The ever thought-provoking Don Marti wrote something in “This is why we can’t have nice brands” on February 17th that reminded me of a standard direct mail practice:

“Twitter has a solution that keeps its ads saleable: just don’t show any ads to important people. I’m surprised they can get away with this, but given the mix of rip-off and real brand ads I keep seeing there, it seems to be working.”

Back in the heyday of direct mail, we had small but important lists of people that we never wanted to promote. That list included the executives in our company*, state attorneys general and staff, and regulators of various types, among other folks.

You’d never want to accidentally put an aggressive direct mail piece into the mailbox of an AG who was running for governor, for example. The suppression file of a few thousand names was worth its weight in gold in avoiding legal issues and was trivially easy to do in the merge/purge.

But how does suppressing one’s own executives in direct mail relate to Twitter?

One of the main rules in direct response advertising is “test shouts, not whispers,” which reminds one that discontinuous and large lifts in profit don’t come from minor tweaks. That means you push the envelope in terms of creative, offer construction, format, and adherence to brand guidelines.

Sometimes the resultant tests, particularly if you’re running a complicated multivariate test, look downright weird. C-level executives have no problem understanding how a completed test–a new control–helps the quarterly earnings. But an odd-looking piece with no results (yet) sometimes gets a call to the big boy floor to explain. You get to talk about creative, test history, and the hypotheses that yielded the test. You might even need to discuss fractional factorial test design. Whoo-boy.

That’s a lot of Powerpoint and time to no good end.

To avoid this waste of time, you suppress the big Kahunas from your merge/purge, at all of their residences. As I indicated above with the asterisk, if they want to see the promotions, you re-insert their names into the bulk stream at the end of the merge so they can see what “everybody else” gets.

Back to Twitter.

If social media and digital marketing are the most targeted media ever created by humans, in theory we should be able to serve the right ad, to the right person, at the right time. (By the way, when you hear that phrase, push on it. You’re probably talking to a charlatan.)

As Don Marti points out, there’s a lot of crap in digital marketing. Mostly the worst-of-the-worst, bottom-funnel and outright scam ads. Why in the world would a brand like P&G run next to that kind of garbage?

The answer is that the executives never realize it.

Just as it’s easy to suppress names in a merge/purge, it’s easy to show your social media platform to be a clean, well-lighted storefront to those executives. You can get rid of the crummy ads and tweak the algorithms so the Russian trolls don’t show up in the feed. The media can be whitewashed for an audience of one or just a few.

Think I’m being a tin-foil hat conspiracy theorist? If Uber can build Greyball to avoid providing service to certain individuals, don’t you think the social media platforms can build tools to adjust the level of service to certain individuals to encourage advertising spend?

Takeaway: Check any digital media you’re going to buy from multiple “normal” accounts. Create the equivalent of seed lists to see the real environment you’re running in. And use independent monitoring services to check your ads in context. And win.

P.S. Yes, it sounds like I’ve run a kind of Greyball program in the past. However, every test I’ve run was always vetted by legal. And I’ve only run one test in my life that I wouldn’t feel comfortable rolling out. Call me to talk about that one!