eCommerce Analytics

One of the toughest problems facing organizations with many customers and large numbers of transactions is determining when a customer has left the company for good. Unlike contractual businesses where the customer is locked in for a given period of time, most companies have to make educated guesses about whether or not a customer is worth spending their marketing dollars pursuing.  While there are complex data analysis models that can be more than 80% accurate in predicting when a customer will buy next, many times a quick and dirty calculation can give your marketing executive a better indication of who to contact in the future.

We sometimes use the following metric to give an initial suggestion of whether or not a customer is still active in a given period. To clarify, period here can mean any observable time length - days, weeks, months, even years can work depending on your type of business and how long your typical customer stays engaged:

Probability (Active) = (T/N)ⁿ

where:

T = last period when the customer made a purchase

N = total periods in your observation time

n = total number of purchases during observation time

To give you an example of how it’s used, let me take the following example out of the book Managing Customers for Profit: Strategies to Increase Profits and Build Loyalty by V. Kumar (highly recommended) to illustrate the power of its simplicity. Let’s say I have two customers with the following purchasing pattern over a twelve month time horizon (X’s indicate when a customer made a purchase):

The probability for each customer to purchase in time period 12 is calculated as follows:

Customer 1 = (7/12)⁴ = 11.58%

Customer 2 = (8/12)² = 44.44%

The initial results may seem counter-intuitive - why does Customer 1 have a lower chance of purchasing again when he/she has 4 purchases in the observation time while Customer 2 only has 2 purchases? Well, as you can see from the table, Customer 2 appears to buy every 4 months and is poised for another purchase in month 12 while Customer 1 was purchasing every 2 months, but hasn’t made a buy in over 5 months so is unlikely to buy again in month 12.  Nice, huh?

While the above metric doesn’t take into account all important information, it does provide a fast and easy way for marketers to get a better indication of whether a customer should be put into the active marketing group or the re-activation group.

Someone asked me the other day, in response to my assertion that one-to-one marketing on a massive scale was the wave of the future, how a company could possibly send out so many personally-tailored emails. Being in the local Irish Pub, The Burren, I almost laughed Guinness out of my nostrils. But I couldn’t avoid the underlying message. One-to-one marketing never really has been embraced because no one really thinks that they do customer segmentation very well, that there are too many obstacles to customer segmentation for it to be entirely useful. Ultimately, this means that few believe they have homogenous enough segments to deliver the personalized goods.

What this also means is that one-to-one marketing is complex due to the fallacies of profiling. I once worked at a company that had such in-depth profiles for each segment that the profiles read like a Faulknerian novels. At this company, I learned that our target female customer in the 35-40 range probably once wanted to visit France but was now stuck with two kids in middle America and made meatloaf once a month for a husband she rarely saw. She obviously consoled herself by buying our software.

What’s my point with all this? This kind of profiling is for low-transaction sales, nothing more. Direct marketing units with high transaction rates should never take the tack of email blasting a segment based on their demographics. Nevermind writing fanciful biographies for said segment. Instead, direct marketing should ignore demographic profiles and concentrate on profiles that accomplish an immediate business goal (see below). Given the immediate needs that direct marketing normally serves, it needs to have a shortsighted, tactical approach, not the strategic approach that profiling represents. Below I look at the goal of getting rid of an overstock of shorts via the email channel. In doing so, I explore two, important dimensions of personalization: what the segment is willing to buy cross-referenced by when that segment most likely opens email.

The Group that Will Likely Buy Shorts Next

The truth is, the customer is not out there to buy from your company. They’re out there to purchase the product they want next and you’re merely there as a direct marketer to insinuate yourself into the buying equation. So which segment of your customers is likely to buy shorts next because that’s the group you want to reach when your shorts have been sitting in inventory for way too long and the leaves are already falling from the trees. So is this a profiling problem? In other words, is it time to blast every demographic who might wear shorts. I suppose you could. But then you’re likely to turn some people off. If you ran your customers past transactions through a classification data mining task, what you’d come up with is a list of people who are likely to buy discounted shorts at that time of the year. In fact, you’d probably come up with a few segments that demonstrate such a propensity. And they would definitely cut across your demographic profiles. You’ll have some moms buying shorts for their sons and some dads buying shorts for next summer’s Hawaii trip.

