected spike in green beer every March, a trend that Google Insights is forecasting to repeat again in 2013. There is no forecasting data for the query craft beer because it’s trend has been unpredictable, specifically in the drastic increase in volume over the past 2 years.

Based on this graph, if you sell green beer, you can allocate nearly 100% of your budget around March. You could also expect to continue increasing your search budget for Craft Beer as it is clearly on the rise. Obviously this is a pretty easy trend to spot without Google Insights, but it illustrates how you can use the forecasting feature to determine budgets before expected spikes in impressions are expected. That then allows you to adjust monthly budgets, and daily budgets before the swing so you never miss out on peak performance times.

Using Insights for Search to determine new geographic areas to target

You can also use Insights for Search to determine if there is a new market you should enter, or a market you should exit. In the graph below the query “Diet Pills” is compared across the US (Blue), Australia (Red), Germany (Orange), and the UK (Green).

You can see that based on this graph Australia is a market that has shrunk significantly, but that the UK is at all time high search queries. So based on this you might want to think about shifting marketing budgets out of Australia and in to the UK.

These are just a few ways that Google Insights for Search can be used for success in PPC. What are some ways you’ve used it?

But what happens when Google releases a new feature, Bing upgrades a targeting capability or the company in question opens up their market to a new product-line? (Remember when Bing changed phrase and broad match keyword behavior back in August of 2011 and many accounts saw an influx of traffic for less-than-relevant queries? Hot mess, that was.) That individual and/or small team gets a few extra to-do’s added to their consistent and typical list, which inevitably means a couple things may fall off to the next week, or even month.

This is where we make our unique offer! Sure, the bulk of your company’s paid search management can be maintained by your trusted in-house team, but here’s your chance to let them keep up that hard work and offset the last minute or extraneous tasks to us! Enlist our team of rock stars to check that list off a little quicker while your team sustains the rest!

Here’s what we’re offering:

• Paid Search Assistance Retainer: For a flat monthly fee, you receive 45 hours from Hanapin to divvy up as you wish! Keyword research, ad copy writing, ad group restructuring for quality score…you name it! The hours are yours; we’re here to help.

If our offer seems like exactly what your team needs, or if you have more questions, get in touch with myself (kayla.kurtz@hanapinmarketing.com/812-330-3134 x111) or Director of Sales, Tom Hootman (tom.hootman@hanapinmarketing.com /812-330-3134 x120) and let us know how we can help your team get a little breathing room!

We’ve covered several Analytics topics this week, including:

• 10 Google Analytics Hot Spots

• Finding New Test Subjects

• Using Google Analytics for Lead Generation

• Google Analytics for Ecommerce Accounts

So what do you like most about Google Analytics? We would love to hear about your favorites in the comments section below. Remember, you can always access the PPC Hero series database by checking out the blog series tab at the top of the page. Thanks for reading!

If you’re running a PPC account for e-commerce, you likely already understand the importance of dynamic reporting capabilities, so I’ll try not to make this article sound like a sponsored plug for Google Analytics. But, as an Account Manager here at Hanapin, I can’t fathom running an e-commerce PPC account without the powerhouse that is Google Analytics. For e-commerce accounts, AdWords data and reporting really won’t give you an accurate understanding of your marketing metrics, not only as they pertain to PPC alone, but also as to how your marketing efforts across multiple channels are interacting with each other.

I’d like to share some basic Analytics reports I find invaluable in managing e-commerce PPC accounts, some ways to use these reports to tease out extraneous data you might not have considered in context, as well as some tips for creating your own custom reports. I’ll also talk about how you can use Analytics to understand the impact of PPC on other marketing channels (and vice versa).

PPC Basic Reports in Analytics

First, let’s start with some basic reports you should be using.

To get basic reporting data on your AdWords account, Analytics actually make this fairly simple and easy to find. If you’re in the newest version of the Analytics interface, you’ll see an Advertising section in your left side bar:

If you click on “Advertising” your menu will expand to display the AdWords sub navigation items. From this menu you’ll be able view data for you campaigns, keywords, search queries, destination URLs, etc. More specifically, you can see how your AdWords campaigns, etc. are performing by number of site visits, time on site, bounce rate, goal completions (usually these equate to your conversion definition in AdWords), and, most importantly, by revenue. Also pay attention to the bounce rate of your campaigns. Why? This could indicate that you have not chosen the most relevant or useful landing page.

