March 23, 2020

Test Panels and Attribution

I recently read an article from a vendor ... the vendor described their multi-touch attribution routine for catalog brands.

If you don't execute four-panel tests between catalog marketing and email marketing, multi-touch attribution routines are hopelessly flawed.

Yes, they're hopelessly flawed.

Your four-panel test gives you all of the information you need to assign weights to transactions.

Look at the email portion of the table.
  • 9% of mail/phone/fax demand is attributed to email marketing.
  • 98% of email demand is attributed to email marketing / almost all of it is caused by email marketing.
  • 12% of search demand is attributed to email marketing.
  • 11% of online marketing demand is attributed to email marketing.
  • 5% of in-store retail sales are attributed to email marketing.
  • 6% of all other online sales are attributed to email marketing.
That's it ... you've got it ... you executed holdout groups, and you know the exact percentage of sales that should be attributed to email. No need for AI or algorithms. You have the facts.

Look at the catalog portion of the table.
  • 93% of mail/phone/fax demand is attributed to catalog marketing.
  • -23% of email demand is attributed to catalog marketing. Catalogs cannibalize email, and if you take catalogs away email does much better ... so you need to subtract email demand from the catalog total to account for the corrosive impact of catalogs.
  • 32% of search demand is attributed to catalog marketing.
  • 25% of online marketing demand is attributed to catalog marketing.
  • 8% of in-store retail sales are attributed to catalog marketing.
  • 53% of all other online sales are attributed to catalog marketing.
You've got it ... you take your matchback algorithm, you apply the percentages by channel, and you know exactly what the contribution of catalogs are to your bottom line. Run a p&l by segment and you are done.

If you execute test panels, you know the truth.

I'm been writing this blog for nearly fifteen years. I've been advocating this style of testing the entire time. Get busy learning the truth about your business.

No comments:

Post a Comment

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

Well, You Got Me Fired

I'd run what I now call a "Merchandise Dynamics" project for a brand. This brand was struggling, badly. When I looked at the d...