The Seven-Day Ghost: How a new cross-media study exposed the most expensive case of misattribution in modern advertising

There is a sentence that nearly every adult in America says approximately twice a week, has been saying for the entirety of their adult life, and will continue saying until they are dead. It is also, by some distance, the single most consequential sentence in consumer marketing, which is a thing nobody in consumer marketing seems to have noticed. The sentence is: "I'll go this weekend."

You see a billboard on Tuesday during the commute, the one for the home improvement chain with the new grills out front. You think, in the kind of half-conscious way that all interesting consumer decisions actually happen, that you want a new grill. You do not buy a grill on Tuesday. You do not buy a grill on Wednesday or Thursday or Friday. You buy a grill on Saturday morning, because Saturday morning is when adults in America buy grills, and you have known this since you were eight years old.

The receipt in the Home Depot point-of-sale system has a Saturday timestamp. The Google Maps query in your phone has a Saturday timestamp. Whichever digital coupon happened to fire in your inbox between Tuesday and Saturday will, in the various dashboards consulted by various people in Manhattan and Mountain View, be quietly credited with the conversion. Somebody on a marketing team will look at a number on Monday morning and conclude that the coupon "drove" the purchase. The billboard, meanwhile, will have spent the entire week bolted to the side of a building, doing nothing measurable. Or so we have been told.

A research project published this month by Kochava and the OAAA quietly demolished this assumption. The way they did it was very simple and somewhat embarrassing for the rest of the advertising industry: they extended the attribution window. Most digital measurement runs on a window of one to seven days, because the digital industry has decided, without consulting any actual humans, that this is how long meaningful consumer decisions take. Kochava ran the window out to twenty-eight days. And what they found, sitting in the data that nobody had bothered to look at, was a ghost.

Or maybe not a ghost. Maybe a pattern so visible and so consistent that the only reason it had stayed hidden this long was that the people responsible for finding it had been measuring the wrong thing on purpose.

The top-line finding will not feel novel to anyone who has actually worked in OOH, but it should feel deeply uncomfortable to anyone who has spent the last decade calling Search and Social "performance channels." OOH delivers twice the median action lift of broadcast TV. Twenty percent for in-person actions. Fourteen percent for digital actions. Versus ten and a half percent for CTV, and ten flat for broadcast. We are not talking about a tie. We are talking about the medium that the industry has spent a generation describing as "the brand channel" outperforming the screens that are bought, planned, and measured under the assumption that they are the performance channels.

There is the usual instinct here, which is to discount the study because OOH stands to benefit from it. The instinct is wrong. The study was run by Kochava, which is a mobile measurement partner. Their entire commercial reason to exist is making digital media measurable. Their customers are the same people who buy Search and Social. The implicit message of the study, which Kochava has now published under their own brand, is that the channels Kochava primarily measures have been quietly siphoning credit from a channel they do not. This is not the kind of thing a vendor publishes for fun. This is the kind of thing a vendor publishes when the data has become impossible to ignore.

The data has become impossible to ignore. Several hundred campaigns. Device-level tracking. Seven verticals. Multiple conversion types. Pre-campaign matched control groups. Synthetic-control methodology drawn from an six-billion-device panel. The entire apparatus that the digital industry uses to defend its own measurement quality, applied for the first time at this scale to a medium that was, until recently, measured in approximate visibility cones and quarterly survey panels.

The number is twenty percent. Two times the median lift of the closest competitor. And it does not even include the part where it gets weird.

Here is the part where it gets weird.

Most marketers, when they think about how an ad performs over time, picture an exponential decay curve. You run a campaign, it generates response on day zero, the response declines steadily over the next week or two, and then it tapers off into noise. This is how broadcast TV performs. It is how CTV performs. It is how nearly every digital channel performs. The decay curve has been the default mental model for media performance for at least sixty years.

OOH does not do this. OOH does this for six days, and then, on the seventh day, the curve goes the other way.

The specific number, from the Kochava data on QSR campaigns: on day zero, 17.27% of OOH-exposed devices generate an in-person conversion. By day six, the number has declined to 4.4%, exactly as a decay curve would predict. On day seven, the number jumps to 10.4%. That is a 136% increase from the prior day, and it has no equivalent in any other medium ever measured. The pattern declines again for six days. On day fourteen, it does the same thing. On day twenty-one. On day twenty-eight. The Kochava team ran the attribution window out four full weeks, and the bump happened four full times.

This is not statistical noise. This is the heartbeat of American consumer behavior, laid out in a graph for what is probably the first time in the history of advertising.

The reason it shows up only in OOH is the reason it has been invisible until now. Digital attribution windows are short because digital channels assume short decision cycles. Click the ad, buy the thing. The Google ad you saw at 8:47 AM should produce the conversion by 8:52 AM, or it never will. This is not how humans actually work. This has never been how humans actually work. Humans operate on a behavioral rhythm that has nothing to do with click-through latency and everything to do with the structure of the week. The work week ends. Saturday begins. The thing you saw on the billboard becomes the thing you go buy. The Sabbath predates the personal computer by approximately three thousand years. The weekly retail trip predates direct response advertising by roughly the same interval. The "I'll go Saturday" mental note is a load-bearing piece of cognitive infrastructure, and the entire field of digital attribution has been operating as though it does not exist, because acknowledging it would require attribution windows that don't fit neatly inside a quarterly performance review.