Here are some additional reports in Analytics that you’ll definitely want to start using, if you’re not already:

1) Source/Medium. Compare revenue from all your online channels, including search engines (organic and paid search) and referrals. This is useful not only as a barometer of PPC success, but can also help identify other issues with clients’ sites. If the gap between PPC revenue and Google search revenue starts to widen too much, for example, I’ll know that my client’s SEO rankings may have started to slip. Why is this important? There is plenty of research out there to support the idea that your PPC campaigns will be bolstered by a site that ranks highly in organic searches, so you’ll want to ensure that both SEO and PPC performance remains steady. Based on my personal searching experience, I can understand why these two go hand in hand. I’m much more likely to click on a PPC ad if a site is showing at the top of my search pages out of sheer familiarity; the more times a user can see your name on a search result page, the more prominent and familiar you’ll feel to them.

A few more things about the Source/Medium report; I also use this report to drill down keyword performance across multiple search engines. You can select keywords as your secondary report dimension, and sort your data however is most useful to you (by search engine or keyword). I also use this report to identify advertising opportunities I might be missing on Google’s Display Network. If I’m getting a large number of site visits from a referring traffic source that isn’t part of another marketing initiative, I’ll target it, or similar sites, as managed placements.

2) Product Performance. Knowing how individual products are faring in terms of revenue generation and unique purchases will go a long way in helping you decide what/how to advertise in your PPC campaigns. I run this report at the beginning of each month to see what’s currently selling well, and I also run it for the past year to understand what my consistently selling items are, i.e. what I should always be pushing via PPC.

Here’s something else I like to do with my Product Performance report: add keywords as my secondary dimension. Why? I like to see that people are actually buying in relation to the keywords they came to my site/ad from. If your keywords are highly relevant to the product that was purchased, you’ll know to get more granular in your ads, focusing on specific products. If you’re noticing less relevancy, focus on more generic text and broadly focused or branded campaigns.

3) Traffic. I know this is obvious in terms of Analytics data, but I thought it was worth mentioning in relation to PPC for a few reasons. Remarketing is a very powerful tool in PPC, so if you want to gauge how powerful this could be for you, review your traffic report. How many unique visitors are you getting a month? At what stage in the buying process are users dropping off? This information will help you formulate effective remarketing campaigns.

Custom Reports

You likely know exactly what information will be most valuable to you on a regular basis for each of your e-commerce accounts, and I can almost guarantee that they’re not in the same place within the Analytics interface. You will save yourself a lot of time if you take the time to create custom reports in Analytics. I have custom reports in each of my Analytics accounts for the weekly and monthly reports I pull, in addition to separate reports for other information I reference on a frequent basis, like an ROI report that shows both cost and revenue.

Creating a custom report in Analytics is very user-friendly, and I’ve honestly never had any trouble creating them, and you can filter your reports to only include data from AdWords if you want. I am comfortable recommending a simple Google help search to get you started creating your custom reports because it’s that easy.

Cross Channel Analysis

The last thing I’d like to touch on briefly is using Analytics to view PPC performance in relation to your other marketing channels. Fortunately Analytics provides a rich, visual platform for you to analyze your multi-channel information.  If you’re in Analytics, under the Conversions menu in your left navigation, you’ll see the drop down menu for Multi-Channel Funnels:

You’ll be able to view the percentage of your conversions that are coming from your various channels in a nice circle graph, which also shows you the amount over overlapping conversions for all selected channels:

While this is a great, quick snapshot of your performance, what I find most illustrative of the impact of my PPC account on my client’s business is actually the Assisted Conversions report (also accessible under the Multi-Channel Funnel dropdown menu). Assisted Conversions, as far as PPC is concerned, are conversions in which PPC was on the conversion path, but was not the last point of interaction from which a user converted/purchased. Translation: someone clicked on my PPC ad, but made a purchase from another access point at a later point (organic search, direct, email, etc.).

Take snapshots of your assisted conversions frequently to understand the larger impact of your PPC campaigns. If your revenue is increasing from PPC assisted conversions increases over time, you’ll be able to show additional value to PPC advertising beyond your normally reported conversions and revenue data. Perhaps increasing assisted conversion revenue becomes a new goal in and of itself.

All in all, Google Analytics will arm you with all the information you need to effectively manage and understand your e-commerce PPC account, not only as a stand only marketing effort, but also as a compliment to your other marketing efforts.

Check back tomorrow for our Google Analytics Test Lab series wrap-up!