Dear Catalog CEOs:
Be honest, you love your matchback algorithms, don't you!
And for good reason. Matchback algorithms validate the need for a catalog. We like the belief that we mail a catalog to a customer, and then the customer visits your website to place an order.
This hypothesis works as long as you mail the customer once or twice a year.
Once we mail the customer at least nine times a year, we make a critical mistake. You see, each matchback window is about six weeks long, so we assume that any online order is associated with a catalog.
At seventeen mailings a year, you have a whole bunch of three week matchback windows, so every single online order is credited to a catalog.
The reality is that once you mail a customer nine or more catalogs a year, you make many mistakes, mistakes that cost you a half million, a million, or two million dollars of profit per year. The vendor community doesn't mind this, because the more mistakes you make, the more you use their services ... you buy more paper, you print more catalogs, you rent more names, you pay for more postage. The entire vendor ecosystem benefits every time you over-credit your catalogs with online demand.
And because you don't want to lose demand, you and your marketing team won't execute holdout tests. As a result, you never get to see just how much demand still happens when you stop mailing catalogs to housefile customers.
My clients and I, however, do get to see how much demand still happens when you stop mailing catalogs. If you're an old-school cataloger with a 60+ year old customer base, you'll still get 30% of your housefile demand (your mileage will vary ... knowing the exact percentage is really important). If you're an average cataloger, you'll still get 45% of your housefile demand. And if you have a younger customer base, you'll still get 65% or more of your housefile demand.
This means that you're wasting a ton of money ... maybe 20% to 30% of your housefile ad costs don't need to be spent. That's a lot of bacon!
Do me a favor. Set up a three month holdout test. Randomly sample 10% of your 12-month buyer file, and split it into two groups. The first group receives catalogs normally. The second group is not allowed to receive one single catalog for the next three months.
At the end of three months, measure the difference between the mailed group, and the holdout group. Then come talk to me about how we determine the optimal number of catalogs to send to a customer.
Tomorrow, we'll take a look at what the optimal contact strategy might look like.
Thank you for the post. I personally have read something similar a couple of months ago written by Yury Mintskovsky but did not quite understand exactly what he meant. But after reading your work everything is clear to me now.ReplyDelete