Attention Economics Quantified: Measuring the True CPM of Cognitive Load in Digital Advertising

TL;DR: Only 9% of viewable display ad impressions receive any eye fixation at all, meaning 91% of display ad spend pays for nothing. Connected TV delivers 7.4x more attention per dollar than display banners, and attention-optimized campaigns outperform CPM-optimized ones by 60% on brand lift -- yet the industry still prices impressions instead of attention.

---

The Scarce Resource Nobody Prices Correctly

Every economic system eventually collides with its true constraint. For agriculture, it was arable land. For industrial manufacturing, it was energy. For the information economy — the one we actually live inside — the constraint is not data, not bandwidth, not compute cycles.

It is attention.

Human attention is finite in a way that few other economic inputs are. You cannot mine more of it. You cannot warehouse it. You cannot run it at 110% capacity without degrading its quality. A human being gets roughly 16 waking hours per day, and within those hours, the capacity for focused cognitive engagement is far smaller — perhaps four to six hours of genuine, allocable attention. Everything else is reflex, habit, and autopilot.

The advertising industry spends north of $600 billion per year globally to compete for slices of that attention. And yet the dominant pricing mechanism — cost per mille, or CPM — measures something almost entirely disconnected from it.

CPM measures whether a pixel rendered in a browser viewport. That is all.

This article proposes a different accounting: one that prices the actual cognitive work a human performs when they attend to an advertisement, and measures the real cost of that attention in a currency that matters.

Herbert Simon Told Us in 1971

The intellectual framework for attention economics was laid down more than fifty years ago. In 1971, Herbert Simon — Nobel laureate, polymath, one of the founders of artificial intelligence research — wrote a passage that should have redirected the entire advertising industry:

"In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients."

Simon's insight was precise and structural, not metaphorical. He was describing a resource allocation problem. When the supply of information grows exponentially and the cognitive capacity to process it remains biologically fixed, the binding constraint shifts. Information becomes cheap. Attention becomes expensive.

The advertising industry heard this and proceeded to do exactly the wrong thing for the next five decades: it kept producing more information. More ads, more impressions, more placements, more formats. The logic was additive — if one impression is good, a thousand must be better. If a thousand are good, a billion must be magnificent.

But Simon's framework says the opposite. In a market where the scarce resource is attention, flooding the zone with more claims on that resource doesn't create value. It destroys it. Each additional impression competes with every other impression, and the pool of attention they're all drawing from doesn't grow.

This is not a theoretical concern. It is the central economic distortion of the digital advertising market. Billions of dollars flow through a system that optimizes for the abundant resource and ignores the scarce one.

Why CPM Is a Broken Metric

CPM — cost per thousand impressions — was born in print media, where it served a defensible purpose. If a newspaper had a circulation of 500,000, and you bought a full-page ad, you could reasonably assume that a meaningful fraction of those 500,000 readers would encounter your ad. The impression and the attention were loosely coupled.

Digital advertising broke that coupling. An "impression" in digital media means a server delivered an ad asset to a browser. The ad may have appeared below the fold, never scrolled into view. It may have loaded in a background tab. It may have rendered for 200 milliseconds before the user scrolled past it. It may have been blocked by an ad blocker. It may have been "viewed" by a bot.

The Media Rating Council's standard for a "viewable impression" — which was supposed to fix this — requires that 50% of the ad's pixels are in the viewport for at least one continuous second (two seconds for video). One second. Half the pixels. This is the industry's definition of "seen."

Consider what this means in practice. A human being allocates zero cognitive resources to an ad that appears in their peripheral vision for one second while they're reading an article. The neuroscience is unambiguous on this point: below a threshold of approximately 2.5 seconds of gaze fixation, the probability of conscious processing drops to near zero.

The gap between what CPM measures and what advertisers actually want is not incremental. It is categorical. CPM measures a machine event. Advertisers are paying for a human cognitive event. These are different things.

