Eighty-five percent of digital advertising impressions receive less than 2.5 seconds of attention. For most of the last decade, that number was read as an indictment — proof that the majority of digital spend evaporates on contact.
Then a 2025 study from VCCP Media and the attention platform Amplified put a finer point on it. Tracking more than 20,000 views across 72 digital video ads, Dr. Karen Nelson-Field found that 1.5 seconds of active attention is enough to encode a memory — provided the ad carries distinctive brand assets. Strip those assets out and the effect collapses. Same media buy. Same seconds. Same audience. Different outcome.
Read that again slowly. Attention was the control variable. What changed was whether anything was left behind.
This is the gap the industry keeps stepping over. The distinction between attention and memory is not academic hair-splitting. It is the difference between an input you rent by the second and an asset you own for years.
Attention measurement was a genuine correction. Viewability had become a fiction — Nelson-Field's work found that roughly 75% of inventory rated "viewable" receives no attention at all. An ad that technically rendered on a screen no human looked at was being counted, invoiced, and reported as delivered. Attention platforms fixed a measurement failure that the industry had spent years politely ignoring.
But they inherited its underlying flaw. Attention metrics still measure the moment of exposure. And exposure is a delivery receipt, not a result.
Nobody buys anything at the moment of exposure. Almost nobody. The person watching your fifteen-second pre-roll is not, in that instant, evaluating a purchase — they are waiting for the skip button. Whatever influence occurs happens later, in a different room, under different conditions, with your brand nowhere in sight. The ad is not the persuasion event. The ad is the deposit. The persuasion event is the withdrawal.
Which means the question that matters is not how long did they look. It is what is still there in ninety days, and what will pull it back out.
The Ehrenberg-Bass Institute has spent decades building the vocabulary for this. Their term is mental availability: the probability that a brand is thought of, or easily recognized, in a buying situation. It is built through category entry points — the specific needs, occasions, and cues that send someone into a category in the first place.
The evidence is not soft. In a 2024 meta-analysis across more than 100 brands, the research platform quantilope found an average correlation of 0.83 between mental market share and actual sales — a single survey metric accounting for roughly 70% of the variance in sales performance. That is a stronger relationship than most marketers will encounter anywhere else in their career.
The strategic consequence is uncomfortable: brands do not define category entry points. Buyers do. They exist independently of your positioning, your messaging framework, and your quarterly narrative. "The server went down again at 2 a.m." is a category entry point. "We need press before the funding announcement" is a category entry point. Your job is not to invent them. Your job is to be the name that surfaces when one fires.
Every influence principle has this same storage problem, and it is routinely overlooked. Authority is treated as something that operates in the moment — a credential displayed, a citation dropped. It doesn't. An authority cue only does work if it is retrieved at the instant the credibility question is asked, which is usually weeks after you presented it, in a conversation you are not part of. Same for social proof. Same for liking. Persuasion is not what happens while someone is paying attention to you. It is what happens when they are not.
In B2B this becomes arithmetic. Research by Professor John Dawes at Ehrenberg-Bass, popularized by the LinkedIn B2B Institute, holds that at any given moment only about 5% of potential buyers in a category are actively in market. The other 95% are not researching, not shortlisting, not reachable by intent data. They are simply working.
The number is derived from purchase cycles: with an average five-year B2B replacement window, roughly 20% of a market enters buying mode in a given year, and about 5% in any given quarter.
So the attention you buy this week is being cashed by a buyer who, this week, does not exist. The transaction is not exposure-to-conversion. It is exposure-to-storage-to-retrieval-to-conversion, and the middle two steps are invisible to every dashboard you own.
This is also why performance measurement looks the way it does. Last-click attribution can only see the 5%. It observes the moment of retrieval and assigns full credit to whatever cue happened to trigger it — the branded search, the retargeting ad, the final email. The system credits the withdrawal and cannot see the deposit. Then budget follows the credit, and the deposits stop.
