The Attention Landscape · emerging-blended-established evidence

Attention Is Not Time-Spent: Why Impressions, Minutes, and Actual Attention Diverge

Last reviewed 2026-07-20. Written by Chandranshu Kumar, Founder, Raveneye Global. · 8 min read

An impression counts a placement that loaded. A minute of time-spent counts a survey respondent's estimate of how long they used a medium. Neither counts whether a human being actually looked. That gap matters because two ad placements bought at the identical CPM can carry very different amounts of measured attention, which means CPM alone cannot tell a buyer which channel is actually cheap. The theoretical case for treating attention, not information or exposure, as the scarce resource dates to 1971. The practical case for measuring it directly, rather than inferring it from a click or a self-reported minute, is newer and still maturing. Distinguishing between the two, without conflating a vendor's proprietary metric with settled science, is the discipline behind the Machine-Readiness Score.

The unit everyone reports is not the unit that decides anything

An impression is logged the moment an ad is served, whether or not a screen was ever in front of a person's eyes. A minute of time-spent, the figure behind most media-planning decks, is typically a survey estimate: a respondent recalling how long they used an app or watched a channel, not a measured behavior. Both are useful for accounting. Neither measures the thing that actually produces a sale, which is a human being noticing an ad long enough to register it.

This is not a pedantic distinction. It is the reason a media plan can hit every impression and time-spent target on paper while quietly buying almost no real attention, and the reason a channel that looks expensive on a raw CPM basis can be the more efficient buy once attention is accounted for.

The theoretical root: information is abundant, attention is not

The economic case for treating attention as the scarce resource, rather than the information or inventory being served, was made by Herbert Simon in 1971. His formulation is precise: a wealth of information creates a poverty of attention, and the practical problem becomes allocating that attention efficiently among an overabundance of sources competing for it.

Applied to media buying, the implication is uncomfortable for anyone still planning off impressions or minutes. If attention, not exposure, is the scarce and valuable resource, then a metric that counts exposure without confirming attention is measuring the wrong thing at the point where it matters most.

What a directly measured attention metric looks like

Lumen Research, an eye-tracking measurement firm operating the industry's largest ad-attention panel since 2015, built a metric it calls attentive seconds per thousand impressions, and a derived figure, attentive CPM, defined as CPM divided by attentive seconds per thousand. The logic is straightforward: instead of paying for a thousand served impressions, an attentive CPM tells a buyer what a thousand seconds of actual measured attention costs.

The finding that matters for a media planner is not any single number Lumen publishes, but the shape of the result: attention supply varies enormously by format and placement even when CPM is held constant. A cheap placement with low attentive seconds can be a worse buy than an expensive one that holds attention, and no amount of impression or time-spent reporting would reveal that on its own.

Why this is an emerging metric, not an established one, and that distinction matters

Lumen's attentive-seconds and attentive-CPM metrics are a single commercial vendor's proprietary methodology. The method is disclosed and the panel is increasingly adopted by platforms and out-of-home operators, which is a meaningfully stronger evidentiary position than an undisclosed black box. It has not, however, been independently peer-reviewed at the level of the older economics this article also draws on.

The defensible position is to treat attentive-seconds as a promising, methodologically disclosed practitioner tool, worth taking seriously as a directional signal, while not citing it as settled science on the level of a replicated academic finding. A buyer should ask any vendor using an attention metric the same question this article asks of Lumen: is the method published, and is the sample disclosed.

The diminishing-returns case for why attention quality, not volume, drives marginal spend

John Little's 1979 survey of aggregate advertising-response models established, decades before modern measurement tools existed, that the relationship between advertising exposure and sales response is typically concave or S-shaped, not linear. Marginal effectiveness changes as spend increases, and it is possible to spend well past the point where an additional exposure buys anything.

Google's 2017 Bayesian media-mix modeling framework, the methodological ancestor of tools such as Meridian and Meta's open-source Robyn, formalizes this into an estimable problem: a channel's response curve and carryover effect can be modeled so a marketer can compute the marginal return on the next dollar in each channel, rather than assuming all impressions are equally productive. Neither of these depends on Lumen's specific metric, but both provide the underlying economic reason a raw impression count was never going to be a reliable input to that calculation.

What this means when two channels report the same CPM

If a buyer is comparing two channels on CPM alone and neither channel's attention quality is measured, the comparison is incomplete by construction. A ten-dollar CPM that reliably holds attention for two seconds is a different economic proposition than a six-dollar CPM that is scrolled past in a fraction of a second, even though the second number looks cheaper on every report a platform hands over by default.