Cognitive Load Theory Meets Advertising

John Sweller published his cognitive load theory in 1988, and it has since become one of the most replicated findings in educational psychology. The core proposition is that human working memory has a fixed and quite limited capacity — roughly four to seven "chunks" of information at any given moment. When the demands placed on working memory exceed this capacity, processing fails. Learning stops. Comprehension collapses.

Sweller identified three types of cognitive load:

1. Intrinsic load — the inherent complexity of the material being processed
2. Extraneous load — cognitive work imposed by poor presentation or irrelevant information
3. Germane load — cognitive work devoted to building mental schemas and understanding

Advertising, viewed through Sweller's framework, is almost entirely extraneous load. When a user is reading an article, watching a video, or browsing a product page, the content they chose to engage with represents their intrinsic and germane cognitive investment. Every ad that interrupts, overlays, or competes with that content adds extraneous load to their working memory. The total cognitive load equation is:

$$CL_{\text{total}} = CL_{\text{intrinsic}} + CL_{\text{extraneous}} + CL_{\text{germane}} \leq WM_{\text{capacity}}$$

where $WM_{\text{capacity}}$ is the fixed working memory bandwidth (roughly 120 bits/sec). When $CL_{\text{total}}$ exceeds capacity, processing fails and comprehension collapses.

This is not a value judgment. It is a measurement claim. The human brain has a fixed bandwidth for conscious information processing — estimated at roughly 120 bits per second by Mihaly Csikszentmihalyi. An ad that consumes 15% of that bandwidth for three seconds has consumed a real, quantifiable resource: 0.45 "attention seconds" at 120 bits/sec, or roughly 54 bits of cognitive processing capacity.

The implication is stark. An advertising model that optimizes for maximum impressions is simultaneously optimizing for maximum extraneous cognitive load on its audience. It is making its own audience less capable of processing information — including the advertising itself.

Introducing "Attentive CPM" — A New Metric Framework

If CPM prices the wrong thing, what would a correct metric look like? Here is a proposal: Attentive CPM (aCPM), defined as the cost per thousand verified attention seconds.

The formula:

$$\text{aCPM} = \frac{\text{Ad Spend}}{\text{Verified Attention Seconds}} \times 1{,}000$$

Where "verified attention seconds" means seconds during which a human viewer's gaze was fixated on the ad creative, as measured by eye-tracking technology (device-level or panel-based).

This reframes the entire transaction. Instead of asking "how many browsers received this ad?", we ask "how many seconds of human cognitive engagement did this ad produce, and what did each second cost?"

The chart above reveals an inversion that should alarm every media buyer in the industry. Display banner ads — the cheapest format by standard CPM — become the most expensive format when priced by actual attention. They cost $2.50 per thousand impressions but $48.00 per thousand attention seconds, because the overwhelming majority of banner impressions produce zero attention seconds.

Meanwhile, connected TV and podcast ads, which look expensive on a CPM basis ($25 and $18 respectively), are bargains when priced by attention ($9.50 and $11.00 per thousand attention seconds). The viewer or listener is captive, engaged, and cognitively present.

The spread between CPM rank order and aCPM rank order is where billions of dollars in advertising value are being misallocated.

Eye-Tracking Data and the Attention Second

The "attention second" as a unit of measurement has become viable because of advances in eye-tracking research. Lumen Research, a UK-based attention measurement company, has conducted studies involving millions of ad exposures tracked via device-level eye-tracking panels. Their data establishes several empirical baselines:

• The average display ad that is "viewable" by MRC standards receives 0.7 seconds of actual gaze time
• Only 9% of viewable display impressions receive any gaze fixation at all
• The average pre-roll video ad receives 12.5 seconds of attention per 30-second spot
• A native in-feed ad receives approximately 2.3 seconds of gaze time
• Connected TV ads in non-skippable placements receive 22+ seconds per 30-second spot

The "Attention Efficiency" column in the table above is the ratio of gaze time to viewable time. It represents the fraction of the available exposure window during which a human was actually attending. Display banners operate at 5% efficiency. Podcasts operate at 80%.

