By: Sarah McWilliams Guerra
Attribution is having a moment. As retail media matures and closed-loop measurement becomes the gold standard, the industry has quietly rallied around a new default: impression-to-purchase attribution. If an ad was served and a purchase occurred within a defined lookback window, credit is assigned. ROAS is calculated. Everyone moves on.
But there’s a fundamental assumption baked into this model that deserves more scrutiny: that an impression is meaningful.
An impression sounds concrete, but in practice it’s surprisingly abstract. It simply means an ad appeared on a screen. Not that it was noticed. Not that it was processed. Not even that it was visible for more than a split second. Yet we’ve grown comfortable treating impressions as if they represent influence – sometimes even weeks after the fact.
This is where attribution starts to blur the line between measurement and imagination.
The idea that a shopper “saw” an ad is already debatable. In a fast-scrolling environment, surrounded by competing content, “on screen” does not automatically translate to “in mind.” And yet, we increasingly accept the premise that a shopper could passively encounter an ad, take no action whatsoever, and then – ten days later purchase the item in store, with the impression receiving full credit for the sale.
That’s a strong claim. Stronger than the data can usually support.
Clicks, meanwhile, have fallen out of fashion. They’re often dismissed as too performance-driven, too lower-funnel, or not reflective of true brand influence. But in doing so, we’ve sidelined one of the clearest behavioral signals available to us.
A click isn’t perfect, but it is intentional. It requires a shopper to notice the ad, find it relevant enough to interrupt their journey, and take an action.
In a world where we’re debating whether someone even saw an ad, a click answers that question decisively.
Closed-loop measurement is powerful, but it doesn’t eliminate bias. Matching ad exposure to purchase data improves precision, but it doesn’t automatically prove causation. Many shoppers were already in-market. Many would have purchased regardless. When impression-based attribution ignores engagement entirely, it risks over-crediting media that happened to be present rather than media that actually influenced behavior.
The longer the attribution window, the more this problem compounds. At some point, impressions stop representing influence and start representing proximity. Whoever was last in the ecosystem gets the credit, even if they did nothing to earn it.
None of this is an argument against impressions or upper-funnel media. Awareness matters. Reach matters. But not all impressions are created equal, and treating them as such flattens the truth instead of clarifying it.
A more honest way to think about attribution is as a spectrum of signals. Impressions create the opportunity for influence. Clicks provide evidence of interest. Purchases confirm outcomes. When we collapse all of that nuance into a single impression-to-purchase line, we don’t get better measurement, we just get more convenient math.
Clicks shouldn’t be ignored or treated as outdated relics of early digital advertising. They should be weighted, contextualized, and used as a reality check. If a shopper clicked and later purchased, that’s a strong signal. If they clicked and didn’t purchase, that’s even more interesting for brands and retailers. Why didn’t they convert? How do we get them to convert next time? If they didn’t engage at all, the attribution story deserves more humility.
As impression-to-purchase becomes the industry standard, the real question isn’t whether it’s scalable or easy to report. It’s whether it’s honest.
Closed-loop attribution is only as credible as the assumptions behind it. And assuming that every impression meaningfully influenced a future purchase doesn’t make our measurement more sophisticated—it just makes it more flattering.
Clicks aren’t the enemy of modern attribution.
They’re accountability.