Triggers And Grids

Our future includes a shift from pure campaign-centric work to grid-centric triggers. Oh, I know, countless pundits have been telling us this for decades ... they tell you how their cloud-based machine learning application will uncover patterns you'd never anticipate. 

But the reality is that we must combine business experience with math ... unfortunately, we tend to omit the business experience portion of the equation.

Take a look at this grid ... the grid illustrates how much a customer with various recency-oriented direct-channel and retail-channel attributes will spend online in the next year:

Grids help us understand customer value. Grids help us understand whether we should change our behaviors or not.

Look down the "Direct Recency = 25 to 36 Months" column. This column shows us how much a 25 to 36 month direct channel buyer will spend online, in the next year, by in-store retail recency segments.

Let's pretend that we have an online buyer with no retail history ... that customer is expected to spend $14.51 online in the next twelve months. Now, look at what happens when the customer buys from a retail store ... the customer moves up the column, and moves into the "0 to 1 Month" recency segment. Online value increases ... from $14.51 to $37.58.

Conversely, read across the bottom row ... let's pretend that the customer buys online instead of buying in-store. Now, the customer moves from $14.51 of online value to $141.77 of online value.

The retail purchase has very little influence on online value. The retail purchase would likely trigger a different email cadence - one that encourages the customer to learn more about the store.

The online purchase has significant influence on online value. The online purchase would trigger a different email cadence - one that encourages the customer to continue to shop online.

Grids quickly become complicated ... it's hard to move beyond two dimensions. A good idea, then, is to create models that combine numerous attributes into one dimension. You predict, on an A/B/C/D/F basis, the value of the customer online, and you predict, on an A/B/C/D/F basis, the value of the customer to a retail store. Then you have a 5x5 grid ... and better yet, you have multiple 5x5 grids that show retail value, online value, and company value. Based on customer value, you trigger the customer into different marketing programs.

In the coming months, we're going to talk more about grids, about models that combine attributes, and the triggers we use to capitalize on customer behavior. The goal is to expand business knowledge ... in terms of your career, business knowledge > machine learning applications.

Sports Commerce Series: Customer Metrics

Many of our readers - especially those who follow along on Twitter, are actively involved in Sports Commerce. They perform analytics for sports teams, or they are part of sports organizations.

So I am going to do something for those folks ... each Thursday for awhile, I am going to run a series of sessions on Sports Commerce. In today's session, I will draw comparisons between Sports Analytics and Sports Commerce. Subsequent sessions will build upon prior content, as I explore how Sports Analytics Experts can generate significant profit for their teams/organizations by applying analytics techniques to their commerce divisions.

If you can see the embedded video below, please click it to view the session.

If you cannot see the embedded video, then please click here to visit YouTube to watch the session. 

If you work for a Professional Sports Franchise or a Collegiate Organization and don't have the resources to analyze your own Team Shop or E-Commerce Store, send me an email message (kevinh@minethatdata.com) and we'll talk about how I can help.

Williams Sonoma

Have you had a chance to read through the Williams Sonoma Annual Report (click here and download the 10-K statement).

Look at page 25, where you will observe selected financial data. Yesterday, a trade brand pointed out that their omnichannel strategy is clearly working. Let's evaluate their hypothesis. Look at the metrics for 2010 (the year before a research brand invented the term "omnichannel"), and compare the metrics to 2014. What do you see?

• Slower growth.
• Lowest gross margin percentage in the past five years.
• Slowest e-commerce growth of the past five years.
• Second slowest retail growth over the past five years.
• If selling square footage have increased (+2.6%), then the retail growth metric (+2.4%) implies a necessary drop in comp store sales.

Clearly, expenses are being managed well, because profit as a percentage of net sales is good. And the metrics we see in 2014 are outstanding ... most retailers would beg to have metrics that look like this.

But my goodness. This isn't the outcome of an omnichannel strategy that is clearly working. This is the outcome of a company that has done things well for three decades.

Williams Sonoma is worth emulating, no doubt about it. But not because of omnichannel excellence ... for if they were excellent at performing omnichannel strategies, their metrics suggest that omnichannel excellence is resulting in a slow degradation of business performance.

Lifetime Value (LTV)

When I started working at Lands' End, back in the early 1990s, I was surprised how few companies calculated lifetime value.

