Skip to content

Voices from the open internet

Opinion

Credit where it’s due: The next era of advertising accountability

Attain Co-founder and CEO Brian Mandelbaum on why measurement only matters if it makes the next decision better.

A progressively growing bar chart, with the tallest bar being made of a price tag.

Illustration by Christian Ray Blaza / Getty / Shutterstock / The Current

By

Co-founder and CEO, Attain

November 17

The question I hear most from marketers is three words long: “Did it work?” It sounds like it should have an easy answer. It almost never does.

Think about how someone actually decides to buy something. They see a TV commercial that introduces a brand. A week later, a digital ad catches their eye on their phone. They mention it to a friend, compare a few options and eventually pick something up at the grocery store on a Tuesday night. Ask that person which of those moments made them buy, and they probably couldn’t tell you. Ask the industry, and you’ll get a different answer depending on who’s doing the selling.

But ultimately, a purchase tells us something that an impression or a click never can. 

That matters more every day, as the infrastructure powering advertising increasingly becomes automated or devoted to AI and emerging technology, because more of advertising is being handed to machines.

I’ve spent the last decade building Attain around a simple, human observation: what people spend money on is a direct reflection of what’s important to them. Someone can spend weeks researching a product and never buy it. They can watch an ad, never click and buy it ten days later. Both behaviors tell us something. Only one of them is a decision.

Over the last 30 years, our industry got very good at measuring what was easy to measure: impressions, clicks, views, time spent. Each one was useful, and each one became a stand-in for the thing advertisers actually cared about, which was whether someone bought. We built an enormous, sophisticated system on top of proxies. The gap between ad engagement and real-world behavior never really closed.

The next era of advertising is about closing that gap, and that will take accountability: advertisers should be able to base their investment on what most closely predicts what their customers will do next, and consumers deserve to understand how their data is used and get something of value in return. 

An average is not a market

For most of digital advertising’s history, a lot of media has been valued on averages. An impression that changed someone’s mind and an impression that scrolled past at the bottom of a page often get credited the same way. When you can’t tell the difference, you price them the same. That isn’t a market. It’s an average. And averages quietly punish the publishers and channels doing the best work.

That’s what makes The Trade Desk’s Open Measurement interesting to me. It starts from something anyone who has bought media already knows: not every impression contributes equally to a consumer decision. By accounting for the quality of different exposures, it creates an opportunity to recognize the value of environments that have been hard to evaluate through conventional attribution, and to stop over-rewarding the ones that have simply been easy to count. For what it’s worth, Attain is the foundational data partner  of this initiative, contributing our real-time verified consumer transaction data as an essential input to the methodology.

Media platforms understand exposure and engagement. Brands understand portions of their own sales. Neither view tells the whole story. Bring them together with verified purchase behavior and you can get much closer to the truth about what advertising contributed. I’m proud that our data is part of that foundation.

Measurement tells you what happened. Signal decides what happens next.

Measurement on its own isn’t the point. 

That may sound odd coming from a company who has a measurement product in market,  but no advertiser invests in media to understand the past. They invest to grow. Knowing what a campaign accomplished only matters if it makes the next decision better. That’s the business case, which is easy to lose sight of when everyone is arguing about attribution models.

I’d argue that it’s the behavioral pattern, or consumer signal, that’s the most valuable to advertisers. That signal decides who a campaign reaches, what a buyer is willing to pay for an impression, if that impression has a chance at converting, and where the next dollar goes. If the signal is weak, even perfect measurement just confirms that the money went to the wrong place.

And the systems making those decisions are increasingly learning on their own. Every outcome becomes an input to the next round of decisions. That recursive learning is powerful, but it isn’t neutral. It gets better at whatever it’s taught. Teach it clicks, and it gets very good at finding people who click. Teach it what people actually spend money on, and it starts to understand how they live.

Picture a family that just moved across town. Over six weeks, they make three trips to the home improvement store. They switch grocery stores because the new one is on the way home. They start shopping for car insurance, because a new zip code means a new rate. They buy a couch and a lawn mower, and they order a lot of takeout while the kitchen is still in boxes.

A model trained on clicks sees a handful of ad engagements. A model watching one product category sees one purchase. A model that learns from the full picture of what that household spends money on — the products they buy, the stores they visit, the services they shop for — sees a family in the middle of a life change. That’s relevant to the insurer, the hardware chain, the furniture brand, the grocer and the restaurant down the street, all at once.

That’s what I mean when I say recursive learning on signal isn’t one-dimensional. The question isn’t only whether someone bought the product in the ad. It’s how their economic behavior changed after they were exposed: where they shopped, what they started considering, which categories they moved into. Every campaign adds to that understanding, and the next one starts from a better place. The learning compounds.

Planning, buying and measuring shouldn’t be three different conversations

Most campaigns are still run as three separate exercises. Planning happens before launch, optimization happens during and measurement happens after, often with different data and different definitions of success at each stage. That’s how you end up with a plan built around one audience, an optimization tuned to clicks, and a recap that tries to connect both to sales after the money is already spent.

Purchase-based insight should run through all three. Before launch, advertisers should know who they’re reaching and how those people actually spend. While a campaign is running, they should see what’s working and adjust. Afterward, they should be able to show results their CFO can stand behind. Open Measurement plans to be a meaningful step toward connecting those stages and making purchase data part of the infrastructure of buying media, rather than something that shows up after the fact.

What this asks of all of us

We’re not there yet as an industry. That infrastructure has to be transparent about how credit is assigned and where the data comes from. As AI takes on more of the work, that isn’t optional. A system you can’t inspect is a system you can’t hold accountable, no matter how good its results look.

Much of the signal feeding buying decisions today is still built on proxies for intent rather than evidence of it. Fixing that asks something of everyone. Data providers have to be clear about where their data comes from. Platforms have to be open about how credit gets assigned. And advertisers have to be willing to judge media on what it actually did for their business, even when the answer is inconvenient.

I also want to be honest about the limits. No methodology will ever perfectly explain why someone bought something. People are more complicated than that. The goal isn’t certainty. It’s a more reliable understanding of what an investment accomplished, and better information for deciding what to do next.

So the next time a marketer asks me, “Did it work?” — I want the answer to come with evidence. And I want that same evidence to make the next campaign smarter than the last.

Credit where it’s due, and a clearer path to turning that performance into growth.


This op-ed represents the views and opinions of the author and not of The Current, a division of The Trade Desk, or The Trade Desk. The appearance of the op-ed on The Current does not constitute an endorsement by The Current or The Trade Desk. The Current is owned and operated by The Trade Desk, Inc.

Open Measurement is currently not yet available. This information is provided solely for background and is not a representation or guarantee of any future availability or performance. Any timelines and information included are not guaranteed and are subject to change. This content contains forward-looking statements under the Private Securities Litigation Reform Act of 1995. Actual results may differ materially due to risks in the Company’s SEC filings (www.sec.gov). Do not rely unduly on these statements. The Company does not intend to update them.

Subscribe to The Current

Industry news, expert analysis, behind-the-scenes podcasts and more—straight to your inbox.