Direct Marketing Guru

Utilizing Association Rules to Improve Personalization of Offers to Customers

2011-07-12T12:17:00.000-04:00

What Association Rules are, how to build them and most importantly, how to leverage them in your marketing strategy.

The concept of Association Rules has been around since the early 1990’s and has become a core tool utilized by data miners. The process is based on the concept of discovering regularities between products in large scale transaction data recorded by systems in retail and product fulfillment. For example, the rule “(Potatoes, Onions)--> Hamburger” found in the sales data of a supermarket would indicate that if a customer buys onions and potatoes together, he or she is likely to also buy hamburger meat. Such information can be used as the basis for decisions about marketing activities such as, e.g., promotional pricing or product placements. This type of analysis is useful in developing recommendations similar to what you see when suggesting “Customers who purchased product X also purchased product Y”.

The first step it to clearly define what you what to associate. Association rules differ from Classification rules in that there are no preset outputs. So Association rules allow more freedom for unique combinations to be discovered. The most typical use is to associate baskets of product purchases, but this technique has also be used to help set pricing strategy or retail product placement analysis. Sometime the Association rules can discover some “odd” relationships.

A famous story about association rule mining is the "beer and diaper" story. A purported survey of behavior of supermarket shoppers discovered that customers (presumably young men) who buy diapers tend also to buy beer. This anecdote became popular as an example of how unexpected association rules might be found from everyday data. There are varying opinions as to how much of the story is true. While I don’t find diapers around many beer sections in stores, I’m sure some enterprising retailer has tested this at some point.

So what tools are able to perform this type of analysis? The usual power house applications from SAS and SPSS are of course quite capable of crunching through volumes of data to develop these rules. Depending upon your budget and expertise either of these tools offer solutions. If you want to keep your tool costs down you can consider a tool I recently discovered (see this link http://www.cs.waikato.ac.nz/ml/weka/) You will need to do a bit of reading to get up to speed, but as this is freeware, I have found the time to get up to speed more than offset by this tool’s very effective performance.

Once you discover the Association rules in your database what do you do with them?

I have found these rules provide an excellent foundation to build a cross media promotional offer engine. I have developed some great campaigns via email where a product that the customer had recently purchased is used to re-establish the relationship and then a couple of items that showed the highest support and confidence rankings via the association rule set were offered with minimal purchase incentives. These campaigns were set-up with test and control groups and the groups receiving the communication based on the Association rules showed a 40% lift in response as compared to the control group.

You can also create tables of product associations to feed recommendations on your website. These can be very powerful as compared to making recommendations based on less robust analytics.

This technique can also be useful in determining how to sequence product presentations in your catalogs and brochures. Just as retailers utilize this analysis to help layout merchandise in stores to be more consistent with customer shopping patterns, this analysis helps do the same for printed materials. Typically the improved sequencing of products in printed material improves the response by 5 to 10 percent.

All these actions result in improved sales, retention rates and ultimately profits.

How much of your Website Business is incremental versus channel shift from your offline efforts: A simple analytical answer along with a couple of methods to measure the quality of web buyers.

2011-04-03T09:42:00.001-04:00

Almost every multi-channel marketer is struggling with the question, “how much of my web business is incremental”? This is a key question, for marketers who have observed that as the web channel grows, typically the other traditional offline channels fall. So how can a company determine if the web sales are purely a channel shift, or might it be incremental?

My experience is to look at what portion of your web business is coming from customer who are new to your file. Obviously, if your web business is dominated by customers who have been on your file prior to the web activity, then the web channel activity is a pure shifting of business. But that is most likely what you won’t find. There will be a proportion that is new to file and the key is to compare that portion of new with your other traditional offline channels.

The table below illustrates this concept well.

New to File Proportion of Business by Channel

	Existing Buyer Business	New to File Business	Proportion existing Business	Proportion New to File Business
Offline Channels	$300	$100	75%	25%

Web Channels (Ver 1)	$25	$10	71%	29%

Web Channels (Ver 2)	$25	$20	56%	44%

Web Channels (Ver 3)	$25	$100	20%	80%

If your proportions looks like those in Ver. 1, your web business is clearly a channel shift only and should not be viewed as incremental at all.

If your proportions looks like Ver. 2, your web business is partially incremental, but your offline channels are still your primary growth source.

If your proportions look like Ver. 3, you should view your web business as incremental, and just like any marketer, you need to understand what do these new customers cost to acquire, and is that an appropriate investment. This brings up the next topic, how to measure quality of the web buyers.

Two of my previous articles will help with this task. First is to run the “Buyer Matrix” on your web buyers to understand at what level do they return to make subsequent purchases, and how do those compare to other similar aged customers. These behavior metrics will quickly help you determine if the investment made to acquire these customers was appropriate or not. It will also provide you a relative measure of quality between your customers acquired from offline efforts versus online. You should also review my previous article on Acquisition Cost to utilize the appropriate financial perspective on the appropriate investment returns.

Another useful tool is the Generic Scoring model. In a similar fashion, you should compare the scores of your new to file buyers across channels. There are many myths that web buyers are of lower quality of the offline buyer, and thus do not warrant the same level of investment. My experience is that this is not always the case. Be sure you understand how your customers behave using their actual purchase data.

