Prove It or Lose It: The Rear-View Mirror Is No Longer Enough to Defend a Media Budget
Most media budgets die the same way. Someone in finance asks a simple question, and the person defending the spend answers a different one.
The question is always some version of the one Alison Sidrane, EVP of Media at iProspect, put to a room of senior marketers on day three of Advertising Week: “If I give you $1, what am I going to get in return?”
The answer that usually comes back is a lift percentage, a click-through rate, a dashboard of site traffic, or a confident slide of attributed ROAS. Every one of those numbers is true. None of them answers the question. The CFO hears the mismatch, says nothing, and next year’s plan arrives ten percent lighter.
InMarket’s annual AWNY Lunch Club was held at Bar Harta on West 39th Street. Kelly Metz, Chief Investment Officer at Spark Foundry, described it as a “lovely stage slash restaurant slash high top table.” The session was billed “Prove It or Lose It.” Catherine Flint Chen, SVP of Measurement at InMarket, moderated with Sidrane, Metz and LeShaun Jones, who leads ads measurement for restaurants and delivery at Pinterest. Chen framed the problem in one line: how do you prove the value of marketing “to a CFO who sees things solely in dollar signs and not in the nuances, the infinite nuances that go into building marketing campaigns?”
Over the next hour the panel kept coming back to one point. The nuance is not the problem. The tense is.
The question finance is actually asking
Sidrane introduced herself with a job description that turned out to be the thesis of the whole session. If her title came with a subtitle, she said, it would be “translator: translating between the business problem, the media, the data, and all of the teams that need to action on it.”
Translation is where most budgets are lost. The metrics media teams use to optimize campaigns are the wrong language for the boardroom, and Sidrane did not soften it.
“Numbers like sales lift or site traffic, or any of those metrics that we would use to optimize our campaigns, I would never even use those words in front of a CFO,” she said. “If I want to justify my media budgets, if I want you to give me more, I want to then take those metrics and figure out how to turn that into: that actually drove this amount of lifetime customer value, or it drove actual sales numbers. That is a number you can take to the CFO.”
Metz described the tension from the investment seat more bluntly. “It’s very challenging, because I like to spend money, and they want to keep the money where they have it.”
Beneath the joke is a structural mismatch. Finance thinks in quarters. Marketing, at its best, thinks in customer lifetimes and brand equity. “They’re focused on short term,” Metz said, “and a lot of the marketers that we work with are focused on perhaps something longer term like lifetime value or brand building.” What closes the gap is not a better report. When Spark Foundry arms a CMO for the CFO conversation, Metz said, the ammunition is “very specific forecasts slash capabilities against sales. That’s where we all end up, because we have to speak their language. They care about revenue and they care about profit. That’s how you win.”
Notice the word forecasts. It came up repeatedly, from the people who control the money.
SPONSOR MESSAGE: translating media metrics into revenue and profit language the CFO can sign off.
The industry’s unicorn
Chen asked each panelist what is overhyped in measurement right now. Sidrane went straight for the industry’s favorite fantasy.
“Overhyped is this notion that there is this perfect single source of truth that can tell me every channel, every action, every consumer, and what that outcome was,” she said. “Different frameworks exist for a reason. They answer different questions, and I think we almost create false precision when we try to make them all fit into a little box that they don’t actually fit into.”
False precision is the right diagnosis. A single number feels like certainty, and certainty is what the finance team wants. But a number that hides its own assumptions is not certainty. It is a liability that hasn’t been called in yet.
Jones, who spends her days reconciling MMM, MTA and lift studies, explained why the unified view is always partly broken. “A unified view is possible, but I think it will always devalue something,” she said. “There’s always going to be something you have to give up. With high-impact media, the goal is as many eyeballs as possible. With lower-funnel media, you want that very high-quality click. I don’t think there’s any solution that’s really going to show you both of those things as the most important.”
Metz agreed it is a unicorn, then named the deeper flaw in the industry’s most trusted tool. “MMM’s been here for a very long time and it still works,” she said. “But the biggest challenge around that is it’s a rearview mirror.”
Sidrane offered the resolution the room had been waiting for. “A unified view, no. But a unified framework, absolutely.” For iProspect that means an audience-led approach that connects brand through to demand, “pulling every channel into an ecosystem that you can then optimize against, all towards business outcomes.”
The distinction is not semantic. A view is something you look at. A framework is something you decide with.
Who marks the homework
When the platform, the agency and the measurement provider sit at the same table, bias is the obvious subject. Chen raised it directly, noting that nobody is being dishonest on purpose. Bias does not need bad intent. It only needs to be built into how credit is counted.