When Is the Best Time to Reach My Shorts Group?

Almost everyone out there sends me email blasts on Tuesdays and Thursdays. Why? Well, the general belief is that it adheres to the customers’ work/open schedule. I have seen elsewhere that most emails are opened on Sundays. That’s a compelling argument. But I tend to believe that each of your potential shorts purchasers has a more personalized schedule as to when they open and read emails. And that leads to the answer to the question in the subtitle. There are many times to best reach your customers who will buy your clearance shorts. The best web article I’ve read on this is by Bill Nussey of silverPOP who argues for tuning your send times per customer based on their last-recorded response. Couldn’t agree more. In fact, I believe that timing is the hidden axis of personalization. I would actually alter Nussey’s belief just slightly. And that’s simply to say that I would average their responses - and give the most recent responses just a bit more weight - to triangulate on the time your shorts buyers are most likely to open your email. For ease of use, you can bucket this into days or half-days so you don’t have to schedule an email every hour. If you record response data to your email blasts (opens), then this really shouldn’t be a problem.

So what do you ultimately have? You have customers that are most likely to want discount shorts and you have the best time to contact each of them. Now that’s personalized marketing.

A few days ago in my post Restaurant Loyalty Programs: 10 steps to understanding your most valuable customers I talked about how to do a rudimentary best customer analysis using census demographic data. One thing I did not mention, however, is the hidden danger in using census data: the temptation to overdo it.

If you followed the steps I outlined, you may have seen that the American Factfinder site offers mountains of demographic data for every zip code in the country. You might be tempted to try to use all of it in your analysis. Resist the temptation and strive to perform your demographic segmentation using as few pieces data as possible. Why?
1) Smaller data sets are easier to manage

Data integration is 90% of modeling. The more you can cut down integration by ignoring inconsequential data the less time it will take to get actionable results.
2) A kitchen sink approach can lead to overfitting

Overfitting simply means that the data is being sliced so thinly that randomness is causing you to see things that aren’t really there. While this isn’t quite as important if you’re doing the quick and dirty analysis I outlined, it becomes quite important when using more quantitative techniques.

How do you know if you’re seeing overfitting? If weird “pockets” of data exist, you may be a victim. For example, if the data show that customers who live in zip codes where males comprise between 50.31 and 50.35 of the population are twice as valuable as zip codes where they are outside that range, you have probably overfit the data.
3) Smaller models are easier to act upon

When deciding whether to keep a variable, consider what action you would propose a company to take if the variable turned out to be meaningful. For example, if you have a data set that can tell you that your best customers live in areas where public transport usage is high, you could consider advertising on subways and busses to attract more high value customers. On the other hand, if your data set tells you that your best customers have commutes between 15-30 minutes, could you act on that? If not, drop that piece of data.
Simple is beautiful when it comes to demographic modeling. Try to resist the temptation to throw all the data available at a segmentation model. You, your database guy, and your marketing staff will be glad you did.

At Istobe, we’ve clearly been preaching the virtues of customer analytics for some time now, so it makes us happy when we read new research that shows an increased interest in customer centric software solutions to make better use of existing customer data. Specifically, a new report by AMR (The Customer Management Market Sizing Report 2007-2012) shows that spending on customer management applications and software has increased by double-digits for the first time since the 1990’s.

The report talks about how “an almost universal trend around customer-centricity and customer experience management will fuel continued customer management investments regardless of economic conditions” and how in the next five years “the customer experience will be the most important potential competitive weapon.”

Additionally, the author specifically singles out marketing analytics as being “even more critical in a more challenging economy as insights into customer behavior provide better visibility into future demand and the effect of promotions and campaigns” and also notes that marketing organizations “are feeling more pressure from executives to demonstrate better insight into return on marketing investment. Today’s marketer needs to rely more on technology to aggregate demand data, analyze customer behavior and close the loop on campaigns.”

The report goes on to talk about the growth in SaaS and other industry trends, but the biggest takeaway I found was that the current economic recession is a great opportunity for companies to gain ground on competitors by implementing customer analytics solutions that better focus on increasing customer retention (see our other post about that here) and improving marketing response rates. While many organizations have already planned for or implemented cost cutting measures in light of concerns over the future, I definitely believe the measure of success between firms in a given industry over the course of the next few years will largely be determined by how effectively they use their existing data to manage their business.