Customer File Health Check: Use this simple analytical approach to gage the health of your business

2010-11-15T14:31:00.001-05:00

The health of your customer file is the key to future growth and profits

A few weeks ago I wrote about utilizing the Generic RFM equation in finding the break-even point for more efficient promotion execution. One suggestion I made in the article, for which I have received many comments and questions, has to do with “freezing” the score ranges for the deciles. In doing this you have a view of how your customer file is changing (in terms of the percent change in each decile) across time. This dynamic view can tell you a great deal about the health of your file. In this article I will show some of the most common file dynamic patterns along with how to interpret what the changes mean for future sales and profit growth.

Keep in mind that the transactions included in calculating the RFM score should include all channel activity. If you are not including all channel activity when you calculate the score the observations made in this article will probably not be applicable. As the score can be calculated back to any point in time this analytical technique is easy to employ. Just be sure there is sufficient time between the freeze date and the end date (my experience has been at least one year) such that file migrations are occurring in sufficient quantity to be projectable to the whole file.

When you run the scoring for your customer file and get the decile counts you will have a report that looks something like the table below:

Decile	Freeze	Later	% Change
1	5,000	7,000	40.0%
2	5,000	6,500	30.0%
3	5,000	6,000	20.0%
4	5,000	5,750	15.0%
5	5,000	5,500	10.0%
6	5,000	5,500	10.0%
7	5,000	5,600	12.0%
8	5,000	5,700	14.0%
9	5,000	5,900	18.0%
10	5,000	6,500	30.0%
Total	50,000	59,950	19.9%

From this you can see that the file grew by about 20%. Also the deciles at the top grew faster than the deciles in the middle, and the bottom decile grew at a rate lower from the top decile. So what would this mean for your business?

First is the good news, you are building value in your customers as the top decile is growing faster than your bottom decile. In other words, your best quality customers are growing faster than your poorest quality and typically this means your customers are purchasing more than defecting. Some of the middle deciles are not growing as fast as the top or bottom deciles. This can mean that customers are migrating quickly either to the top or bottom of the file. This suggests that the life cycle of customer may be short and you should know quickly if new customers are connecting with your brand or not. In this case you may want to focus in on a sample of new buyers who are sinking to the bottom and contrast with customer floating to the top. What are the differences? Does a product category or price point differentiate the two groups? Can anything be done to prevent the customers from sinking to the bottom?

If you see that the top three deciles are larger and growing faster than the bottom three deciles then you can conclude that the file is fairly healthy. If on the other hand you observe that the bottom three are larger and growing faster, then you have a problem. You need to understand what your brand is missing or has the competition begun to offer some product or service that you might have missed?

Understanding how the customer is migrating through deciles tells a lot regarding how the customers are reacting to your brand and offers. This type of analytics is critical when answering the questions regarding the health of your customer file.

Improve your Brand’s Perception and Position now; don’t wait for the “storm’ to pass.

2010-11-02T10:29:00.000-04:00

Some simple steps to improve and enhance your brand using some straight forward Direct Marketing Metrics.

While speaking with many business owners over the past months I have heard too many stories of the business “hunkering down” and waiting for the storm to pass. This is the best time, in fact, to spend a little time and resources to understand why your customers do business with you as opposed to your competitor and what could you do now to take more share. Such activity is not expensive, yet it has potential to pay off handsomely if executed and followed up in a methodical manner. So what am I talking about?

What could you do now to ensure your business retains and expands its leadership position in the marketplace?

My recommendation would be to first profile your customer base. This requires answers to a few key questions including:

· Why do your customers buy from you?

· What do customers think your brand stands for?

· Why did the customer purchase from you as opposed to one your competitors?

· How many of your customers intend to purchase from you in the future?

These profiling questions would best be answered at your customer segmentation level (i.e. Best Customers versus Marginal Customers). To create such segmentation please refer to my article on the Generic RFM Scoring methodology.

If you already have this kind of information from customer primary research studies you are well on the way to securing your position in the marketplace. If not, I have the expertise to get this process moving for you if you wish.

Having years of experience analyzing customer transactions in marketing databases I would then want to analyze the following marketing metrics for the buyers profiled above:

How many customers purchased from you in the last 12 months?
How many times did they order and what was the average order size?
How many new customers did you acquire last year?
How many of last year’s new customers are you expecting to buy this year?
How have these metrics changed in the past few years?
What marketing programs have the most impact on these metrics?

Answers to these metrics questions coupled with the customer profiling above would give your brand a clear road map of how to grow the customer base at rates much greater than your competitors. While at the same time you would have improved information on key metrics driving your business. The metrics are critical in setting appropriate benchmarks for ROI goals on your marketing programs. In that way you will have a better and timelier understanding of what is working and what is not. Then you can expand the programs that work and fix or stop programs that are not working.

Doing these things now is critical, especially if your competitors are “hunkering down” and waiting for the storm to pass. For when the competitors raise their heads again you will have secured a more defensible position in the marketplace from which they will be required to play catch up, which almost always is a more difficult and costly exercise.

Customer Retention: How to improve it utilizing Customer Behavior Analytics

2010-10-19T09:34:00.000-04:00

Understanding how often your customers repurchase items can yield large paybacks in terms of sales and profits

Many companies are not aware they are sitting on a gold mine of sales and profits and all they have to do is look for it. Over the years I have analyzed customer purchase behaviors and in many instances found opportunity to capture more sales, improve retention rate all while increasing profits. Does this sound too good to be true? The key to identifying purchase behavior that could be cultivated into one of the most profitable marketing programs is to look at item sales across time to identify if the same customer purchases the same item multiple times. The easiest method of identifying such behavior is to list item by customer number who purchased. If it is observed that over twenty percent of your items are repeat purchases from the same customer, you’ve struck gold.