When I work with companies in 2015, I am surprised how few companies calculate lifetime value.

#Sigh.

I know, I know, you don't have all the proper cost metrics and you use web analytics software that only allow you to evaluate campaigns and as a result you perceive that it is impossible to perform lifetime value analytics and you don't care anyway as long as your campaigns deliver an acceptable ad-to-sales ratio so so what?

All of the magic in your business comes from understanding how customers behave. No, not how they behave in campaigns ... but how they behave.

Do you have the table illustrated above? Do you review the table, at least quarterly? No? Why not?

The query is terribly simple.

• Identify all customers who purchased as of April 6, 2014.
• Segment those customers based on recency (three months increments) and life-to-date orders (1x, 2x, 3x, 4x-6x, 7x+).
• Then, for each segment, calculate the average amount of demand spent by the customer in the next twelve months.
• Produce the table outlined above.

How hard can that be?

And yet, there's a lot of value to be had from reviewing the table.

Notice how few high-value customers there are?

• 0-12 Month 7x+ Buyers.
• 0-6 Month 4x-6x Buyers.
• 0-3 Month 3x Buyers.

That's it.

Now look at all the low-value buyers (green and blue). Notice that a first-time buyer quickly moves into low-value status if the customer does not repurchase within six months. Don't you think that's something your whole company should know? Shouldn't every employee know that you have six months to convince a first-time buyer to purchase again, or the customer will descend into low-value status, requiring you to acquire another customer? Wouldn't your marketing team want to craft programs to convert the customer to a second purchase within that vital six-month period of time?

You don't have to perform a full lifetime value analysis to do the right thing. Just replicate the table above. It's not hard to replicate the table - the queries are terribly easy to replicate. Then identify the blue/green segments above, and do something to prevent customers from falling into the blue/green segments. By doing just this small amount of work, you replicate 80% of the value of a lifetime value program. Take the table above and convert it to profit, and you are 90% of the way there. And you've done virtually no work!

If you cannot produce the table, contact me (kevinh@minethatdata.com) and I'll do it for you.

Influential Books

When you begin your career, you gobble up information ... everything is new and interesting and you are a sponge, soaking up knowledge.

If your career is measured in fifteen year thirds, then there is a point somewhere in the latter half of the first third of your career where you realize you have to find your own identity. You stop doing things the way everybody else does them, and you chart your own course (or not).

There are six books that influenced my path.

1. Statistical Methods / Snedecor and Cochrane (click here). I have a version from 1938. The book talks about the kind of tests that modern online marketers claim to have invented. Without Snedecor and Cochrane, Google wouldn't be what Google is. Did I mention that these folks wrote about stuff back in 1938 ... stuff that digital gurus love talking about today? Maybe everything new is old.
2. Correspondence Analysis in the Social Sciences / Greenacre and Blasius (click here). The book illustrates ways to take a simple cross-tab table in Excel and map the relationships between the rows and columns. There is no better lesson than one where you learn that complex information can be reduced to simplified information.
3. Modelling Biological Populations in Space and Time / Renshaw (click here). The most influential book of my career. The book explains how species interact with each other ... how wolves need/eat rabbits until there are no rabbits and the wolves starve, for instance. Everything in this book relates to modern channel interactions. You learn that e-commerce needs stores, but stores don't need e-commerce ... you learn that e-commerce preys on stores, driving store performance down, causing stores to close ... stores are rabbits and e-commerce is a wolf (and by the way ... e-commerce will be a rabbit to the mobile wolf, it's coming, and e-commerce folks choose to not recognize this, to their detriment). Once you draw the comparison, your view of the world changes - you are able to see the future clearly.
4. Permission Marketing / Seth Godin (click here). You may have noticed that everything about my business model has roots in this book. The book offered a blue print for building a small business. Still does. Mr. Godin was a beloved guru - then he refused to play the social media game of engagement (like he could engage with 400,000 readers on a one-to-one basis) and so many marketers abandoned him ... but that's their loss.
5. The Demography of Corporations and Industries / Carroll and Hannan (click here). This book parallels "Modelling Biological Populations in Space and Time", but looks instead at how industries evolve. For me, the "ah ha" moment was how you start with a ton of small businesses ... the small businesses eventually coalesce around a handful of winners who become huge ... and those winners are then defeated by a ton of small businesses who capitalize on a new technology or change in the system ... and then those small businesses coalesce around a handful of winners who become huge. This pattern repeats ... and is in the process of happening once again (mobile).
6. Moneyball / Michael Lewis (click here). I don't like this book because of the metrics. This book is important because of how it discusses culture. You can have all the metrics in the world, but if the culture isn't aligned, well, tough, you're not going to make progress. And here's a hint for the analytics community - you need an evangelist at a high level - people don't just accept your arguments and do what you say - the world doesn't work that way. Best of all - the book outlines how you can do everything right, and in the end, you don't succeed - and somebody with more resources steals your ideas and gets all the glory.