The Kochava study makes the cost of this denial specific. Measuring to day seven, which is what the digital industry does, captures 52% of OOH conversions. The other 48% happens between days eight and twenty-eight, with measurable resurgence bumps at days seven, fourteen, twenty-one, and twenty-eight. If you stop measuring at the digital industry's preferred window, you are missing 48% of the value generated by the OOH portion of your media plan.

Forty-eight percent. Not a rounding error. Not "directional." Half of the medium's actual performance, gone, because the industry decided to measure on a calendar that humans do not live on.

Let me try to explain what this means in a way that will be deeply unwelcome to a specific kind of reader.

The Kochava team did something almost no one in this industry has bothered to do at scale, which is map the actual sequence of advertising exposures that precede a digital conversion. They took 10.8 million app install events from a major QSR brand, isolated the 86,082 devices that were exposed to both OOH advertising and paid Search during the campaign, and asked a simple question: when both exposures occur, which one happens first?

The answer is OOH. 95.7% of the time, OOH was the first exposure. The Search interaction came afterward. The Search ad got the last-touch credit. The Search team got the bonus. The OOH line item got cut next quarter to fund more Search. This has been the loop for fifteen years.

For Social ads, the same study, the same campaign, the same kind of analysis: 93.8%. Same direction. Same magnitude. Same outcome.

If you have ever watched a baseball game with someone who cares about pitching statistics, you have witnessed an argument about saves and wins. The pitcher who throws seven shutout innings does not get the save. The relief pitcher who comes in for the ninth and strikes out three batters does. This is a known weirdness about baseball statistics. Nobody who actually understands the sport believes the closer is the one who won the game. Everyone who actually understands the sport understands that the starter did the work.

Performance marketing has spent fifteen years giving the save to the closer and calling it a win. The Kochava study is the box score that says otherwise.

The phrase "performance marketing," in the way the industry currently uses it, describes the channel that closes a conversion, not the channel that causes it. These have always been different operations. The Search ad converts an existing intent. It almost never creates one. Somebody had to want the brand first. Something had to put the brand in their head. That something, 96% of the time in the QSR data, was OOH. The Search budget gets the credit. The OOH budget pays the rent.

This is not a defense of OOH. This is an indictment of how the rest of the industry has been doing math.

There is a separate finding in the Kochava data that contradicts a different piece of received wisdom, and it is worth pausing on because it is the cleanest data in the entire study.

The received wisdom is that high-frequency advertising produces diminishing returns and consumer irritation. Show somebody an Instagram ad ten times and they will block your brand. Show them the same YouTube pre-roll three times in a row and they will associate your product with frustration. This is true. It is true because digital advertising is active advertising. It requires the consumer to either consume it or refuse it, and refusing it requires effort. After enough refusals, the consumer's effort calcifies into resentment.

OOH does not work this way. OOH is passive advertising. You don't choose to see the billboard. You do not have the option to mute it or skip it. But you also do not develop a conscious resentment toward it, because it lives in the periphery of your attention rather than at the center. It is the difference between somebody whispering a fact in your ear every morning during your coffee and somebody emailing you the same fact ten times. Both deliver the message. Only one of them gets you to file an HR complaint.

The Kochava data quantifies this distinction in a way that should produce approximately a hundred internal-strategy memos at every holding company. From one exposure to ten exposures, OOH digital conversion rates climb from 0.24% to 1.28%. A six-fold increase. The cleanest upward curve of any media type in the study. Broadcast TV, by comparison, plateaus at 0.22% and barely moves. CTV climbs gently. OOH just keeps going up.

This means that the planning instinct in digital media, which is to cap frequency and rotate creative aggressively to avoid burnout, is precisely backward for OOH. Extended flight durations beat short bursts. Market saturation beats narrow targeting. Sustained presence beats clever creative refreshes. The exact things that digital media has been optimizing against for fifteen years are the exact things that OOH has been optimized for the entire time, and only now, with the kind of frequency curve that Kochava just published, can the strategy actually be defended with numbers instead of intuition.

The intuition was correct. The numbers caught up.

Step back. The thing the Kochava study is actually doing, underneath the percentages and the precedence rates and the seven-day bumps, is dismantling a structural assumption that has organized the advertising industry for two decades.

The assumption is that advertising channels can be cleanly sorted into "brand" and "performance" buckets, and that the budget conversation is a question of how much you allocate to each. Brand budgets are evaluated on lift studies, recall surveys, and qualitative attitude measures. Performance budgets are evaluated on conversions, CAC, and ROAS. The two are managed by different teams, measured on different timelines, and frequently funded out of different P&L lines. There are conferences dedicated to the brand-performance divide. There are entire job titles built around bridging it. The Procter & Gamble CMO and the DTC founder are sometimes spoken of as if they are different species of marketing professional.