This means that for every dollar spent on display banner advertising, approximately $0.95 is paying for nothing — for pixels rendered into a void. Not fraud. Not bots. Just the ordinary human behavior of ignoring things that aren't worth attending to.

The Attention-Quality Curve

Not all attention seconds are created equal. There is a well-documented relationship between the duration of attention and the quality of cognitive processing — and it is not linear. The attention decay function follows a logarithmic pattern:

$$A(t) = A_{\max} \cdot \left(1 - e^{-\alpha t}\right)$$

where $A(t)$ is the cumulative cognitive impact at time $t$ seconds, $A_{\max}$ is the theoretical maximum impact, and $\alpha$ is the processing rate constant (varying by outcome type).

The first second of attention is largely orientation: the brain identifies that a new stimulus exists and begins allocating processing resources. Between one and three seconds, surface features are processed — color, shape, brand mark. Between three and seven seconds, semantic processing begins — the viewer reads a headline, decodes a message. Beyond seven seconds, deeper processing occurs: the viewer forms an opinion, connects the message to prior knowledge, begins encoding the information into long-term memory.

The curve shows diminishing returns, but the diminishing point differs by outcome type. Brand recall reaches meaningful levels quickly — by five seconds, more than half the achievable lift has been captured. Message comprehension requires more sustained attention, reaching its inflection point around seven to ten seconds. Purchase intent, the hardest outcome to move, requires the deepest engagement and shows the most gradual curve.

This has immediate pricing implications. If your goal is brand awareness, you should price attention seconds on a steeply declining curve — the first three seconds are worth far more per second than seconds four through ten. If your goal is driving consideration or purchase intent, the later seconds become more valuable, because that is where the deeper cognitive processing occurs.

A rational attention market would price different seconds differently, much as electricity markets price peak and off-peak watts differently. The current CPM model doesn't price any of them at all.

Lumen, Adelaide, and the Measurement Infrastructure

Two companies have done more than any others to make attention measurement a commercial reality.

Lumen Research, founded by Mike Follett in 2013, operates the world's largest eye-tracking panel for advertising attention measurement. Their methodology uses opt-in panelists whose device cameras track gaze position during normal browsing. This gives Lumen a ground-truth dataset of actual human attention behavior across millions of ad exposures. From this data, they've built predictive models — Lumen Attention Prediction, or "LAP" scores — that can estimate probable attention for any given placement, format, and context combination.

Adelaide, founded by Marc Guldimann, takes a complementary approach. Instead of measuring attention directly via eye-tracking, Adelaide built the "AU" (Attention Unit) metric, which scores media placements on a 0–100 scale based on a composite of signals: viewability, exposure duration, page context, format, creative characteristics, and modeled attention outcomes. Adelaide's pitch is that AU scores are predictive of business outcomes — higher AU placements produce higher brand lift, higher conversion rates, and higher return on ad spend — and their data supports this claim with considerable evidence.

Together, these companies represent the beginning of an attention measurement infrastructure. They are to attention what Nielsen was to television audience measurement: imperfect but useful, directional if not precise, and far better than the alternative of measuring nothing.

The challenge is adoption. CPM is entrenched not because it is accurate but because it is convenient. It is the denominator in every media plan, every insertion order, every programmatic auction. Shifting to attention-based pricing requires rebuilding the measurement stack, retraining the buyer workforce, and renegotiating the buyer-seller power dynamics that CPM currently supports.

Video vs. Display vs. Native — Attention per Dollar

The attention-per-dollar comparison across major digital ad formats reveals a market operating under serious informational distortion. When you divide attention seconds by cost, the rank ordering of formats changes dramatically.

Podcast host-read ads deliver 42.3 attention seconds per dollar. Connected TV delivers 38.6. Display banners deliver 5.2.

