Kevin Hillstrom: MineThatData

There are (too) many digital marketers who, when you talk to them about profit, say that they don't measure profit. "I don't need to measure profit, I measure ROAS, ROAS is a best practice".

The fun part of the comment is that the digital marketer IS measuring profit, s/he just doesn't realize it.

Here's a table for a digital marketing initiative, broken down into deciles for the sake of illustration.

We see total results on the left - I converted the results to incremental outcomes by decile. In this example, deciles 7/8/9/10 lost money, they were unprofitable.

Now look at Incremental ROAS on the far right. An approximate Incremental ROAS of $2.25 is unprofitable. Anything below that is unprofitable.

If you have the discipline to keep incremental activities above a 2.25 ROAS, you're generating profit. Good for you!

        

Non-catalog readers, take the day off.

Once again, I share these images to show you just how wonked-up AI actually is. Spelling errors. Missing legs. If it's this wonked-up with a simple cartoon, imagine what else the gurus are trusting that in reality is completely borked?

But I digress.

Last week a member of the #printisback community on LinkedIn decided that #brands need to go back to 1996, as if J. Peterman were about to publish the Urban Sombrero on the cover.

The root of the lie being told to you is called "Matchback Reporting". Matchbacks ignore incrementality. Those who believe in Matchback Reporting ignore reality.

Time for a parable. Remember when I told you a member of the retargeting community took me to lunch, and for the cost of a Caesar Salad wanted me to convince the Management Team at a large retail brand that because 95% of ecommerce customers saw at least one of his ads he deserved credit for 95% of ecommerce orders (over $300,000,000 in annual sales) and he wanted a percentage of each transaction. If he presented those facts to you, a smart cataloger, you'd say "hey, Goober, get out you Lemonhead." Then you'd go back to your cubicle, pull out your Matchback Report, and perform the EXACT SAME ANALYSIS AS GOOBER DID and treat it as Gospel. You'd think that Goober was an idiot, you'd view you as a savvy marketer. And yet? You're the same person, doing the same thing, sans Caesar Salad.

Let's assume that your catalog is active across a four-week period. Let's assume you were going to mail your entire twelve-month customer file. You randomly select 50,000 customers to be in the mailed segment, you randomly select 50,000 customers to be in your no-mail / holdout segment.

You then sum all demand across all marketing channels for the four weeks when the catalog is active. Compute an average (i.e. divide each sum by the 50,000 customers in the segment). You'll produce a table that looks like this:

This is the way test results generally look. If you don't mail a catalog, your call center is quiet as 78 year olds cannot call you if they don't have a catalog in their hand. Website / Direct Load is where most demand happens, most of the demand will happen if you don't mail the catalog. Email Marketing is usually not impacted by catalogs, though your mileage will vary. Search is a channel that is clearly impacted by catalogs - catalogs cause customers to search for competing products (both a strength and a weakness of catalog marketing because your dumb catalog drives your smart customers to the competition). Social is almost never impacted by catalogs - completely different audiences.

The magic in the table happens in the Total Demand column.

Your matchback reporting takes full credit for the $5.65 of total demand. It ignores what would have happened if the catalog were never mailed. If the catalog were never mailed, the $5.65 total becomes $4.25 ... not $0, but $4.25.

Here's what the p&l might look like for the $5.65 figure.

Everything looks good here ... you appear to generate $1.54 profit per catalog/book ... in modern parlance you generate a ROAS of 7.53. It's the 7.53 figure that the #printisback community on LinkedIn like to refer to ... it's a much higher number than that 3.88 figure you get for paid search or 2.97 for paid social or ... wait ... they never quote email marketing because email marketing has the best ROAS, period.

Remember - your holdout group did $4.25, not zero. Take the control / holdout group average ($4.25), divide it by the mailed group average ($5.65), and you get 75%.

This means that 75% of what is outlined on the matchback report is a lie. A fabrication. It would have happened had the catalog not been mailed ... and you know this is true because in the table YOU DIDN'T MAIL THE CUSTOMERS IN THE HOLDOUT GROUP!!!!

We cannot run the p&L on the $5.65 that is likely reported in your matchback report. We have to run it on the incremental total ... $1.40 ... which is (1 - 75% = 25%) of the $5.65 total.

The p&l changes, friends.

This is where things get really dicey. The incremental outcome (a loss of $0.18 per catalog) is REALITY. The matchback-reported totals column of $1.54 profit is FANTASY.

On LinkedIn, the #printisback community communicates FANTASY results to you, misrepresenting the outcome as reality.

It's pretty obvious why they'd do this.

• $5.65 per book / $1.54 profit = Best ROI, which means you should hire them.
• $1.40 per book / ($0.18 profit) = It might be time to shut down your catalog division.

        

I gave a presentation in 2016, in front of about 1,500 people. I spent nearly an hour explaining to the audience how running an ecommerce brand would become comparable to running a sports franchise.

The audience was not impressed.

I do recall a pair of Associate Athletic Directors working for FCS Colleges reaching out to me to suggest I "had it right" - they suggested their world was heading in this direction as well. My industry didn't agree, these people who weren't in the industry believed in the thesis.

Here we are, in 2026.

As you can see, AI has a way to go. That's a catastrophic effort at creating a cartoon for me. And you need to see the cartoon to understand AI limitations. How will you know when AI completely butchers your marketing efforts? It's going to happen, and it's going to be spectacular.

Last year I stood in a luxury store with an Executive. The Executive told me to watch the customer. The customer was spending somewhere north of $5,000 ... and she had a glow that reached from Phoenix to El Paso. It was the kind of glow an Eagles fan might feel after beating the Cowboys 34-28 on a last-second touchdown pass.

The customer felt special.

The theoretical Eagles fan would feel special.

It isn't hard to see the future of ecommerce bifurcate.

1. AI-Agents shopping on your behalf at AI-Marketplaces. That's cold and boring and sterile.
2. Brands who make customers feel special.

        

Join Daniel/Aaron from Orita.ai and I on Monday at 4:30pm EDT / 1:30pm PDT as we talk about ecommerce and bridging the gap between Executives and those of us with facts that need to be acted upon.

Click Here, now!!

        

By 1994 my promotion wasn't published anymore. By 2026 you self-published your promotion on LinkedIn (and earned 77 "likes").

But 1992?

In 1992 you'd submit org structure changes to DMNews so that you could communicate to the entire Catalog Industry just how sophisticated your intentions were.

• "LL Bean announced today the promotion of Shannon Ellison to Director of Circulation. She brings with her nearly eight years of experience. Ellison mentioned that she's '... looking forward to partnering with our printer to bring perfect binding to the Christmas catalog'. At press time, LL Bean has not decided whether the Christmas catalog will be 256 pages or 260 pages."