I think it is really important to have your own "system", your own set of beliefs that you implement where you work, and then you experience success, success that you use to further build upon your own, unique system. You need to have your own way of doing things. Of what possible good is it to be a "Google Analytics" expert or an "omnichannel expert" when you are doing things that ten thousand people can do equally well? Craft your own way of doing things, provide value, and you'll go places.

Gliebers Dresses: Good Friday

Yes, this is a fictional business story, where I try to explain topics that folks don't necessarily want to have real discussions about. If this isn't your cup of tea, then why not read this article (click here) from Google about how much they love Macy's omnichannel approach?


Without further delay, I take you to the Gliebers Dresses Executive Conference Room


Glenn Glieber (CEO, Owner): Why don't we get started, folks? I'd like to make this quick, so I can attend the Good Friday service at church.

Lois Gladstone (Chief Financial Officer): Where is Pepper?

Glenn Glieber: No need to be smug, Lois. You already know that I suspended her with pay for one month.

Roger Morgan (Chief Operations Officer): She's at church right now with the girls.

Meredith Thompson (Chief Merchandising Officer):  But why suspend her?

Glenn Glieber: I held her accountable for working with a vendor that is costing us money. You and Lois wanted somebody held accountable. I held Pepper accountable.

Meredith Thompson: I wanted the vendor held accountable.

Lois Gladstone: She should have been fired.

Roger Morgan: I think even Brittney Spears would agree that you've become toxic, Lois.

Lois Gladstone: Omicron is not performing well, and we can't get away from them because they are too important to our business. Somebody has to be held accountable.

Glenn Glieber: Can we focus on business? How were sales yesterday?

(Roger, Meredith, and Lois all sit quietly, unable to answer)

Glenn Glieber: Nobody knows the answer?

Meredith Thompson: Pepper isn't here to answer the question.

Roger Morgan: Lois is responsible for marketing in Pepper's absence. Lois, answer the question.

Lois Gladstone: I'm getting a lot of questions in Pepper's absence. Yesterday, our merge/purge vendor wanted to know how many names should be in our June holdout test.

Roger Morgan: What did your answer?

Lois Gladstone: Well, the vendor told me to not put sales at risk.

Meredith Thompson: See, that's a vendor who is trying to protect us.

Lois Gladstone: So I said we should only have 200 customers in the holdout test.

Roger Morgan: What?

Lois Gladstone: That's what the vendor said! Then the vendor asked if I was sure I wanted 200 customers in the holdout test. I thought about what the vendor said. So I changed my mind.

Roger Morgan: Thank God.

Lois Gladstone: I told the vendor to only include 100 customers in the holdout test.

Roger Morgan: You need at least 10,000 customers to achieve statistical significance. You just ruined a test.

Meredith Thompson: Hold her accountable, Roger!

Lois Gladstone: I didn't ruin anything.

Roger Morgan: Look, in my left hand, I am holding a report from Woodside Research. This $495 gem talks all about how Google invented A/B testing and it outlines how important it is to have enough customers in a holdout test to be able to properly measure results.

Lois Gladstone: But the vendor told me to not put sales at risk.

Meredith Thompson: This is your fault, Lois.

Lois Gladstone: I didn't mean to do something wrong. It's hardly my fault.

Meredith Thompson: In Pepper's absence, you are accountable for marketing. You should be fired. Honestly.

Lois Gladstone: I didn't even know what I was doing. You should only fire folks when they know what they are doing.

Glenn Glieber: Why would we ever not mail customers catalogs?

Roger Morgan: Woodside Research says there is a sweet spot in testing. For a company our size, it is around 10,000 customers.

Lois Gladstone: I also got a call from our Email Service Provider. They told me that we should back up from five email campaigns a week to three, so that we could eliminate some of the opt-out challenges we have.