Restaurant loyalty programs are a great way to get customers coming back for more. But what about the vast amount of data that you are generating from your loyalty program? Are you making the most of it?

In most cases, restaurants are sitting on mountains of data that they know is valuable, but they are unsure how to use it. This should not be too surprising since crunching data may very well be the least useful skill in the restaurant business.

It’s too bad, because there are a hundred ways to turn restaurant loyalty data into money. One of the most useful is to identify the demographic attributes that correspond to loyal, high spending customers. When it comes time to open the next restaurant you can focus on areas which have profiles similar to your ideal customers.

Many data mining and geographic information system consultants will be happy to do this analysis for a healthy fee. But even if you can’t afford consulting you can do a rough analysis yourself by following these 10 steps:

1) From your loyalty program application, export the following data to an Excel spreadsheet:

Customer zip code
First visit date
Last visit date
Total dollars spent over lifetime

2) Create a column called “Dollars Spent Per Day” in the spreadsheet and calculate it by dividing total dollars spent over lifetime by last visit data minus first visit date.

3) Create a pivot table that uses zip code as the row label and the average of Dollars Spent Per Day as the value. You now see which zip codes produce the highest spending customers on average.

4) Ignoring zip codes with only a few data points, find the two or three zip codes with the highest average spends.

5) Go to the American Factfinder site at census.gov and select Data Sets->Decennial Census from the left hand navigation bar.

6) Select “Census 2000 Summary File 1 (SF 1) 100-Percent Data” and then click “Quick Tables”

7) In the “Select a geographic type” drop down, choose “5-Digit Zip Code Tabulation Areas” and select the two or three zip codes you chose above. Click the “Add” button to add those zip codes to the result set and press “Next”.

Select the 5-10 most relevant tables and press “Show Result”. You’ll see a breakdown of demographics for this zip code. You may want export this to Excel.

9) Repeat step 8 for the other one or two zip codes you chose. Also, try going back to the American Factfinder main page and repeat the process using “Census 2000 Summary File 3 (SF 3) - Sample Data” information.

10) You now have detailed demographic breakdowns for your best performing zip codes. The next time you’re looking to open a restaurant, compare the demographics of your prospective locations to this gold standard. If it looks similar, you may have a winner. If it looks very different, however you may want to think twice about expanding into that area.
It is important to note that this technique is far from perfect but it’s not bad to use as a rough cut for which areas look good and which not so good for expansion. If you need more precise (and statistically sound) analysis, it may be worth the money to hire a consultant who can develop more formal estimation models using additional data sources.

Cross-sell should work particularly well for a company that has one best-of-breed product and a bevy of products on the come-up with which it really wants to gain traction. The reason for this: more accurate prediction of which groups of customers will buy a second product and when. You may even be able to determine which product is the likely second choice of those customers, meaning that you know which of your immature products to pitch them. By leveraging data mining, your best-of-breed products can really slingshot your other products to stardom.

If you have a best-of-breed product, then it’s likely that the majority of your customers were introduced to your company via this product. They became your customers because of this product, are likely customers who seek best-of-breed products, and are unlikely to buy one of your offerings which is not best-of-breed. In fact, they might not even think of you when they think of a product category in which you have a lesser offering. When these best-of-breed customers go looking for a new product category (in which you do have an offering, albeit an inferior one), you won’t be top-of-list. This is where timing is key to you. Knowing when your customers will buy a second product is a proxy for when they have money to buy and a proxy for when they’ve turned their attention to another product category. Thanks to data mining techniques such as neural networks and decision trees, you know when they have money before your best-of-breed competitors, which means you have a headstart - an early opportunity - to suggest a product category that may just be entering your customers’consideration set.

Once you get this timing down, you can figure out the best way to tout the integration or synergies that make your product better. Heck, you might even offer a discount on your non-best-of-breed product. But it’s the timing that gets you there. And never forget that these are your customers and you understand them - their behavior - better than your best-of-breed competitors. Use it to your advantage by investing in some state-of-the art database analytics techniques.

As I discussed last week in Why Matchback Analysis Overstates the Importance of Catalogs, one of the most effective ways of figuring out how our direct marketing efforts drive online sales is to do holdout testing. Holdout testing is nothing more than a controlled experiment and, done correctly, is a low-risk way of producing the accurate results that matchback analysis can’t.