Jones described it from inside a platform. “We’re in a world where search and social are playing tug of war to get the credit for a conversion or a purchase,” she said. “The bias a lot of times comes in with last-touch attribution or site analytics, or even MMP attribution, where there’s a leaning to want to give the credit to the most recent purchase, but it doesn’t take into account the entire purchase journey.”
Last click survives for one reason. In Jones’s words, “last click is the easiest way to determine performance.” Ease is the trap. A system that rewards the final touch will, over time, starve everything that made the final touch possible.
Metz pointed to a quieter source of distortion: “partner-funded studies.” She did not dismiss them. “There’s a ton of great testing that can happen with each individual partner,” she said. But the agency’s job is to “bring it all together into a cohesive point of view.” What she wants from all of it is “a confidence level,” not a verdict.
Chen asked whether data transparency had improved. “Yes and no,” Metz said. “With certain partners it’s absolutely gotten better, but with other partners it’s gotten worse. It’s like whack-a-mole. One day you’re getting really good signal, and the next day something’s been obfuscated that’s unexpected.” Her conclusion was operational, not philosophical: “The data check process is really essential to everything that we do.”
SPONSOR MESSAGE: “The next step is pushing the modeling capacity further towards actionability at the granularity marketers can use every day, week, month and quarter. Fospha’s always-on infrastructure is the best place I know to start with.” said Gufeng.
The math is never the fight
Chen then asked a deliberately provocative question. If everyone in the room switched to fully causal measurement tomorrow, what internal fight would break out, and who would win?
Sidrane’s answer was the most human moment of the session. “It’s not the math,” she said. “People sit down and they understand the methodology. It’s the change management.” She described a scene she has lived through many times across CPG clients: “I can’t tell you how many times I have sat in front of a media client and he’s said, ‘Well, that’s great, Allie, but I get bonused on what my MMM says.’ And then you’re back to square one.”
That is the mechanism under most measurement stalemates. People defend models the way they defend anything their pay depends on. A better methodology does not beat an incentive structure. Trust does, slowly. “You can’t just throw the baby out with the bathwater,” Sidrane said. Chen predicted that would be the line that got published, and she was right.
Metz reframed the fight entirely. “Media always wins. You want to know why? Because it drives sales,” she said. “It’s never a question of will they spend. If it is a question of will they spend, I fear for their entire business.” The real argument, she said, is about mix: which channels, in what proportion, and to what end.
Then she added the point that separates people who measure to defend from people who measure to learn. “We also have to get comfortable having the flip side of that conversation. What would have happened regardless of where I spent my media dollars, or what that channel mix looks like. That also helps drive credibility, when you’re willing to have that other conversation.”
A marketer who will discuss the counterfactual gets believed. One who only brings good news gets audited.
Quarterly is no longer fast enough
Jones raised the issue of speed, and it may matter more than anything else on the list.
“MMM isn’t underhyped. It’s an appropriate measurement solution that a lot of advertisers use for accurate results,” she said. “But the speed with which you get your MMM results has been changing, and the desire to have a more agile model that gives you monthly readouts. Quarterly is no longer fast enough. Monthly is the name of the game.”
She also noted that matched-market testing has had “a big resurgence in conjunction with MMM and MTA,” as advertisers try to triangulate across methods rather than trust any one of them.
Consider what a quarterly readout means in practice. A model that reports in April on what happened in January is describing a market that no longer exists. The creative has rotated. The auction has moved. A competitor has doubled its budget. The finding may be correct, but it arrives too late to change anything.
Metz added a gap nobody in the industry has closed: “Impact of creative. I don’t think we’ve got great solutions or frameworks in place to deal with the impact of creative on our results.” Most of what drives performance changes week to week, and most measurement still reports on it months later.
This is the point where the session stopped being about measurement and became about something harder.
A report explains. A forecast decides.
Asked where the promise of measurement lies, Metz did not name a new attribution method or a better dashboard.
“With any measurement framework or data source that we’re looking at, and what the promise holds for the future, it’s forecasting models,” she said. “Forecasting: our ability to actually understand the past in a way that we can influence into a forecast for where we may want to go in the future, so we’re not just looking in our rearview mirror the whole time.”
“I think all the CFOs would love to hear you say that,” Chen replied.
They would. A CFO allocating capital is making a bet on the future, and every other function in the business brings a forecast to that conversation. Sales brings a pipeline projection. Operations brings a demand plan. Marketing too often brings a retrospective and asks to be trusted with what comes next.
