Matchback Bias

You're probably partnering with your list organization, data warehouse vendor, or co-op on the never-ending scope of matchback analytics.

The goal, of course, is to prove that catalog marketing is a vital piece of the modern marketing puzzle. You're trying to truly understand the ROI of this activity. That's why you try so hard to attribute every online order back to one of the dozens of catalogs you mailed in the past year.

Now let me ask you this.

Do you go through the same effort to attribute every phone order back to the original online source?

You don't?

I met with a business that is doing just that. They combine their web analytics tool and their matchback analytics platform to attribute phone orders back to the online marketing activity (which is usually organic/natural search) responsible for driving the phone order.

Why is it that our industry is so bent on proving that catalog marketing drives online orders, but doesn't invest the energy to prove that online marketing drives phone (and store) orders?

Our view of the world is biased, folks. And that bias favors co-ops, printers, the USPS, the paper industry, and the list rental/exchange industry.

Your thoughts?

Geek Alert!! Channel Preference And The Hyperbolic Tangent Function

The final step of most of my Multichannel Forensics projects involves predicting channel preference.

This is an important step, because customers that are likely to purchase from self-serve channels in the future (online, stores) may require less advertising than customers who purchase from full-service channels (catalog ordering over the telephone).

A handy mathematical transformation for estimating channel preference in a two-channel situation is the "Hyperbolic Tangent Function" (this was used extensively at Lands' End in the early 1990s to isolate customers likely to return most of their merchandise, allowing us to suppress mailings from these folks).

In the modeling process, you assign your dependent variable a value of -0.999 (telephone), and the other channel a value of +0.999 (online). If a customer splits dollars across both channels, the value is 0. If the customer splits dollars 2/3 phone, 1/3 online, you do a weighted average, yielding -0.333.

Then you transform the dependent variable ... (0.5)*LN((1+x)/(1-x)), where x is the value listed above.

Now you run your ordinary least squares regression against the transformed dependent variable, predicting the channel customers will purchase from in the future.

The current customer file is scored using this model. Once each customer has a score, you transform the score back to a numerical value ... (EXP(2*s)-1) / (EXP(2*s)+1), where "s" equals your score.

Customers with a highly positive prediction are likely to buy online (in this example), and therefore, may not need catalog advertising.

This works for e-mail marketing as well. If you are an online pureplay, -0.999 represents customers who do not ever buy from e-mail marketing, +0.999 represents customers who always buy because of e-mail marketing. Score the file, identify those likely to require e-mail marketing to purchase, and market accordingly.

The typical process employed by many online and catalog marketers these days involves the following steps.

1. Run a Multichannel Forensics analysis on the customer file to determine channel migration patterns.
   
2. Predict the probability of purchasing in the future using Logistic Regression.
3. If the customer purchases, predict future spend per purchaser using OLS Regression.
4. Multiply Step 2 by Step 3, yielding future spend.
5. Calculate future ad spend per customer, or model the relationship ... the relationship is built on the incremental value generated by advertising, not all demand spent by the customer.
   
6. Calculate future profitability by individual customer.
7. Use OLS Regression and the Hyperbolic Tangent Function to calculate channel preference.
8. Given profitability and channel preference, create a contact strategy for each customer.

When done well, the online or catalog brand can identify ad savings that can be re-allocated to customer acquisition activities.

Audience Development: Here's A Mistake I Made

Back in November, I asked readers what they wanted to learn about. I received many responses from folks who wanted to learn more about my thoughts concerning Catalog Choice.

So I spent a lot of time writing about Catalog Choice in December and January.

Traffic, and more important, subscribers, increased by almost 20%. Almost instantly!

Good for Kevin, right?

Wrong.

Audience development is all about cultivating the right audience. In my case, I attracted an additional two hundred subscribers, folks who were actually offended by some of the topics I wrote about. I received e-mails from individuals who challenged my integrity and knowledge of my industry. All of a sudden, a vocal minority didn't like me!

I developed the wrong audience.

I began to receive unsolicited e-mail from organizations friendly to Catalog Choice, asking me to help spread the word about various ecological issues (ironic, given that stopping unsolicited mail is the objective of the folks marketing to me --- but unsolicited digital mail was ok).

When I stopped covering Catalog Choice, subscriber counts plummeted. The unsolicited e-mail campaigns slowed, but to date, have not stopped. It takes a lot longer to correct an audience development mistake than it takes to build a non-congruent audience.

For direct merchants, building a productive customer file is probably second to merchandising in importance. And yet, we make mistakes comparable to the mistake I made all the time.

I purchased an item from a company six months ago, at full price. Since then, nearly every e-mail campaign sent to me by this brand offers me up to sixty percent off my next purchase, if I use the code offered in the e-mail campaign. Clearly, this brand is trying to develop an audience that enjoys the thrill of "x" percent off merchandise offerings.

