September 08, 2013

Dear Catalog CEOs: Really, Really Big Catalogs

Dear Catalog CEOs:

Yes, this is the kind of feedback I get:
  • "I recently received a five pound catalog from Restoration Hardware. No, really, it was five pounds. I weighed it. Restoration Hardware wouldn't send out a five pound catalog if it didn't work. Should we move into really, really big catalogs? Is that the future of cataloging?"
In calculus, there's a concept called a "local maxima". Go read this article (click here). That article is the source of this image:
A 500 page catalog is an example of a local maxima, especially in B2B. It's a peak that may appear to be the most profitable and most prudent decision. It may even be a better tactic than the catalog tactic you're currently employing.

Matchbacks almost demand that a retail brand, or a B2B brand, mail really, really big catalogs. We all know how horribly flawed Matchbacks are, yet catalogers use them religiously, simply because they overstate how effective catalogs are. Matchbacks are the magic that allows the catalog leader to continue to embrace cataloging.

In a retail environment, the majority of sales happen at retail. And at retail, a majority of sales happen if all marketing is discontinued. The Matchback simply and erroneously takes credit for sales that would happen anyway.

So, let's pretend that you have a 500 page monster catalog, and you're, say, Pottery Barn (who doesn't mail a 500 page catalog, but we'll use them for illustrative purposes). You mail it. Then you get the results. You apply your erroneously overstated Matchback to the results. Here's what you see after 90 days:
  • Total Sales, Online = $5.00.
  • Total Sales, Retail Store = $20.00.
  • Total Sales = $25.00.
  • Profit Factor = 35%.
  • Contribution Before Mailing Cost = $8.75.
  • Mailing Cost = $5.00.
  • Total Profit = $3.75.
It works! It works! A 500 page catalog is a great idea. Thank goodness we have Matchbacks!

Except for one little problem.
  • 50% of all online sales would still happen if the catalog is discontinued.
  • 90% of all retail sales would still happen if the catalog is discontinued.
You know this, of course, because you execute mail/holdout testing.

Now look at what your profit and loss statement looks like:
  • Total Sales, Online = $5.00 * (1 - 0.50) = $2.50.
  • Total Sales, Retail Store = $20.00 * (1 - 0.90) = $2.00.
  • Total Sales = $4.50.
  • Profit Factor = 35%.
  • Contribution Before Mailing Cost = $1.58.
  • Mailing Cost = $5.00.
  • Total Profit = ($3.42).
Oh oh.

In the real world, the catalog actually generated an enormous, embarrassing, humbling loss of $3.42 per customer.

This is where somebody in the creative department would say "Hold on, Kevin, hold on, you simply cannot measure the branding impact of a vehicle that sits on the coffee table for a year, I have a friend who displays it right next to a bowl of nuts, and picks it up repeatedly. That's some sweet branding, Kevin, and you and your cold numbers cannot account for the impact of branding".

There are three possible outcomes.
  1. Really, really big catalogs work.
  2. Really, really big catalogs don't work, they accomplish a local maxima that prevents a business from finding the true peak of marketing performance.
  3. Really, really big catalogs don't work, but we measure performance via Matchbacks, which make them appear to work so that we can keep executing them.
The last point is the sad reality of most catalog marketing measurement, especially in B2C marketing.

The middle point is the dangerous reality of most marketing tactics. In B2B marketing, big catalogs are common, and no amount of measurement proof will stop people from creating them. B2B marketers crave the local maxima, and will use the "I saw the big catalog sitting at the Jiffy Lube" argument as proof that visual confirmation equates to sales increases.

The data suggests, on the surface, that really, really big catalogs work.

Slightly advanced techniques (mail / holdout tests) show that really big catalogs usually don't work.

This leaves you in a very interesting situation.

You can believe in the fairy dust known as matchback, where everything works.

You can believe in the realm of testing, where almost nothing works.

Or you can invent the future.


  1. Anonymous3:27 AM

    In physics, to get an object from a state of local optimisation to global optimisations sometimes requires a large amount of energy.

    To put that into marketing terms, last year we froze over 20 like for like cells to test different contact strategies. This was a lot of work, but we are going much stronger into the new year.

    1. What did you learn when you conducted your tests?


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

MRV (Merchandise Residual Value): A Jewelry Example

Yesterday I showed you how Home merchandise negatively impacted how the customer behaved in the future. Here's Jewelry for the same bran...