For years, Bob Hoffman has been writing about the delusion held by advertisers that people want to interact with us or our ads.
I was an early adopter of that now-disproven fantasy. I bought in back in ’89-’90 when I worked Prodigy’s “interactive marketing” internal agency Look! Creative Group. It was fun (not) jamming JWT creative for Ford into 2 lines of 60 characters into some NAPLPS stencil.
I thought it was cool and I thought it was the future of advertising. I was also fooled by cold fusion, another big topic at the time. Mostly I was just looking for money for food while in graduate school and would have believed just about anything, if it paid.
We shared the same delusion about “interactivity” back then as we do now, but it took a while for digital online advertising to take off. Two main reasons: The first was that computers weren’t everywhere, but we knew Moore’s law would solve that problem eventually. The second was that there wasn’t any decent tracking. We definitely didn’t know what happened if the viewer didn’t click on the ad. As a result, advertisers were skeptical and were slow to move to digital.
Enter surveillance marketing. The ubiquity of tracking pixels, scripts, cookie synching, etc. has enabled the adtech and martech industries to sell us a bewildering array of B.S. that claims to track our every move and help serve the right ad to the right person at the right time.
The engine on the digital marketing B.S. train is view-through attribution. What this means is that we can supposedly tell if a viewer saw the ad–but didn’t act–and then attribute some of the subsequent action they take to that ad. Via a multi-touch attribution (MTA) model, which algorithmically attributes fractions of the conversion, we now have the “evidence” that non-interactive ads created interactivity. And that opened the advertiser floodgates, controlled mostly by gullible magpies.
The idea that a banner ad from yesterday should get almost any credit for the pair of shoes you buy today is laughable. Not because advertising doesn’t have an effect over time. It does. It’s just that the list of assumptions to believe view-through attribution is so long. Such as:
- Was the ad even served?
- Was it served on the site I thought it was?
- Was it served to a person?
- Was it served the right person?
- Was it served in a viewable location?
- Finally, was it even seen?
As I’ve written about before, you can’t assume any of the above to be true. Don’t believe it? Just follow Augustine Fou and Shailin Dhar. Within a week you’ll see their research and analysis on all of the above, and more.
Want to get a sense for how problematic this is? Start asking your marketing managers and agencies the following questions about view-through attribution:
- What percentage of the sales attributed to digital marketing have an action (i.e. click*) associated with it?
- Show me this equation: sum of all fractional view-through attribution / all “digital” sales
- How long is your view-through attribution window, by source and media?
Don’t be surprised if, for the first two, you see numbers exceeding 90-95%. Once you calm down, ask them to detail, specifically, what they are doing to ensure that the conditions I bullet-pointed above are being met. Ask for proof.
If your view-through attribution window is longer than 24… minutes, ask how it was arrived at and if an in-market clean and univariate test was used to arrive at the window(s). And ask to review the results**
Now freak out, once you realize that 95% of your digital marketing budget is being allocated by an algorithm via laughably long windows, built on assumptions that have never been validated.
Takeaway: The view-through is the money machine that powers digital marketing B.S. If you don’t understand exactly how it’s being used, you’re being taken to the cleaners.
* Yes, I know clicks aren’t everything. I also don’t believe most clicks are human anyway.
** Please use a disinterested third party to review the test design, analysis, and the data. You’d be shocked at how many people that could no longer pass middle school Introductory Physical Science are analyzing your tests.
You must be logged in to post a comment.