This was never a real distinction about how advertising works on human beings. It was a distinction about what tools were available to measure each medium. Digital channels generated click streams, so digital became "performance." Physical channels generated estimated impressions and recall surveys, so physical became "brand." The labels stuck even after the measurement gap closed. The brand-performance dichotomy is, almost entirely, a measurement artifact that hardened into industry doctrine.

The Kochava data is what the dichotomy looks like when both sides are measured honestly. The physical advertising drives the conversion. The digital advertising captures the receipt. Both are necessary. Neither one is "performance" or "brand" in any meaningful sense. They are different points in the same decision cycle, doing different jobs, and the entire conversation should be about sequencing and attribution share, not about which channel goes in which bucket.

This is going to be uncomfortable for the ad tech vendors whose entire business model depends on enforcing the dichotomy. It is going to be uncomfortable for the agencies whose org chart is built around it. It is going to be uncomfortable for the CMO who has spent five years cutting OOH to fund Search and now has to explain that decision to a board that has read the same study you are reading.

Good. The dichotomy was wrong. The data was always going to win.

I am going to be direct about something for the next four paragraphs, because pretending I have no commercial position in this conversation would be both transparent and stupid.

The reason this study exists is that the measurement infrastructure for OOH finally caught up to what was always happening in the physical world. Device-level viewshed mapping. Cross-channel attribution. Synthetic-control methodology applied to physical exposure data at scale. Extended attribution windows running on dedicated identity graphs. None of this existed ten years ago. Most of it did not exist five years ago. The reason the Kochava study can make the claims it makes is that the underlying data plumbing has finally been built, by a lot of different companies, doing a lot of different work, mostly in parallel and mostly without coordination, over roughly the last decade.

AdQuick has been building the buy-side half of this infrastructure since 2017. The marketplace that makes OOH inventory transactable at digital speed. The APIs with Outfront, Lamar, Clear Channel Outdoor, Global, and others that let media planners pull live inventory directly into a programmatic workflow. Our various data partnerships bring third-party audience composition into the planning surface. The Circle Graphics production API that handles creative delivery to physical fabrication. The Trust Medium content series that documented how badly OOH measurement was broken before any of the modern fixes existed. Kochava has been doing the measurement-side version of the same work on a completely separate track. The two efforts did not coordinate. They did not need to. They are different pieces of the same decade-long thaw, and the Kochava study is one of the first big artifacts the thaw has produced. It will not be the last.

For most of OOH's history, the medium's biggest problem was not that it did not work. It worked fine. The Kochava data is just the formal proof. The problem was that nobody could prove it worked in the language that modern media planners speak, and nobody could buy it in the workflow that modern media planners use. Both of those problems were solvable. They just required somebody to spend a decade solving them.

Now they are solved. And the budget-reallocation conversation gets a lot more interesting. Every CMO who has been pulling OOH dollars to fund Search now has to defend that decision against device-level lift data that says they have been pulling the wrong lever for ten years. There is a version of the next five years in advertising where this study is the inflection point, and every media plan written after May 2026 has more OOH in it than the one written before. There is also a version where the industry ignores the data, because the industry has ignored data before. Neither version is in my control, and I am not going to pretend it is.

What I will say is this. The data is now in the field. The plumbing is in the ground. The story is on the table. What the industry does with it is the industry's problem.

Return to the grill.

You are standing in the aisle on Saturday morning, holding a credit card and a small assortment of marketing residue you do not consciously remember acquiring. Your phone has a Maps query from this morning, a Search session from Wednesday night, and a coupon notification that fired sometime between Thursday and Friday. The Search ad will get the conversion in Google's dashboard. The coupon will get the conversion in the email tool. The Maps query will get the conversion in some analytics dashboard. Somewhere on Monday, three different marketers will look at three different dashboards and quietly congratulate themselves for a closed sale.

The grill is in your cart because of a billboard you cannot remember seeing. It worked anyway. It has been working this way for the entire history of the medium, in approximately the same fashion, for every category of consumer purchase, against every other channel that has ever competed for the credit. The Kochava data is the first cross-media study at this scale that bothered to measure the work instead of the receipt. The work is twice the size of what the digital industry has been crediting itself with. The receipt has always belonged to whoever was paying attention at the moment of transaction. The work has always belonged to whoever put the want in your head.

For twenty years, OOH did the work and digital took the meeting. The meetings, going forward, are about to get a lot more awkward.

The billboard never needed the data. The billboard has always known what it was doing. The marketer who buys the billboard always needed the data, because the marketer who buys the billboard has spent twenty years justifying the line item to a CMO who did not believe the medium worked. Now the data exists. Now the line item has a number on it. Now the medium that was always doing the work has paperwork to prove it.

The receipt was never the decision. The seven-day ghost was always there. The only thing that changed is somebody finally took the trouble to look for it.