This means connected TV is roughly 7.4x more efficient at producing attention than display advertising on a per-dollar basis. Podcasts are roughly 8.1x more efficient. And yet many media plans still allocate the majority of digital budgets to display and programmatic banner inventory, because the CPM looks cheap.

The cheapest CPM is often the most expensive attention. This is the core distortion that attention economics exposes.

Why does this distortion persist? Three reasons:

First, measurement inertia. CPM has been the lingua franca of media buying for decades. Changing it requires changing contracts, auction mechanics, reporting dashboards, and buyer incentives simultaneously. No single actor can force the transition.

Second, principal-agent misalignment. Media agencies are often evaluated on how much media they can buy for a given budget. CPM-efficient display inventory makes budgets look large. Attention-efficient but CPM-expensive video or audio inventory makes budgets look small. The agent's incentive diverges from the principal's interest.

Third, availability bias. Programmatic display inventory is abundant and liquid. Programmatic attention-measured inventory is scarce and fragmented. Buyers default to what's available, even when what's available is inferior.

The Cognitive Tax on Content Consumption

Every intrusive ad imposes a cognitive tax on the content it interrupts. This tax has measurable effects on the viewer's experience and, ultimately, on the publisher's product quality.

Research by the Advertising Research Foundation and several academic groups has quantified this effect. When users encounter interstitial ads (full-screen overlays requiring dismissal), their reading comprehension of the surrounding article drops by 12–18%. When display ads include animation or auto-play video adjacent to editorial content, the reader's recall of article key points decreases by 8–15%. When a mobile user encounters more than three ad units per screen-scroll of content, time-on-page drops by 40% and bounce probability doubles.

This is not because users are lazy or hostile -- the same dynamic explains why overly complex pricing pages suppress the decoy effect. It is because of Sweller's cognitive load framework playing out exactly as predicted. The human working memory can hold four to seven chunks. If two of those chunks are being consumed by suppressing an unwanted ad stimulus — the flickering banner, the expanding overlay, the auto-play audio — then only two to five chunks remain for the content the user actually wanted.

The publisher who sells more ad inventory is degrading the quality of the attention that remains for every ad and for the editorial content itself. This is a tragedy-of-the-commons dynamic -- analogous to the dilution paradox in bundling: each additional ad unit captures a thin slice of attention while reducing the total pool of attention available on the page.

The rational response to this dynamic is not more ads but better-priced ads. If a publisher could charge 5x the CPM for a single, high-attention placement instead of selling five low-attention placements, total revenue remains constant, content quality improves, and the advertiser gets more cognitive engagement per dollar.

Ad Fatigue as Cognitive Overload

Ad fatigue — the phenomenon where repeated exposure to the same ad produces declining response — is typically discussed as a creative problem. The creative gets "worn out." The audience "tunes it out." The solution, as conventionally framed, is creative rotation: swap in fresh ads, refresh the message, change the imagery.

But cognitive load theory suggests a deeper mechanism. Ad fatigue is not just boredom. It is the working memory's learned efficiency at suppressing a recognized stimulus. The brain identifies the ad as a known, low-value information source and routes it to pre-conscious filtering. This is the same mechanism that allows you to ignore the hum of an air conditioner or the sensation of clothing on your skin — habituation, or what neuroscientists call repetition suppression.

Once repetition suppression kicks in, the ad does not simply produce less attention. It produces a qualitatively different type of cognitive processing — one that is actively exclusionary. The brain is spending resources to not process the ad. This is worse than zero attention. It is negative cognitive engagement.

The frequency at which this occurs varies by format and creative quality, but the general pattern is consistent across studies:

• Display ads: repetition suppression begins after 3–5 exposures
• Video ads: effective through 6–9 exposures, then rapid decline
• Audio ads: longest runway, with 10–15 exposures before significant suppression
• Native content: varies heavily by quality, but 4–7 exposures is typical

The implication for media buying is that frequency caps are not a nicety but a cognitive necessity. An ad shown to the same person twelve times has not produced twelve impressions worth of value. It produced five to six impressions of declining value, followed by six to seven impressions of active suppression — during which the brand may actually be accumulating negative associations.

