Andrew Covato keeps seeing the same thing happen with retail media networks: a brand asks for data they can feed into their marketing mix model, the RMN misunderstands the request, tries to sell them an attribution dashboard instead, and the brand walks away.

Covato has spent 15+ years building measurement programs at Google, Meta, Netflix, and Snap. He now runs Growth by Science, a consultancy helping brands build custom incrementality testing and MMMs — so he's seen this from every angle. And on a recent episode of the Unlocking Retail Media podcast with host James Avery, he laid out why RMNs keep losing credibility here: it's not that the ads don't work. It's that they can't prove it in the language their most valuable advertisers speak.

Here are three things I took away from this conversation about what RMNs are getting wrong on measurement.

Retailers know that a marketplace model can dramatically boost product assortment, shopper engagement, and total revenue.

But, to get the most out of your marketplace, you need an ad tech solution that can really engage sellers.

Mirakl Ads is powering the future of retail media for leading retailers — to activate both 3P sellers and 1P brands.

Learn more

1. The big ad platforms stayed measurement-agnostic. RMNs didn't.

Andrew's point here is that platforms like Meta, Google, and Snap didn't succeed at measurement because they're altruistic. They're optimizing their own revenue — but they figured out that the way to do that is to support whatever measurement the advertiser wants to bring. Incrementality, MTA, MMM — if enough demand existed, the platform built for it.

"Ad platforms are not necessarily in the business of optimizing advertiser outcomes," Andrew said. "They're in the business of optimizing their own revenue." But the smart ones realize you can't do that without being credible — so they stay measurement-agnostic and build products based on what advertisers are actually asking for.

RMNs tend to do the opposite: pick an attribution model and hand it to every advertiser regardless of what they actually need.

2. RMNs are answering questions their biggest advertisers have stopped asking.

Andrew sees RMNs over-indexing on granular user-level path-to-purchase data — the idea that proving a touchpoint preceded a conversion is enough to demonstrate value. And he's pretty direct about why: "RMNs are not typically the primary revenue stream for folks, right? This is retailers that realized in the last few years that there's some great monetization potential. So they're by nature not caught up with what I would say as natives to ad tech."

In other words, they're retailers playing at ad tech, and the measurement reflects it.

The most sophisticated advertisers have moved on to modeling, experimentation, and aggregate data feeds. "That's kind of the old-school way of doing things," Andrew said of the path-to-purchase approach. His advice to RMNs: figure out what an experiment requires, what an MMM requires, and build the APIs that let advertisers pull data in those formats. Don't abandon attribution — supplement it.

This tracks with something I keep hearing from brand-side measurement teams: they're not asking for more dashboards. They're asking for clean data exports they can feed into their own models. The RMNs that grasp this distinction will hold onto their most valuable accounts.

3. First-party lift is grading your own homework. Just be honest about it.

James asked Andrew how he'd design incrementality measurement for an RMN, and Andrew laid out a graduation framework:

  • Step one, for new advertisers: Start with directional attribution. Acknowledge it's imperfect but use it to prove that people seeing the ad are buying the product.
  • Step two: First-party lift studies. Low cost, no third-party dependency — but Andrew didn't sugarcoat it: "I would look at first-party lift as more of a sales tool versus a measurement tool. Go into it fully acknowledging — look, we are doing our best to be objective. But we understand that you may not look at this as purely objective because it's our own platform." That kind of honesty, he argued, actually builds more credibility than pretending the study is neutral.
  • Step three: Third-party geo testing and MMM integration for bigger spenders who need independent validation.

Most RMNs are stuck at step one.

But what Andrew is describing here is what sophisticated brands need. A huge chunk of brands aren't there yet. When I spoke with Liz Roche, VP of Media and Measurement at Albertsons Media Collective, about their iROAS research published this week, she made the point that many brand partners receive a report and simply don't have the team to question what's behind it. That research found that 83% of campaigns can flip from positive to negative iROAS based on methodology alone — same media, same spend, same audience.

Which means transparency isn't just a top-tier CPG concern. It's essential for the mid-market brands taking the number at face value. The RMNs willing to show their working — including the uncomfortable parts — will earn more trust than the ones posting the highest possible number.

So what?

Retail media has a measurement gap at both ends. The most sophisticated brands can't get the data formats they need. The mid-market brands can't evaluate what they're getting. Andrew's advice to RMNs is the same regardless: be upfront about what your measurement does and doesn't prove, build the APIs and geo-targeting that advanced advertisers require, and bring in people who really understand the "why" behind the ask.

The RMNs doing this now will hold onto their most valuable advertisers. The ones still pitching the same attribution report to everyone will keep losing deals at the top and credibility in the middle.

Check out the Unlocking Retail Media podcast for the full conversation.