Measurement that only explains the past leaves the most important question to instinct: where should the next dollar go? Answering that takes a system that can model the response curve of each channel, show where diminishing returns begin, and test a reallocation before the money moves. That system can tell a CMO, in revenue terms, what happens if paid social goes up twenty percent and search comes down ten. Without it, every planning cycle repeats last year’s split with a few adjustments.
SPONSOR MESSAGE: forecasting and budget planning; scenario-testing reallocations against predicted revenue before the plan locks
Before the plan locks
With 2027 plans being finalized across the industry, Chen asked each panelist for one thing to do before budgets are locked, not after.
Jones: “Make a learning agenda that’s closely tied to your measurement system of truth, so that you’re getting ahead of it and you’re not introducing bias into how you’re interpreting results. When you just jump into it, you tend to over-rely on past performance and really not test and learn to see what works in the long term.”
Sidrane: “Before you spend $1, make sure that you understand exactly what it is that you are looking to measure and what you are looking to learn. Don’t go to measure it after the fact.”
Her underhyped pick earlier in the session was the same idea, “just plain old measurement design.” She apologized that it made her “sound really old school.” Chen drew out the consequence: get the design right and “the storytelling after it gets to be that much easier,” because you built the measurement to answer the question finance was always going to ask.
Metz asked for something rarer in an annual plan: room to move. “I’m looking for fluidity. I’m looking to chase what’s working on behalf of my clients,” she said. “I like to see large chunks of money held to do very specific things. That’s what I ask for from a CMO and a CFO as we start looking at annual plans: basically preparing for the unexpected.”
“Things move too fast to lock everything in at the beginning,” Chen said.
Taken together, the three answers describe one discipline. Decide what you need to learn. Design the measurement before the spend. Model the plan forward so you know where the next dollar works hardest. Hold back enough budget to act when the model tells you something you didn’t expect. None of it is new. Very few teams actually do it.
SPONSOR MESSAGE: always-on measurement that tells you where to move held-back budget in-flight
The number that actually matters
Chen shared a piece of evidence that showed the whole argument at work. Across 35 campaigns InMarket ran with Pinterest, the average incremental lift was 33%. In dining it was sharper still: more than 44% incremental lift among new diners, and 0.4% among existing ones.
That gap is the most useful number from the lunch. Loyal customers come regardless. The ads mostly preach to the converted. The return lives at the margin, with people who would not have walked through the door without the media. A last-click model or a blended ROAS would never separate the two.
Jones’s response showed why she has run so many of these studies. A strong result is not the same as a useful one. “There’s always going to be some result that is really good, and we’re very excited about it,” she said. “But if it’s not tied to a business objective, or what the CFO is using to make business decisions, it’s not as impactful. If we see results in something that’s a low priority, that’s great, but it’s not necessarily going to move the needle.”
The 44% only matters if new customer acquisition is what the business is trying to buy. Whether to fund more of it depends on what the next tranche of spend will produce, not on what the last tranche did.
A world without clicks
Chen closed by asking what metric or method the panel expects to be discussing two years from now.
Sidrane was clear. “It has to be AI-mediated journey measurement. We are no longer in this perfect world where a consumer sees an ad, clicks on the ad after they’ve searched, and then buys,” she said. “We shouldn’t repeat the past mistakes of attribution. We should really look at causality.”
Metz followed that through to its conclusion. “In a world without clicks,” she said, “no one’s going to click.” Her prediction was not a new metric but a return to old ones: “I think we’re going to see a little bit of history repeating itself. We’re going to revert to things we know work as leading indicators of consumer behavior, but informed by the AI of it all and how the journey is changing.”
Jones wanted an end to relying on any single metric: “balancing holistic measurement with more precise measurement, the zoomed-out picture with getting in the weeds.”
“What’s old is always new again,” Chen said, “even when the new is really new.”
When the click disappears, every model built on the click loses its footing. What remains is causality, the counterfactual, and a forward view of where money should go next. Measurement will have to return to the question it should always have been answering: not who gets the credit, but what happens if we spend.
Sidrane’s question is still the one that matters. If I give you a dollar, what will I get back?
Nobody has ever won a bigger budget by explaining what already happened.
Lee Arthur attended InMarket’s AWNY Lunch Club, “Prove It or Lose It: How the Industry’s Biggest Brands and Platforms Are Measuring and Optimizing Media,” on Wednesday 7 October 2026 at Bar Harta, New York, during Advertising Week New York.
Leave a Reply
You must be logged in to post a comment.