Meredith Thompson: Did you listen to the vendor?

Lois Gladstone: Their argument made sense. They said that by not sending out email campaigns, we would reduce opt-outs, and by reducing opt-outs, we'd create customer advocates who would evangelize our brand for us. They told me that their strategy would act a lot like free marketing.

Glenn Glieber: I love free marketing!

Roger Morgan: So you cancelled two email campaigns for next week?

Lois Gladstone: Yes.

Meredith Thompson: Which two?

Lois Gladstone: Does it matter?

Meredith Thompson: Does it matter? Does it matter!! We buy inventory to meet email marketing demand. When you kill two campaigns, you kill demand for the two featured items. Now I have to liquidate those items. That costs us gross margin dollars.

Roger Morgan: Lois, aren't you held accountable for generating gross margin dollars?

Lois Gladstone: The vendor said we'd reduce opt-outs.

Meredith Thompson: Every time Roger brings up some stupid vendor idea endorsed by Woodside Research, we yell at him. You've been accountable for marketing for just one week and you're already falling under their spell.

Lois Gladstone: I'm not under a spell.

Meredith Thompson: Either we fire you, or we fire the vendors, Lois. Somebody has to be held accountable.

Lois Gladstone: I guess I didn't see how all of this stuff was connected.

Roger Morgan: Interesting.

Lois Gladstone: At least our social media agency isn't costing us sales.

Meredith Thompson: That's because social media doesn't work when your core customer is older than fifty-five years old.

Lois Gladstone: They wanted us to stand up for equal treatment of all individuals by tweeting stuff using the hashtag #wecareabouteverybody.

Meredith Thompson: Sounds risky to me.

Lois Gladstone: You should have seen all the nasty replies we got.

Meredith Thompson: Why would you listen to a social media vendor?

Lois Gladstone: It sounded like a good idea. What is so bad about caring about everybody? You know I learned yesterday? There's a lot of people out there who don't care about anybody!

Meredith Thompson: Those aren't our customers. Trolls don't buy dresses, they hide under bridges.

Roger Morgan: You were crucified for taking an unpopular stance.

Lois Gladstone: I had to temporarily shut down our Twitter account.

Meredith Thompson: You should hold the vendor accountable.

Lois Gladstone: All I did was tweet that we care about everybody. Why should anybody be persecuted for that?

Glenn Glieber: Are you telling me that you ran my good name through the mud?

Meredith Thompson: Never mix politics, religion, and business. Pepper knew where the line was, and she knew not to cross it.

Lois Gladstone: Well, I learned a valuable lesson this week.

Roger Morgan: What's that?

Lois Gladstone: Marketing is a lot harder than I thought it was.

Roger Morgan: Maybe next time you'll be nicer to Ashley Cole and the good folks at Omicron.

Lois Gladstone: Oh heck no, they need to be held accountable.

Catalogers and Co-Ops

I found this article (click here) on the home page of the NEMOA website two days ago.

I will say this - I would be very frustrated to be a non-Executive working at a co-op. You have too many strikes against you, limiting you from doing a good job.

1. Few individuals understand statistical models. You have to defend models that you are not trained to understand. That's hard work. The modeler should have to sit in front of the client and address the questions you get.
2. Catalogers do not give you good information. Catalogers should be fully transparent with you, demonstrating five year or ten year trends across comparable segments, showing you, to the penny, the decline in profit they are experiencing. How can you help them when they will not give you detailed profitability information by segment?

That's why I wrote the Ashley Cole character in last week's Gliebers Dresses (click here). She was being beaten silly by folks who don't know anything about marketing (the CFO), and is held accountable for mistakes that Gliebers Dresses made over a twenty year period of time.

I also wrote the Gliebers Dresses story (more coming, folks, more coming) to defend catalogers. The speaker at NEMOA wouldn't have offered a passionate ten minute argument (from what many of you have told me) unless his client base (like mine) experience frustrations when working with the co-ops. The late Don Libey went through this in 2006 - he addressed client concerns in his newsletter and at a conference - and was promptly demonized by some in the vendor community ... the predictable response of an article in a trade journal that sponsors conferences and gets ad revenue from the very co-ops in question is a time honored tradition. Notice that trade journals do not ferret out cataloger stories to validate if there are challenges or not ... isn't that what a journalist would do? 

