April 06, 2010

Worst Practices: Catalog Matchbacks

Sometimes, we need to look again at an old concept, to see if it is still viable.

Today, the concept is the "matchback". This is the catalog process where online orders over, say, a sixty day window, are "matched back" to the customers who received a catalog. The assumption is that when a customer orders online, the customer orders because something stimulated her to order. More often than not, the cataloger believes that the catalog caused the order.

The best way to validate whether matchback results are accurate or not is to execute mail and holdout tests. Take your house list of 200,000 twelve month buyers, randomly sample 40,000 names, then split those names into two groups. One group receives your next catalog, one does not receive your next catalog.

Pay close attention to online demand in the three weeks after the catalog is mailed. Let's assume you see this outcome:
  • Mailed Group Online Demand = $4.00 per customer.
  • Holdout Group Online Demand = $0.20 per customer.
When you observe this outcome, you know that catalog matchbacks represent a "best practice" for your business --- online orders simply would not happen if the catalog is not mailed.

But if you see this outcome, you've got issues:
  • Mailed Group Online Demand = $4.00 per customer.
  • Holdout Group Online Demand = $3.00 per customer.
When you observe this outcome, it means that your matchback algorithm is taking credit for all of your online orders, when in reality, 75% ($3.00 / $4.00) of the orders would happen anyway, regardless whether the catalog was mailed.

In this situation, catalog matchbacks are a "worst practice". They are causing you to dramatically over-mail your customer file, matchbacks may be costing you 15% to 50% of company profitability.


I mean it. That is what happens! I analyze a lot of customer data across a lot of companies. I see this on a routine basis.

Long-time readers are looking at this and are saying "... Kevin, you've been harping on this topic for more than four years, when are you going to stop?" Well, I'll stop when we, as an industry, begin to avoid "worst practices".

I want your business to be highly profitable.

I don't want for you to waste money on cataloging, when you could save that money and re-invest it in customer acquisition or re-invest it testing new methodologies or new technologies.

Execute a test. Determine for yourself if you are executing "best practices" or "worst practices".


  1. But don't just do it for catalogs, do it for email too. The results are often very similar to those described above!

  2. Yes, this works for e-mail, too!

    Mind you, folks, I'm not saying you shouldn't do matchbacks, I'm saying that it is good to know the context of your matchback results, because the context makes all of the difference!


Note: Only a member of this blog may post a comment.

Good Vendor Employees Are Working All Around You

So I'm on a Zoom yesterday, and the individual representing the vendor did SUCH a good job. What does doing a good job look like? Patien...