Many companies have this behavior hidden in their transaction data and are not aware of its value. I have found such behavior at both B2B and B2C companies. The majority of this buying behavior can be associated with products that are “consumed” or regularly replaced. Some good examples of consumable products include paper supplies, business forms, annual replacement items, nutritional supplements, filters, equipment that fails after use and numerous other examples. If in doubt, the basic analysis is straight forward and should be done, it only takes a small percentage of your items to generate impressive ROI on the resulting program.

So let’s say you found your product portfolio contains consumable products, now what? The next step is to find what the minimum time interval is between purchases. For illustration purposes, you find that the average time between purchases is nine months. The key is to contact the customer prior to the average replacement window and secure the reorder as opposed to allowing the customer an opportunity to explore alternative sources for the product.

In this example, I will use some actual data from a B2B marketer I worked with who sold supplies which have a clear replacement cycle as the products ultimately fail with repeated use. In this study there are 4,100 customers who purchased one (or more) of seven similar items used to deliver services to their customer. These customers had purchased one of the items within a window of a minimum of 6 months and a maximum of 24 months ago. For this test 2,950 will receive a special communication regarding the item replacement while another 1,150 are held out as a control group to measure the incremental lift.

The communication’s content should be personalized to the recipient such that the purpose and action is clear. For this test the communication contains a picture of the product purchased, date of purchase and a brief explanation that the product’s useful life may soon expire, and photos of products with various component failures are shown to help the customer evaluate the remaining life of the item. The customer is urged to examine the product and if any signs of potential failure are found, a replacement should be ordered immediately. The communication was sent via snail mail, but the technology has advanced quickly and this customized communication can be sent via the customers preferred channel of communication. The control group continued to receive all the regular communication except for this replacement notification.

The performance of the test and control groups is documented below.

	Control - Total Qty 1,150		Test - Total Qty 2,950

Product	Data	Total	Data	Total
Category	Buyers	103	Buyers	330
Performance	Sales	$19,359	Sales	$54,460

	Sales per Member	$16.85	Sales per Member	$18.45
	Response	8.96%	Response	11.18%
			Sales Increase	9.5%
			Response Increase	24.7%



SKU	Data	Total	Data	Total
Performance	Buyers	48	Buyers	207
	Sales	$5,109	Sales	$20,185

	Sales per Member	$4.45	Sales per Member	$6.84
	Response	4.18%	Response	7.01%
			Sales Increase	53.8%
			Response Increase	67.9%




Overall	Data	Total	Data	Total
Performance	Buyers	243	Buyers	762
	Sales	$90,607	Sales	$338,106

	Sales per Member	$78.86	Sales per Member	$114.53
	Response	21.15%	Response	25.81%
			Sales Increase	45.2%
			Response Increase	22.1%

Not only did the effort result in elevated levels of customers reordering the items included in the communication, but additional items in the product category and products outside of the category were purchased too. Overall retention was improved by 22% and with an increase of average order value sales increased by an average of 45%. These are actual results achieved, not all programs will have such strong results, but I have always seen this program deliver acceptable results when implemented as outlined. For some businesses that have a highly consumable product portfolio this type of program typically becomes the mainstay of top line sales and bottom line profits.

Acquisition Cost: How to calculate it and how to determine what your business can afford

2010-10-04T12:01:00.000-04:00

Understanding what it costs to acquire a new customer and how long it takes to break-even on the investment is a critical step to growing your business profitably.

In an earlier post I spoke of the need to acquire new customers in order for your business to grow. Understanding how much those new customers cost to acquire and how long it will take to recoup your investment is the key to growing your business profitably. So how is it done?

Going back to the Buyer Matrix post, you will need to identify all new customers acquired during a given time frame (I typically use a year of activity) and attempt to link them back to the promotion that brought them in. If you have a large portion of your first orders with accurate key code capture this is an easy task. Most companies are not as fortunate and some analytics needs to be conducted to get customers associated with the promotion that converted them to customers. One common technique is to do a match back. This process requires a file from your merge purge house of all prospects who were promoted during the period you are measuring. You then match all your new customers against this file, regardless of what channel they converted through. (I wonder if JC Penny’s did this prior to deciding to eliminate all their catalog mailings.) If your new customers have a valid source code use that as your source, if not, and the customer hits the match back file, use the list from the match back as the source, if the customer still does not meet either of those criteria label it as “miscellaneous” and leave it in the channel converted through.

You will also need some additional information including:

Marketing costs by channels and specific effort – How much you spent on catalogs, Direct Mail Pieces, List Costs, Printing and postage, pieces or units sent, PPC, Banner ads, Email.
Gross margin – what was your cost of goods sold, try to get this down to the order if possible, if not, use the average
Fulfillment costs – what it costs to take an order and get the product to the customer, If your S&H is done at a profit, take this off the this cost, if it is done at a loss add it in.

You now have enough information to begin the calculation process. You should have the new customers broken down into what promotion/channel the order was placed through (i.e. phone, web, email, SEO, PPC and other). Get your total new customers counts down to specific effort and then any miscellaneous new customers in that channel get allocated across all that channel’s efforts. You now have information telling you what channels and specific promotions delivered customers during the analysis period. Now it is time to calculate Acquisition cost. See the example below.