If memory is the asset, the operative question becomes how efficiently a given second of attention converts into a durable trace. And here the VCCP and Amplified research is blunt.
The study ran what it called a "bad twin" test: video ads from eight brands — Cadbury, Domino's, easyJet, O2, Sage and others — placed in real social environments, once with their distinctive assets intact and once with those assets removed. Colors, characters, sonic branding, visual signatures.
Well-branded ads were 2.5 times more effective even in low-attention environments than poorly branded ones. One brand code delivered 3.5 times the attention-adjusted return of its unbranded version. The gap between the two conditions is what the researchers called a "distinctive asset tariff" — a self-imposed cost of up to 66 pence of every pound spent, or roughly £66 billion in lost value globally each year. Legacy brands fared worst, losing an average of 69 pence per pound against 59 for challengers.
The mechanism is not mysterious. Human memory is cue-dependent. Something has to trigger retrieval, and that something has to be linked to you specifically. A distinctive asset is a pre-built path back to the brand. Without one, the viewer still encodes something — they encode the category. You have paid to make the buying occasion more salient, and the brand that owns the cue will collect on it. Sometimes that brand is your competitor.
Which reframes the entire creative brief. The purpose of consistency is not aesthetic discipline or internal governance. It is retrieval engineering. Every time you refresh a look, rename a framework, or let a new agency reinterpret the system, you are demolishing paths that took years and millions to build — and you are doing it for reasons that will not appear on any scoreboard.
Les Binet and Peter Field established the pattern more than a decade ago across roughly 1,000 IPA case studies: activation produces a sharp, fast-decaying lift; brand building produces a slower effect that compounds. Their 60/40 guideline is the best-known artifact of that work, and the most frequently misapplied.
At the IPA Effectiveness Conference in 2025, Binet's argument was that the industry has optimized itself into efficiency, targeting, and short-term metrics at the expense of scale, reach, and durable effect.
It is worth being precise about why this keeps happening, because the usual explanation — marketers are short-sighted, CFOs are impatient — is too easy. Short-termism persists because memory is invisible on a dashboard until it converts. Activation shows up immediately in a system built to detect immediate things. Brand building shows up as a slow drift in base rates that no weekly report is designed to see. Given a measurement system that can only perceive one of the two, budget will migrate toward the visible one indefinitely, regardless of which one is actually producing the growth.
The market does not reward the better strategy. It rewards the more measurable one. That is a solvable problem, but only if you change what you measure.
Four replacements. None of them report weekly, which is precisely the point.
Mental market share by category entry point. Not aided awareness — aided awareness measures whether someone recognizes your name when handed it, which is not a situation that occurs in commercial life. Ask instead which buying situations retrieve you, and how that set is widening or narrowing over time.
Distinctive asset attribution. Remove the logo and the name. Show the color, the sound, the shape, the character. Can buyers still name you? If not, you are paying full price for exposure and banking a fraction of it.
Branded attention, not attention. Count only the seconds in which the brand was present and identifiable. A ten-second view where you appear for the last half-second is not ten seconds of brand attention. Your invoice disagrees.
Base-rate drift. Unprompted recall and organic branded demand, tracked quarter over quarter with no expectation of weekly movement. This is the slowest number in your stack and the only one that reflects the asset you are actually building.
Competitors can match your pricing within a quarter. They can rebuild your product in a year. They can hire your team.
What they cannot do is delete the association already sitting in a buyer's head — the one that fires unbidden the moment the problem appears, before any evaluation begins, before your category is consciously entered. That association was not bought. It was deposited, repeatedly, with discipline, by a firm that understood it was building storage rather than renting eyeballs.
Attention is rented by the second, and the price goes up every year. Memory is owned. Spend accordingly.
If your measurement stack cannot distinguish between attention and memory, it is optimizing against the thing that actually produces growth. Book a strategy consultation and we will show you what your current reporting is hiding.