Ad measurement and attribution are in crisis – but maybe that’s a good thing.
“Actually, I’d even say it’s a great thing,” says Andrew Covato, founder and managing director of measurement consultancy Growth by Science. “It’s time for change.”
Covato launched Growth by Science, initially as a side project in 2011 while working full time building marketing technology within Netflix, eBay, Google, Snap and Meta (when it was still Facebook).
He was head of product for measurement and marketing science at Snap in 2019 when he first got wind that Apple was planning to release what later became its AppTrackingTransparency (ATT) framework.
ATT, which requires third-party apps to get permission from users before tracking their activity or sharing data, throws a wrench in the gears of campaign measurement by cutting off a key source of mobile tracking data. Since going into effect in 2021, the framework has been responsible for billions in lost revenue for Facebook, Snap, YouTube, Twitter and others.
But rather than setting his hair on fire in reaction to ATT, as did many others in the industry, Covato says he was pumped.
“Honestly, I was so excited when I heard about ATT,” Covato says, “because it felt like, ‘Heck yeah, this is inherently going to kill stupid measurement and bad ad tech practices.’”
That didn’t happen, of course. Despite significant platform changes and increased regulatory scrutiny, workarounds abound, which is only natural. The status quo may be unsustainable, but that doesn’t mean the industry isn’t going to try and maintain it.
Still, changes like ATT have fundamentally altered how advertisers and publishers need to think about campaign measurement, Covato says. “And to a certain extent at least,” he adds, “they’ve opened the door to new and smarter measurement platforms.”
I caught up with Covato for some real talk on true incrementality, data clean rooms, Apple’s SKAdNetwork and why advertisers should take platform-provided measurement with a grain – or even a shaker full – of salt.
AdExchanger: Why strike out on your own after so many years working at the big platforms?
ANDREW COVATO: Measurement is an unsung hero of the broader economy – unless you’re able to measure correctly, you don’t know how well you’re performing and you’re not able to optimize toward and connect with the people who actually buy things. When advertisers waste money, the only ones who end up winning are the ad platforms.
Just because there was an exposure on a platform doesn’t mean it deserves credit or even that the conversion was incremental.
That sounds like a pretty damning indictment of platform-provided measurement solutions coming from a man who used to be on the inside.
If you look at what a platform is calling out as its ROAS [return on ad spend] and what is an objective incrementality-based measure of that ROAS, there’s always a vast disparity. And which one looks better? It’s always the platform. It’s like marketing’s worst-kept secret.
Why do people knowingly look at incorrect data and think it’s fine – just because they have a number they can take back to their CFO?
Would you say that by now the ad industry has mostly lapped the effects of Apple’s privacy changes?
It’s a mixed bag. Some are trying to follow the Meta playbook, but they’ll never catch up. Meta has so much data that if it wants to create probabilistic conversion modeling, which everybody’s trying to do now to some degree, Meta is going to be better at it than anyone other than, maybe, Google.
I think a question advertisers need to ask themselves is: Do I want Meta to be the one telling me which ad exposures might have led to a conversion? Because that’s like double grading their own homework. They make up the question, answer it, then grade it.
What do you think of Apple’s SKAdNetwork?
It’s recreating the past because it’s a privacy-safe way to do last-click attribution. It’s an ingenious setup, and I have no qualms about the privacy aspect, but that’s not “good” measurement. Anything that is post-exposure is not good measurement.
Are data clean rooms actually good for privacy?
I have a lot of respect for the technology, but the problem I have is that clean rooms are just a better, more robust version of SKAdNetwork. I’m not talking about the methodology but, rather, philosophically, in that they’re recreating a pre-privacy world and trying to keep all the same data connections.
Clean rooms are a good privacy tool, but people will remain addicted to matching data at the user level because they’ll still be able to do calculations at the user level, just without seeing the underlying data.
This interview has been lightly edited and condensed.
Thanks for reading! And speaking of thanks, wishing you all a happy Thanksgiving, and may you enjoy your meals as much as this little guy. As always, feel free to drop me a line at [email protected] with any comments or feedback.