Acquisition cost Calculation example

100	Acquired Customers

Revenue	$12,500	Gross Revenue generated (does not include S&H)
AOV	$125	Average Order Value
GM%	50.0%	Gross Margin percent
COGS	$6,250	Cost of Goods Sold
Marketing	$10,000	Marketing Costs to acquiring these customers
Unit Cost	$0.80	Marketing expense divided by units sent
Response	0.80%	New Customers divided by Units sent
Fulfillment	$10	Cost to take and ship order
Fulfill Cost	$1,000	Total Fulfillment Cost
Profit(Loss)	($4,750)	Cost to acquire these customers
Acquisition	($47.50)	Cost to acquire one customer

The example shows an acquisition cost of $47.50 per customer for this promotion. The more granular you can get this calculation, the more actionable your efforts. By this I mean if you can get this down to specific lists or search terms the more you will be able to evaluate that specific effort. If you can only do this at a channel, then you can’t evaluate specific efforts within that channel.

Now you need to determine if $47.50 is an acceptable acquisition cost. To do this you need to utilize the Buyer Matrix to understand future behavior patterns of your new customers. Typically I model customer behavior by channel the customer came in on and then compare those behaviors to see if there is any material behavior difference across the various channels. The key to this kind of statistical comparison is to have sufficient customers in each channel to reduce “statistical noise”. A few hundred customers is my rough standard for this comparison exercise.

The example below shows the behaviors for new customers across a few years of activity. I only use two years since you should have a positive return on your acquisition invest in that time line. If the acquisition investment takes longer than that, you are probably spending too much to acquire customers.

NTF Count	100
Same year Orders	30
AOV	175
Margin	0.5
Fullfilment	10
Demand	$5,250
Cogs	$2,625		Cost/Unit	Frequency
Marketing	360		$0.80	4.5
Fullfil	300
GM	$1,965
Per FTB	$19.65

Year 1
Pct. Return	0.35	35	Customers Returning
Orders Per	1.5
AOV	189
Demand	$9,923
Orders	53
COGS	$4,961		Cost/Unit	Frequency
Marketing	$720		$0.80	9
Fullfil	$525
GM	$3,716
Per FTB	$37.16

Year 2
Pct. Return	0.8	28	Customers Returning
Orders Per	1.7
AOV	189
Demand	$8,996
Orders	60
COGS	$4,498		Cost/Unit	Frequency
Marketing	$576		$0.80	9
Fullfil	$595
GM	$3,327
Per FTB	$33.27

From the above illustration in the same year acquired an average of $19.65 profit per customer is earned. Over the next year a total of $37.16 is earned per buyer. Both these figures are averages across the 100 customers being analyzed. So with an acquisition cost of $47.50, the payback occurs at about the ninth month of the next year.

This gives you the basis from which to begin evaluating various channels and promotions in terms of their respective effectiveness at delivering new customers at an acceptable cost. Obviously if the investment cost cannot be recouped in two years, you are probably spending too much. If on the other hand, if the costs are recouped in less than a year, you should probably be looking for ways to expand the volume and investments, without raising the costs too fast.

This type of balanced approach to your acquisition efforts will help grow your customer base and sales while generating sufficient profits to keep the growth engine running smoothly.

First Time Buyer Retention: Why it is so important and how to improve it

2010-09-25T09:10:00.001-04:00

Improving retention rate of your new customers can have a huge sales impact; I’ll even share a program that has improved new customer retention at both B-to-B and B-to-C companies.

Each year your acquisition efforts need to bring in customers sufficient to replace those that are lost. If you want your sales to grow, then the number of new customer should exceed the number lost. So if you’re losing large numbers of customers you are even more dependent on acquiring more customers just to stay even. Understanding your retention issues becomes obvious when you think about it.

When you analyze the customers who were lost you will probably find what I have, the single largest cohort group is the new customers acquired in the prior year. This loss of new buyers can be quantified by utilizing the Buyer Matrix process outlined in an earlier post. Obviously the fewer new buyers lost each year can ultimately translate into a sales increase, as opposed to a sales decline.

From the Buyer Matrix article I want to focus on just the 100 customers that were acquired in 1999. In 2000, 50 of the 100 customers returned to purchase, the retention rate for first time buyers was 50%. (see the first table below) What would the impact be on sales if you were able to improve the retention rate from 50% to 55%, 60% or 65%? The analysis below shows the impact of improving retention by these increments in terms of the cumulative 4 years of customer sales for the total 100 customers acquired in 1999. The incremental sales are only from the additional customers retained in 2000. The second table below documents just the 65% returning rate and the summary figures are below that.

Buyer Matrix Forecast
		Buyers who placed an Order
Year of First Order							Forecast
		1999	2000	2001	2002	2003	2004
	1999	100	50	40	32	24.992	19.49376
	Returning Pct.		50.00%	80.00%	80.00%	78.10%	78.00%
		Orders Per Buyer

		1999	2000	2001	2002	2003	2004
	1999	1.2	1.5	1.5	1.6	1.7	1.7
		Average Order Value

		1999	2000	2001	2002	2003	2004
	1999	$90	$100	$101	$102	$103	$104	Total
		Sales						Sales
								2000 to
		1999	2000	2001	2002	2003	2004	2004
	1999	$10,800	$7,500	$6,060	$5,222	$4,376	$3,446	$26,605