An executive recently told me that his e-mail marketing list of over a million individuals only responds to free shipping, buying more than four times as much merchandise if free shipping is offered than if it isn't offered. He developed an audience that only responds to free shipping. He cannot get away from free shipping unless he develops a new audience. It won't happen by wishing, only by hard work.

I've made countless mistakes developing an audience that enjoys and participates in Multichannel Forensics. Let's learn from our mistakes, let's develop audiences relevant to the niches we serve.

E-Mail Marketing And Customers Who Return A Lot Of Merchandise

Sometimes, our instant access to metrics cause us to screw up.

This happens to most of us.

1. We execute an e-mail campaign on a Tuesday morning at 9:00am.
2. By 10:27am, we have a forecast for how well the e-mail campaign will perform. We know open rates (or render rates as the experts now say), click-through rates, and conversion rates. We may even know $ per e-mail.

Three weeks later, twenty percent of the customers who purchased from the e-mail campaign returned their merchandise for a refund.

Did the e-mail marketing campaign work?

One of the things we can do is identify customers who are "high returners". I've done this analysis for many companies. Typically, a small subset of the audience (maybe 1% to 5% of your twelve month buyer file) are responsible for a disproportionate amount of returns.

An easy way to address this problem is to identify customers with a high return rate, and see if those customers will have a high return rate in the future. If so, you run a profit and loss statement on future sales. You talk to your folks in finance, folks who know the actual cost to process each item returned to a company.

At Eddie Bauer, we knew that if a customer had ordered at least three times in the past, and returned two-thirds or more of the merchandise she purchased, she would be unprofitable to market to in the future.

In e-mail marketing, this one is a slam dunk! You simply create a suppression list for this tiny subset of the customer file, and don't send e-mail marketing campaigns to this segment.

And then you bask in the glow of the increase in profit you obtain because of your strategy.

You are likely to see a drop in your metrics --- high returns customers are typically your most active customers --- they open e-mails, they click-through to the website, they buy stuff. And given your returns policy, you should let them buy stuff. However, there is no rule that says you must also market to the customer. So generate additional profit for your company. Stop e-mailing customers who return too much merchandise!

New Catalog Marketing Blog From Belardi / Ostroy

For those of you who enjoy catalog marketing, here's a new blog from the folks at Belardi / Ostroy, called "Directly Speaking". Give it a try!

Attention Catalogers: Co-Ops (Abacus) And Matchbacks

If you do customer acquisition via catalog marketing, you undoubtedly elected to drink the co-op kool-aid. And why not? Based on our reporting (sometimes provided by co-ops like Abacus), co-op lists outperform outside lists.

I've mentioned this before, and I want to mention it again, because the topic keeps coming up in various projects I work on. Co-op customers tend to be more likely to purchase over the telephone than rental/exchange customers.

And since phone orders are nearly 100% attributable to the advertising vehicle sent to the customer (whereas online orders are at best semi-attributable if matchback analytics are performed properly), co-op names may "appear" to perform better simply because of the channel preference of the customer selected by the co-op.

This has long-term implications for the brands we shepherd. If co-op names work "best", with co-op customers more likely to order over the phone, we then "have" to mail catalogs in the future to get the demand. And by having to mail catalogs, we have to keep feeding the entire catalog ecosystem --- printers, merge/purge houses, USPS, the paper industry, and the co-ops.

By feeding the catalog ecosystem, we anger some customers and prospects, which feeds the rampant growth of Catalog Choice.

We create our own problems, folks!

If you are a heavy user of co-ops, please consider extensive matchback analytics. At minimum, use the Migration Probability Table as outlined in Multichannel Forensics to understand future channel preference of co-op sourced names. You're in for a treat if you do!

Effective Use Of Your Database

At some point in the past five years, you probably invested in your customer database.

Maybe you built a series of normalized tables ... elegantly designed in a way that would make any IT professional proud. Maybe you hooked up Business Objects or MicroStrategy to the database, believing that you could answer any question you could think of.

Maybe you integrated your web analytics tool with your centralized customer data warehouse, expecting lightning bolts to appear from the sky about the casual visitor who browsed eight important landing pages before buying something in the store.

Or maybe you outsourced your database to a quality vendor who specializes in said activity.

I'm guessing that you're still dissatisfied with what you have.

You've probably learned the following equations:

• People > Database Design
• Database Design > Software
  

In other words, when crafting your database marketing platform, you focus on people first. One gifted query analyst or statistician means far more than any database you'll ever design. Far more!!

Database design means more than software. You need a series of summarized tables for campaign management. Don't ever let your software vendor or IT leader tell you not to store detail-level data (one row per item purchased, one row per page viewed). Your data expert needs the detail-level data to answer all the questions that cannot be answered by summarized fields.

Once those two aspects of the equation are solved, get good software.

Effective use of a database requires us to realize that people are more important than database design, and that database design is more important that software. This spring, many of you are communicating to me that your organizations view this the other way around ... you are outsourcing your analytical staff to India, you are outsourcing control of your databases, and you are relying on simple BI tools to query against summarized fields that don't adequately answer questions.

Let's turn this trend around!!