By the way, Mr. Libey's concerns were not fundamentally different than the concerns outlined by catalogers in 2015 - nearly a decade later.

In the article above, the author suggested that no solutions were offered at NEMOA. I'd like to offer a few solutions. Since employees from all co-ops read this blog on a frequent basis, let's see if we can't get a few solutions generated.

Solution: Give every catalog client the specific equations used in their model. Housefile modeling organizations like Clario happily do this for clients. Please do this, it's not hard, right? And have the modeler Skype with the client, explaining why the modeler is using the methodology the modeler is using. This is a great first step in being open and transparent.

Solution:  For each select, list the percentage of the select that have purchased from the client's product category in the past ninety days. When the percentage is low (i.e. fewer than 50% of the select buys widgets), explain why. Use prior results to clearly demonstrate that non-widget buyers are worth more than prior widget buyers. Or if you cannot do that, make sure that widget buyers are in each select!

Solution: We all understand that all households are identified by all co-ops. Let's try something different. On a monthly basis, could the co-ops please publish a count of what I'd call "Outstanding Catalog Buyers" on a public blog? Nobody cares that 100,000,000 households are being captured and that a low minority of e-commerce transactions are captured. Everybody cares that there are 8.1 million "Outstanding Catalog Buyers", and that the total has dropped from 8.7 million last year and from 9.1 million two years ago. This helps catalogers an awful lot. The co-ops then raise the issue that the "Outstanding Catalog Buyer" audience is shrinking, and therefore, it isn't the co-ops fault that the audience is shrinking, but instead, it is a reflection of the catalog industry. Similarly, comparable metrics for the e-commerce ecosystem, social ecosystem, and mobile ecosystem should be published monthly, so that catalogers can see that the world is changing.

Solution:  Every month, each co-op publishes a private report for each cataloger, telling how many times cataloger contributed names were mailed by co-op members. Also, for all the names taken by a cataloger, a report outlines how many times those names were mailed by the cataloger in the past year, and how many times those names were mailed by other co-op members in the past month. All reporting is outlined in a this year vs. last year vs. two years ago format, for comparison purposes.

Solution:  Create an online selection system that allows catalogers to pull their own names. If catalogers are not obtaining acceptable performance, then let catalogers choose the names themselves, using the attributes that co-ops typically model against. This serves two purposes - first, you cannot be accused of not being transparent - second, catalogers will quickly see how hard your job is after their selects perform at 80% of your modeled results.

Solution:  Demand that all catalogers share a common set of metrics with you. Response Rate, Average Order Value, Dollar per Book, and Profit per Order. Profit per Order is most important. Demand that every cataloger share profit per order with you, on every select, every segment, every decile, for the past five years. This way, you can compare performance equally across time. If catalogers are being honest with you, and most are, then you will have clear proof that performance is suffering. But demand that profit per order is evaluated, across all marketing activities. Also demand that catalogers share Google performance, retargeting performance, you name it. If you are generating more new names at a reasonable cost, make sure folks know.

Solution:  Given catalogers the opportunity to not share their names with e-commerce businesses. Co-ops were created for catalogers.

Solution:  If you are not requiring e-commerce businesses to contribute names, and you are offering cataloger names to e-commerce businesses, give the cataloger the opportunity to split list income 50/50 with you - since the cataloger was kind enough to freely contribute the names to your database.

Solution:  Open your books, and if your performance is struggling as cataloger performance struggles, share that fact with catalogers. Demonstrate that you are in this challenge with catalogers. If your performance is improving and the cataloger you work with is struggling and that cataloger is having positive housefile performance, offer to do something to help the cataloger out.

Solution:  Invite each client to your headquarters, so that your clients can meet the modelers and Executives and back-of-the-office staff who make new customer acquisition possible. Hold a day-long conference, and share your vision as to how you plan to use data from all sources to make catalogers profitable for the next ten years or more.

Solution:  Report on the percentage of each select that are names that were contributed by pure e-commerce businesses.

Solution:  For each select, report on the average age of the name offered. Also list 25th, 50th, and 75th percentiles for age. If ages skew old, prove to the cataloger that the older names outperform younger names, via reporting of prior results.


Use the comments section to offer more solutions. And since so many co-op employees read this blog, why not jump on here and leave a comment with your thoughts about the solutions?