Forecast based on 65% Retention
		Buyers who placed an Order
							Forecast
		1999	2000	2001	2002	2003	2004
	1999	100	65	52	42	32	25
	Returning Pct.		65.00%	80.00%	80.00%	78%	78%

		Orders Per Buyer
		1999	2000	2001	2002	2003	2004
	1999	1.2	1.5	1.5	1.6	1.7	1.7

		Average Order Value
		1999	2000	2001	2002	2003	2004
	1999	$90	$100	$101	$102	$103	$104	Total
								Sales
		Sales						2000 to
		1999	2000	2001	2002	2003	2004	2004
	1999	$10,800	$9,750	$7,878	$6,789	$5,689	$4,480	$34,586

						Incremental
				2nd Year	Total 2000	Sales per
				Retention	to 2004	Customer
				Rate	Sales	Promoted
				50%	$26,606	N/A
				55%	$29,265	$53.18
				60%	$31,926	$106.39
				65%	$34,586	$159.60

Thus if a retention program were to improve the retention rate from 50% to 65% the sales from the group would increase from $26,606 to $34,586, a thirty percent increase across 4 years. The incremental sales per customer promoted figure is only for the customers who purchased in 2000 that would not have unless the retention program was implemented. The total 2000 to 2004 sales from those additionally retained customer is divided by 50 (the number of customers included in the retention program who had not purchased a second time yet). The analysis indicates there is a significant upside to sales as the retention rate increases. The COGS should also be considered to understand the gross margin generated. The next key question is what programs could a company implement to increase the new buyer retention rate?

Over the last twenty years I have implemented programs to improve the new buyer retention rate and the most successful is to call customers who have not yet placed a second order yet. The call should be timed to happen within three months of the first purchase so that the chance of remembering the order is improved. The calls should be prioritized to the largest spenders first as these buyers have shown to have the greatest future value. The call should be scripted along the lines of customer service representative checking in to make sure the customer’s experience, both in terms of the service provided and the product received, lived up to their expectations. Most calls will result in the customer indicating their experience was great, and that comment should be followed-up with an offer to place another order with some form of a time-limited incentive.

On those calls where the customer feels you did not meet their expectation, you now have the chance to make it right. This can be critical to saving the customer and should be done quickly, preferably while you have the customer on the phone. I have done follow-up analysis of customers who had an issue resolved by customer service and found the retention rate and overall future activity was better than customers who did not report any problem. Over the years this program described above has improved new buyer retention from 10% to 30% with a cost per customer contacted of about $10. The return of about $160 (assuming the 30% improvement) on the $10 cost provides a respectable ROI!

When running such a program you will need to create test and control groups to obtain the appropriate measurement of retention improvement. Refer to my article on Horizontal Measurement for more details.

This program, when consistently rolled out across time, can deliver substantial sales with a very high ROI. When considering the amount invested to acquire new customers, the ROI becomes even more attractive. Having the dialog with your customers will also provide qualitative insights as to what is important and how those preferences may be shifting with customers. It is also a great method to find their contact preferences in terms of Email, mail and outbound.

All this activity will go a long way to maximizing the Long Term Value of your customer file and ultimately improving sales and profits.

Customer File Segmentation Simplified: Generic RFM Scoring Equation

2010-09-16T11:13:00.001-04:00

This approach is easier to execute and more effective than traditional RFM cell segmentation.

When direct marketers send offers and promotions to customers on their database, there needs to be some segmentation done to rank which customers should receive the communication and which should not. Typically for many direct marketing companies this involves the use of RFM cells to segment customers. This methodology looks at customer (R)ecency or how long since the customer last ordered, (F)requency or how many timed they have ordered and (M)onetary or how much they have spent. The cells are are then defined by these three attributes. So you might have a cell of customers who purchased within the last six months, ordered 2 times and have spent over $200. As the cells get more granularly defined, the number of cells increases dramatically. For example, if you use five categories each to describe recency, frequency and monetary, you will have a total of 125 cells to wrestle with. Another issue with RFM cells is it is not always easy to sort the cells from best to worst probability of response. As a result it can be difficult to evaluate at what point customers should not be sent the offer. While this methodology is core to segmenting a customer file there are issues associated with its use as described above.

A variant of the cell segmentation is to use an equation to convert the RFM information into a score. The score can then be used in the same way to sort customers from most likely to respond to least likely. But unlike the cells, the customers can now be sorted into “deciles” allowing for efficient determination of who should and should not be promoted. The equation that I have used numerous times is as follows:

Score = (R)^{-1 *} (F+1) * (M+1) ^.5

Notes:

R = Months since last purchase

F = Number of times purchased last 12 months

M = Cumulative dollars purchased across life

Example of use:

Customer placed one order in last twelve months, ordered 7 (seven) months ago, and has lifetime purchases of $500.

Score = (1+1) * (7)^-1 * (500+1) ^.5

= 2 * 0.142857 * 22.38303

= 6.395151

Customer has not ordered in last 12 months, customer ordered 24 months ago, lifetime purchases of $250.

Score = 1 * (24) ^-1 * (250+1) ^.5

= 1 * 0.041667 * 15.84298

= 0.660124

Once the customers have been scored, then next step is to sort them into deciles. This is done by sorting all the customer scores from highest to lowest, and noting what score range represents the top 10% of your file. Then do the same for next 10% until you reach the bottom. Then use these ranges to assign customers to the corresponding decile that their score falls into. Then each time you score the customer file use the same ranges to assign customers to their decile.