Let’s say we’re a cataloger and we want to know which of our online-only shoppers we can stop sending catalogs to. The simplest way to find out is to test it:

1) Separate the online-only customers into behavioral and demographic segments

If you already have a customer segmentation schema in place you can skip this step and use your existing segmentation instead. If you don’t have a schema, you have a couple of options.

You can do a manual segmentation by thinking about who your main customer groups are and what attributes they have. You can then developing rules based on those attributes to do segmentation (i.e., Age > 55, suburban address, often buy children’s items is classified as a grandparent).

If you want a more quantitative based approach and have a statistician or data miner on staff, consider using a clustering technique such as k-means or two-step. These will produce statistically sound groupings which are perfect for holdout testing. Sometimes, however, it’s no so clear what to call each group or what they look like.

2) Randomly choose a set of customers in each segment who will serve as the experimental group

One of the more common mistakes is selecting an experimental group that is needlessly large. We want to ensure the test doesn’t impact the business too much so it’s important to try to keep these groups small. This table to give you a rough idea of how big your sample should be per segment:

If you typically have a higher response rate you can afford a bigger margin of error in your testing. The reverse is also true. If your response rates are smaller, you’ll need a tighter margin of error in your testing to ferret out valid results.

3) Stop sending catalogs to the randomly chosen customers in each segment and track the results

For best results, run this test over a few months and see how the response rate of the control group who still receiving catalogs differs from the experimental group in each segment. If the experimental group’s response rate is only slightly smaller than the control group’s, the loss in revenue may be small enough that you can save money by not sending catalogs to that segment.

This experimental technique succeeds where matchback fails and helps you identify segments that no longer need your marketing dollars to spur spending. Finally you’ll know whether the catalog does indeed drive online sales.

Tim Ferriss makes some excellent points in his post The Margin Manifesto: 11 Tenets for Reaching (or Doubling) Profitability in 3 Months which it got me thinking about how margins are changing in the software business and why enterprise software companies must start “firing” their high maintenance customers.

The software industry for some time has been forgiving of poor fiscal discipline. With 90% margins, it is possible to blow lots of cash on unprofitable sales and marketing campaigns and still make a mint. Furthermore, Wall Street has always rewarded new license revenue growth over cost control. In this kind of environment any new revenue is good revenue, regardless of its ultimate price.

Sadly, the days of inflated margins are nearing an end. The price of software is crashing, and SaaS along and the consumerization of IT is turning software into a commodity. Enterprise software companies doing $500k deals on six month sales cycles will have to reduce their cost structures quickly to survive this disruption to their model.

With these changes afoot, plenty of blog space has been devoted to exploring how software companies can cut sales and marketing costs through search engine optimization, pay-per-click advertising, and viral marketing. Comparatively little has been written about Tim’s #10 point, however: firing high maintenance customers.

Despite the huge improvement in margins that result from firing poor customers, there are three reasons that the idea rarely gains traction in an organization:
1) Cultural resistance due to short-sighted metrics

Most companies are reluctant to fire customers. After all, no sales or marketing organization can be convinced that it’s a good idea to forgo revenue in pursuit of improved profitability down the line. This is especially true when they are measured on how much they drive top line growth, as they almost always are.
2) Data integration challenges

Even if sales and marketing can be convinced of the value in firing poor customers, there are still huge technical barriers to integrating the data required to for analysis. Challenges abound in getting CRM data to merge neatly with support and billing databases. Unless IT has a lot of spare capacity (an occurence as common as a Bigfoot sighting), significant budget will have to be allocated for data integration.
3) Inability to make sense of the results

Finally, once all the data is integrated, a healthy dose of marketing analytics know-how is required to make sense of it all. Without highly trained business analysts on staff, it is very difficult understand which customers are profitable and which aren’t. Furthermore, unless you want to keep spending money acquiring bad customers, statisticians and data miners will need to be called in to help build attribute profiles of unprofitable segments.

While firing unprofitable customers is a powerful way to improve margins and profitability, these three barriers ensure that it rarely gets done.  Unfortunately for most enterprise software companies, it will be too late by the time they realize how criticial it is to shed themselves of poor customers.   Those with the foresight and fortitude to make it happen sooner than later, however, can expect great rewards.