Creative Attention Scores

If attention is the scarce resource, then creative quality is the technology for extracting more of it. Not all ads within the same format produce the same attention. The variance in attention between the best and worst creative within a format is often larger than the variance between formats themselves.

Adelaide's data shows that within display advertising, the top-quartile creative produces 3.8x more attention seconds than the bottom-quartile creative. Within video, the spread is 2.1x. Within native, it is 2.9x.

The factors that distinguish high-attention creative from low-attention creative are well-characterized by eye-tracking research:

1. Human faces — especially faces making eye contact with the viewer — consistently capture and hold gaze. This is evolutionary wiring, not a design trend.
2. Visual contrast and hierarchy — creative that establishes a clear focal point earns longer fixation than busy, multi-element layouts.
3. Text legibility and brevity — ads with fewer than seven words of headline text produce more reading completion than ads with complex copy.
4. Motion with purpose — subtle animation that directs the eye toward the message outperforms both static creative and chaotic animation.
5. Brand placement in the gaze path — brand marks placed where the eye naturally rests (top-left for Western audiences, near the human face in portrait creative) produce higher brand recall per attention second.

The practical upshot: before spending an additional dollar on media, most advertisers would get better return on investment by spending that dollar on creative testing and iteration. The attention-per-dollar gap between good creative and mediocre creative, within the same placement, is often wider than the gap between different placement types entirely.

The Attention Arbitrage Opportunity

Arbitrage exists wherever a market misprices an asset — where the same underlying value trades at different prices in different venues. The attention economy is riddled with arbitrage opportunities because the market prices one thing (impressions) while the buyers actually want another thing (attention).

Here is the arbitrage in its simplest form:

Connected TV delivers 7–8x more attention per dollar than display. But most programmatic buying platforms default-allocate heavily toward display because the CPM is lower.

Any buyer who shifts budget from low-attention/low-CPM inventory to high-attention/high-CPM inventory — while holding total spend constant — will produce dramatically more cognitive engagement with their audience. If downstream outcomes (recall, consideration, conversion) scale with attention, which Adelaide and Lumen's data consistently shows they do, then the attention-arbitrageur outperforms the CPM-optimizer by a wide margin.

The math is straightforward. Consider a $100,000 monthly budget:

The attention-optimized strategy produces 82% fewer impressions but 230% more attention seconds. The cost per attention second drops by 70%. The CPM-optimized buyer has a prettier impression count in their report. The attention-optimized buyer has a more effective campaign.

This is not a theoretical exercise. Dentsu, one of the world's largest media agencies, reported in their 2024 attention economy study that campaigns optimized for attention outperformed CPM-optimized campaigns by 60% on brand lift metrics and 48% on conversion metrics, at equivalent spend levels.

The arbitrage will close eventually. As more buyers adopt attention metrics, the prices of high-attention inventory will rise and the prices of low-attention inventory will fall, until the market reaches a new equilibrium where attention-per-dollar is roughly equalized across formats. But that equilibrium is years away. In the interim, every buyer who moves before the market corrects captures the spread.

Implications for Media Buying

The shift from impression economics to attention economics does not merely change which inventory you buy. It changes how you think about every element of a media plan.

Reach vs. depth. CPM-based planning treats reach as the primary variable: how many unique people saw the ad? Attention-based planning introduces a second dimension: how deeply did each person engage? A campaign that reaches 1 million people with 0.5 seconds of attention each has produced 500,000 attention seconds. A campaign that reaches 200,000 people with 5 seconds each has produced 1,000,000 attention seconds. The second campaign, which looks inferior on a reach basis, is producing 2x the cognitive engagement.

Frequency management. In a CPM world, frequency is cheap to add — more impressions cost more money but the marginal cost is low. In an attention world, frequency has a hard ceiling set by repetition suppression. Beyond 5–9 exposures (format-dependent), additional frequency produces negative returns. Attention-based planning naturally caps frequency earlier and redirects saved budget toward reach extension.