You can back test to find the appropriate break point to stop promoting. Once this method has been used for a number of mails, you will gain insights as to what decile you should mail to depending on seasonality and offer. I have found this methodology works almost as well as customer regression modeling, but can be implemented quickly (days) and costs nothing to develop other than some simple code writing. In addition it greatly simplifies the extraction process of customers to include in your mailing while also providing easier reporting of response performance versus expectation.

Feel free to contact me at this blog if you have questions on implementing this simplified segmentation approach.

Gaining a Better Understanding of Your Customer’s Buying Behaviors: The Buyer Matrix

2010-09-10T15:52:00.003-04:00

The Buyer Matrix helps you to develop marketing campaigns that will build your business and improve profits, while helping you to forecast sales in future years. This technique is available to any business that keeps manual or computer records of each customer’s transactions.

How many customers purchased from you last year? On average, how many orders did each of those customers place? What was the average value of those orders? How many new customers did you acquire last year? How many of them do you expect to buy from you this year? What have been the trends of these metrics for your business? What marketing programs are most effective in improving these trends? Based on these metrics, what should your sales be next year?

Marketers are frequently asked to forecast sales, and identify what marketing programs can improve sales. But the process of forecasting sales and projecting marketing program results has been generally approached as an “art”. While there is always some level of uncertainty in forecasting sales, the Buyer Matrix methodology will help to reduce many of the errors associated with predicting future business.

The Buyer Matrix was developed out of the need to understand how key customer behavior trends are related to future sales. There was also a need to measure what effect Marketing Programs had on sales, and if those programs were delivering sufficient sales to justify their expense.

The Buyer Matrix is based on the following principal: if you can predict how many buyers you will have in the future and how much will they spend, you should be able to forecast what your sales will be. But there is a great deal of variability in predicting buyer counts and their behaviors. After years of experience I have found the best method of projecting buyer counts is to assign your buyers into unique cohort groups and measure the group’s behavior across time. Experience has shown that cohort groups defined by the date a customer first purchased from your company works very well. Each of these cohort groups exhibits a unique predictable behavior pattern. When all the cohort groups are rolled up, a good estimate of the number of buyers for the forecast period is available. Let’s examine a specific example to help understand this concept.

ABC Company was started in 1999. The company sells supplies to other business across the United States. All transactions are captured in an order entry/shipping system. In 2002, management of the ABC Company increased the advertising budget dedicated to acquiring new customers by 20%. As the business continued to increase in 2002, management increased the budget for 2003 by another 20%. Management was forecasting another banner year for sales in 2003, a gain they based on the ever-increasing sales volumes since 1999. In late 2003 however, the company suspected they were falling short of expectations, and did not know why.

In early 2004, in an effort to understand what was happening to the business, the company utilized the Buyer Matrix methodology. The transactions were extracted from their system and the following Buyer Matrix report was created:

Year of	Buyer Placed Order
First Order	1999	2000	2001	2002	2003
1999	100	50	40	32	25
2000		110	55	44	35
2001			110	55	44
2002				100	50
2003					95
Totals	100	160	205	231	249

The matrix report shows that in 1999, the company acquired 100 new customers. In 2000, the company acquired 110 more new customers, and 50 of the customers who were acquired in 1999 returned to purchase. By the end of 2001, ABC now had 205 total buyers, consisting of: 40 from the group acquired in 1999, 55 from the group acquired in 2000 and 110 newly acquired customers in 2001. In 2002, total buyers grew to 231, consisting of 32 from the 1999 group, 44 from the 2000 group, 55 from the 2001 group and 100 newly acquired in 2002.

The matrix shows the volume of acquired customers fell as the acquisition-marketing program was rolled out in 2002 and the negative trend continued into 2003. Management did not realize the customer acquisition program was not working effectively until early 2004, and by then the acquisition budget for 2003 had been exhausted. Profits were negatively affected in 2003 as a result of not having a measurement system in place.

Many companies have adopted loyalty programs (rewards programs), but without a Buyer Matrix mechanism to measure historical customer retention rates and measure changes in those rates, companies have no way of knowing if a loyalty program is effective and delivering value. If the hypothetical ABC Company started a new customer loyalty program in 2002, with the goal of increasing the number of buyers who have had activity since 1999, would the same matrix suggest the program was effective? By using the same Buyer Matrix report, and reading it differently, the answer can be found.

The 1999 cohort group had 50 customers return in 2000. Then in 2001 only 40 customers returned to buy, a fall-off of 20%. Then in 2002, when the loyalty program was initiated, the 1999 cohort group fell to 32 buyers, another 20% fall off. The program did not change the buyer behavior pattern in 2002. In 2003, the trend was still not affected by the loyalty program as the buyers continued to fall off by over 20%. These metrics were the same for the 2000 and 2001 cohort groups, indicating the loyalty program was not effective. What if the loyalty program’s goal was to increase the activity of the buyers as opposed to the number of buyers, would the program then be effective?

In the buyer matrix below, the measurement has been changed from buyer counts to orders per buyer:

Year of	Orders per Buyer
First Order	1999	2000	2001	2002	2003
1999	1.20	1.50	1.50	1.60	1.70
2000		1.20	1.50	1.60	1.70
2001			1.20	1.50	1.60
2002				1.20	1.50
2003					1.20
Totals	1.20	1.40	1.45	1.55	1.65

In 2002, the order per buyer for the 1999 cohort group increased from 1.5 orders to 1.6. The trend continued in 2003 with orders per buyer increasing to 1.7. The 2000 and 2001 cohort groups show similar order per buyer gains too. These measures show the loyalty program is working, assuming the goal is to increase orders per buyer with the existing customers.

Another key metric in the Buyer Matrix is the average purchase amount. Similar to the Orders Per Buyer report above, this report shows the trends of the average purchase amount across time for the respective cohort groups. This is a useful metric for marketing programs designed to increase the average order value, such as premium and discount offers.

When these three key metrics are combined, you have a customer behavior based sales forecasting tool. Sales are broken down into the equation: Sales = Buyers * Order per Buyer * Average Order Value.

For a marketer to make an accurate sales forecast, the buyer counts, orders per buyer and average order value metrics associated with each cohort group are forecast out. The assumptions associated with each group’s specific behaviors should be a reflection of past performance for that specific metric, adjusted for any marketing campaigns designed to affect the metric. In adjusting for campaigns, behavior trends should not be changed significantly beyond historical trends. The illustration below should help to clarify the process.

2004 Forecast based on Buyer Matrix
		Buyers who placed an Order
Year of First Order							Forecast
		1999	2000	2001	2002	2003	2004
	1999	100	50	40	32	25	20
	Returning Pct.		50.0%	80.0%	80.0%	78.1%	78.0%
	2000		110	55	44	35	28
	Returning Pct.			50.0%	80.0%	79.5%	79.5%
	2001			110	55	44	35
	Returning Pct.				50.0%	80.0%	79.0%
	2002				100	50	40
	Returning Pct.					50.0%	80.0%
	2003					95	48
	Returning Pct.						50.0%
	2004						100


		Orders Per Buyer
Year of First Order
		1999	2000	2001	2002	2003	2004
	1999	1.2	1.5	1.5	1.6	1.7	1.7
	2000		1.2	1.5	1.6	1.7	1.7
	2001			1.2	1.5	1.6	1.7
	2002				1.2	1.5	1.6
	2003					1.2	1.5
	2004						1.2

		Average Order Value
Year of First Order
		1999	2000	2001	2002	2003	2004
	1999	$90	$100	$101	$102	$103	$104
	2000		$95	$100	$101	$102	$103
	2001			$97	$101	$102	$102
	2002				$97	$101	$102
	2003					$97	$101
	2004						$97

		Sales
Year of First Order
		1999	2000	2001	2002	2003	2004
	1999	$10,800	$7,500	$6,060	$5,222	$4,378	$3,448
	2000		$12,540	$8,250	$7,110	$6,069	$4,872
	2001			$12,804	$8,333	$7,181	$6,027
	2002				$11,640	$7,575	$6,528
	2003					$11,058	$7,196
	2004						$11,640

Total Sales		$10,800	$20,040	$27,114	$32,305	$36,260	$39,711

ABC Company’s sales forecast for the next year is $39,711, assuming all key behavior metrics are consistent with 2003 historical rates. If the company developed a new customer acquisition program they believed would work substantially better than the old program, adjusting new customers acquired to 120, instead of the 100, might be appropriate for 2004. These kinds of what-if assumptions can be easily analyzed to understand the possible value of new marketing initiatives. The matrix provides key measures to goal marketing programs against, providing companies with an early indication of a program’s success, or the need to re-think it.

The Buyer Matrix is a proven mechanism that has helped many companies understand what marketing programs are effective, and how those programs impact overall business. The matrix cohort groups can be based not only on date of first purchase, but also on any easily identified characteristic that is a differentiator of your business. Some other cohort grouping include:

First product purchased (can be a product line or a specific product)
Source of the new customer (could be from certain ads or web site)
Promotional customers (purchase from sale or discount offer)

When this technique is combined with in an in depth understanding of why customers purchase from you, top line performance is maximized, and ensures marketing program investments are delivering sales consistent with expectations.

The news that the web is dead is grossly exaggerated!

2010-09-08T13:17:00.000-04:00

Recently I saw a post on the NEMOA Linked In chat board with the headline “The web is dead, long live apps”.

Remember back in the mid 90s when the mantra was “Catalogs are dead, get on the web”? While catalogs have lost ground to web, they are clearly not dead. In fact, many DM companies still have the majority of their sales generated through catalogs. So what’s one to think now that the “demise” of the web is around the corner?

Clearly, just like the catalog, the web’s death prediction is a bit of an exaggeration. What’s really happening is customers have yet another option as to how they will interact with your company. What is critical for any marketer is to understand what communication methods your customers prefer to utilize and what they expect to get from them. If you find through primary research that your customers are using mobile apps to purchase your products, then you should really be looking at building that. If on the other hand, your customers are using apps for other non-business informational purposes and prefer to do business with you via web or catalog, then development of apps may not be the way to go at this time.

The bottom line is to have your customers decide what are the ways to interact with your company, and not be seduced by cool new technology.

Remember, “Just because you built it, don’t assume they will come”.

Promotional Measurement - Horizontal vs. Vertical Analysis

2010-08-29T16:59:00.000-04:00

Using Horizontal Analysis to Better Understand Promotional Test Results: Get the Complete Picture

Measurement of Strategic Marketing Promotions should include a comprehensive view (horizontal), along with the traditional source coded analysis (vertical). A Marketing Program should not be rolled out before you have a complete understanding of how customers react.