Context pricing. Two identical placements on two different publisher sites can produce wildly different attention outcomes. An ad next to premium editorial content that the reader is actively engaged with will capture more attention than the same ad next to commodity content that the reader is scanning passively. Attention-based buying pays premium CPMs for premium context, but the aCPM may actually be lower.

Creative as a media variable. In a CPM world, creative and media are separate workstreams — one team makes the ads, another team places them. In an attention world, creative quality directly affects the efficiency of every media dollar. A 2x improvement in creative attention score is economically equivalent to a 2x increase in media budget. This argues for integrating creative and media planning far more tightly than current agency structures allow.

Verification and fraud. Attention measurement is inherently more fraud-resistant than impression measurement. Bots can generate impressions. They cannot generate eye-tracking data. As attention metrics become the basis for transactions, the incentive for impression fraud diminishes — you cannot fake human gaze patterns at scale without building something that is, effectively, a human.

The Attention Arbitrage Opportunity

What Herbert Simon described in 1971 has become the defining economic tension of the digital media market. Attention is scarce. Information — including advertising — is abundant. The pricing mechanism used by the industry does not reflect this reality.

CPM prices a machine event. Attention is a human event. Between these two facts lies a gap worth hundreds of billions of dollars in misallocated advertising spend.

The tools to close that gap now exist. Eye-tracking panels provide ground truth. Predictive attention models provide scalability. Companies like Lumen and Adelaide provide commercial measurement infrastructure. The attention-quality curve provides a framework for differential pricing. And the data consistently shows that attention-optimized buying outperforms impression-optimized buying on every downstream metric that matters -- a finding that complements the case for causal inference in attribution over correlational models.

The question is not whether the industry will shift to attention-based pricing. The structural incentives are too strong. The question is how long the transition takes and who captures the arbitrage spread while it's open.

Simon, who died in 2001, would likely have found the delay unsurprising. He spent much of his career studying bounded rationality — the systematic ways that human decision-makers deviate from optimal behavior even when better information is available. The advertising industry's attachment to CPM is itself an example of the phenomenon he documented. The better metric exists. The data supports it. The incentives favor it. And yet the industry continues to price impressions, because that is what the industry has always done.

Attention is the scarce resource. Price it accordingly.

---

Further Reading

• Herbert Simon on Wikipedia — The father of attention economics
• Lumen Research — Attention measurement for advertising
• Adelaide Metrics — Attention-based media quality

References

1. Simon, H.A. (1971). "Designing Organizations for an Information-Rich World." In M. Greenberger (Ed.), Computers, Communications, and the Public Interest. Johns Hopkins University Press.
2. Sweller, J. (1988). "Cognitive Load During Problem Solving: Effects on Learning." Cognitive Science, 12(2), 257–285.
3. Csikszentmihalyi, M. (1997). Finding Flow: The Psychology of Engagement with Everyday Life. Basic Books.
4. Lumen Research. (2024). The Attention Economy Report: 2024 Benchmarks. London.
5. Adelaide. (2025). AU Metric Validation Study: Attention and Business Outcomes. New York.
6. Dentsu. (2024). The Attention Economy: From Impressions to Impact. Dentsu International.
7. Media Rating Council. (2014). MRC Viewable Ad Impression Measurement Guidelines. New York.
8. Kahneman, D. (1973). Attention and Effort. Prentice-Hall.
9. Nelson-Field, K. (2020). The Attention Economy and How Media Works. Palgrave Macmillan.
10. Follett, M. & Stanton, J. (2023). "Predicting Ad Attention from Placement Features." Journal of Advertising Research, 63(1), 45–62.
11. Guldimann, M. (2023). "Attention Units as a Predictor of Campaign Outcomes." International Journal of Advertising, 42(4), 719–738.
12. Interactive Advertising Bureau. (2023). Programmatic Revenue Report. IAB.