When you run a promotion and measure the performance, do you include all customer activity, or only the activity associated with the promotion code? The easiest and most common method is to measure only those purchases generated by the promotion codes. But, by only considering these purchases in the program’s assessment, you may only be measuring a small portion of the customer’s overall purchases. In addition, customer activity before and after the promotion may be critical to your determination of the program’s effectiveness. Customers may also have purchase activity during a promotion that is not associated with the specific promotion. This can also materially change the determination of a program’s effectiveness.

Including all activity, regardless of when or how the purchase was made, increases the accuracy of measurement of customer behavior. Analyzing customer activity horizontally typically requires more up-front work. The complete view of all customer purchases through horizontal analysis greatly improves the chances of identifying programs that make real significant changes of customer behavior.

Vertical analysis traditionally includes only purchases within the promotion activity timeframe. These are called source-coded orders, as a special promotion code is associated with the purchase. Asking the customer to give a special code or providing a coupon at the POS accomplishes this. Horizontal includes not only the vertical purchases, but also all other purchases within the promotion, along with pre-promotion and post-promotion purchases. Let’s examine a specific example to see the implications of these two analytical approaches.

A direct marketing company sold products through multiple sales divisions to businesses across the United States. One of its divisions, which sold a highly seasonal product line, was testing a Marketing Program designed to increase the number of customers returning to purchase the product by offering an “early order discount”. The test results indicated that the number of customers returning to purchase was significantly higher for the customer group receiving an early order discount when compared to a similar group that did not receive the early discount. The analysis raised several questions including:

Did the results look at all customer activity (horizontal), or only the orders whose promotion codes matched the test promotions (vertical)?
If the test results were so positive, why did overall existing customers return at the same rate as the prior year and not some higher rate?
Did the early discount test group place a higher percentage of their orders earlier than the non-discount group?
Should the Early Discount Program be rolled-out to all customer next year?

The table below shows the coded results for the test.

Early Discount Test Results - Source Code based

	Description
Data	Discount	No Discount
Group Size	7,000	7,050
Total Orders	816	401
Response Rate	11.7%	5.7%
Sales	$303,270	$167,640

With a lift of 105% (11.7% versus 5.7%) in response rate, the sales team felt the program was a huge success. The results were suggesting that a significantly higher proportion of customers who received the early order discount had ordered. Even though the company had lower margin on each order to the discount group, the sales team believed the margin loss was more than made up by the overall increase in business. But the results were based on source code results only. The company was a direct marketer and each mail piece sent to customers had a unique source code that was printed on the order form or requested from the customer by the telephone sales representative at the time the customer placed their order. The sales team believed that such a large lift in response had to be real and significant. But what was not known was if there was any systematic bias in the way this information was captured from the customers that may have skewed the results of this test. Only a horizontal view would answer the questions and give accurate answers to such an important questions.

Customer numbers were obtained for all individuals in each of the two groups from the mail house that had processed the promotional mailing test. The activity of each of those customers for the specific seasonal product line was extracted from the order entry system. The data was then summarized into the following report.

Early Discount Test Results - Horizontal based
	Group
Data	Discount	No Discount
Group Size	7,000	7,050
Total Order	4,129	4,035
Response Rate	59.0%	57.2%
Sales	$1,594,035	$1,600,931

This process yielded very different results. The response rate lift was now only 3.1% as compared to the 105% in the coded analysis, and the sales generated per person mailed was almost a dead heat. These results explain why the division had not seen an overall increase in the buyer retention rate. The promotion had not generated much incremental buyers, and it certainly did not generate any incremental sales. And the ROI was lower for the discount group as a result of the reduced margin.

These results were coupled with one-on-one interviews with customers of the division. When customers were asked what factors were most important in their decision to purchase, it was product and service attributes that were universally cited by customers as most important, and pricing was only occasionally mentioned.

The timing of the order flow between the two groups was also examined using a horizontal view to understand if the early order discount had moved the proportion of early orders for the discount group versus the non-discount group. The results are shown below.

Timing of Buyers between Discount an No-Discount Groups
			Month

Group	Data	1	2	4	5	6	7	8	9	10	11	12
No Discount	Buyers in Month	42	18	2	3	32	222	288	468	932	1,500	483
	Cumulative Buyers	42	60	62	65	97	319	607	1,075	2,007	3,507	3,990
	Cuml. Pct.	1.1%	1.5%	1.6%	1.6%	2.4%	8.0%	15.2%	26.9%	50.3%	87.9%	100.0%
Discount	Sum of Buyers	56	23	3	2	22	196	272	503	936	1,483	543
	Cumulative	56	79	82	84	106	302	574	1,077	2,013	3,496	4,039
	Cuml. Pct.	1.4%	2.0%	2.0%	2.1%	2.6%	7.5%	14.2%	26.7%	49.8%	86.6%	100.0%

The early ordering discount had no effect on moving orders up for the group.

After review of the horizontal results, the decision was made not to roll out the early order discount program. If the horizontal results had not been generated, the company would have probably rolled the program out, and would have been wondering why more customers were not coming back to place orders the following year.

This situation brings to light another key ingredient for any Marketing Program analysis. When testing new Marketing Programs, be sure you know what the performance hurdles are up front, and make sure you have the means with which to measure those hurdles. Many companies develop the hurdles and measurement mechanisms after the test is over. There is usually a heated debate about the results and how to interpret them, and many times the roll out decision is made with incomplete information since a deadline is at hand. Don’t let this happen at your company. If you spend the time to set up measures and hurdle rates, you will make better and timelier decisions about your marketing programs.