This is precisely the failure mode this evidence exists to correct: cheap inventory measured the wrong way looks like a bargain and can be the more expensive channel once the actual attention purchased is accounted for.

Reading the evidence honestly

The scarcity argument for attention over information is established economic theory, not a marketing claim. The observation that response to advertising is non-linear and subject to diminishing returns is established and has been formalized into working statistical models. What is emerging, and should be labeled as such, is the specific practice of measuring attention directly at scale through eye-tracking panels rather than inferring it from clicks, impressions, or self-reported time.

None of this argues that impressions and time-spent are worthless. They remain useful accounting units. It argues that they are proxies, and that a business making a channel decision on proxies alone is making that decision with less information than is now available.

The evidence

Key findings, with their sources

A wealth of information creates a poverty of attention, and attention, not information, must be efficiently allocated among the overabundance of sources competing for it.

established Herbert A. Simon, "Designing Organizations for an Information-Rich World," in Computers, Communications, and the Public Interest, Johns Hopkins Press, 1971.
Attentive seconds per thousand impressions (APM) and attentive CPM (aCPM, CPM divided by APM) show that two ad placements bought at an identical CPM can carry very different amounts of measured human attention.

emerging Lumen Research, "Attention Technology: What Advertisers Need to Know," lumen-research.com.
Sales response to advertising is typically concave or S-shaped rather than linear, meaning marginal effectiveness changes with spend and there is a point past which added exposure buys little.

established John D. C. Little, "Aggregate Advertising Models: The State of the Art," Operations Research, 27(4), 1979.
A channel's diminishing-returns response curve and carryover effect can be modeled so a marketer can estimate the marginal ROI of the next dollar in each channel, rather than treating impressions as uniformly productive.

established Y. Jin, Y. Wang, Y. Sun, D. Chan, J. Koehler, "Bayesian Methods for Media Mix Modeling with Carryover and Shape Effects," Google Inc., 2017.

Reference

Glossary

Impression: A logged instance of an ad being served, regardless of whether a person actually looked at it.
Attentive seconds: Lumen Research's eye-tracking-derived measure of how many seconds of actual visual attention a placement received per thousand impressions, distinct from an impression count.
Attentive CPM (aCPM): A placement's CPM divided by its attentive seconds per thousand impressions, intended as a cross-channel comparability metric for attention quality rather than raw exposure.
Diminishing returns (advertising): The economic finding that additional advertising exposure produces progressively smaller incremental effect, formalized in concave or S-shaped response curves.

Straight answers

Frequently asked questions

Is attentive CPM a proven, industry-standard metric?

No, and it should not be treated as one. It is a single commercial vendor's proprietary methodology, methodologically disclosed and increasingly adopted by platforms and media owners, but not independently peer-reviewed at the level of the academic advertising-response literature. Treat it as a promising practitioner signal, not settled science.

If impressions are not attention, why does the industry still report them?

Impressions remain a useful, simple accounting unit for what was served, and platform-reported impressions are consistent and auditable in a way attention estimates are not yet. The issue is not that impressions are worthless, it is that they are a proxy for delivery, not for whether a human noticed.

Does a higher attentive CPM always mean a worse buy?

Not necessarily on its own. Attentive CPM is one input into a channel decision, and it needs to be read alongside what that attention is worth for a specific business, audience, and purchase cycle, which is exactly the kind of per-domain question a Corpus-style read is built to answer rather than a single ratio.

How does this connect to the Machine-Readiness Score?

The Machine-Readiness Score is built on the same principle this article argues for: measuring the thing that actually determines whether a buyer finds and chooses a business, across search, AI answers, reputation and technical foundation, rather than relying on a proxy metric that looks reassuring but does not reflect real visibility.

Provenance

Sources

Herbert A. Simon, "Designing Organizations for an Information-Rich World," in Computers, Communications, and the Public Interest, ed. M. Greenberger, Johns Hopkins Press, 1971 (established)
Lumen Research, "Our Technology" and "Attention Technology: What Advertisers Need to Know," lumen-research.com (emerging, vendor-disclosed methodology)lumen-research.com
John D. C. Little, "Aggregate Advertising Models: The State of the Art," Operations Research, 27(4), pp. 629-667, 1979 (established)doi.org
Y. Jin, Y. Wang, Y. Sun, D. Chan, J. Koehler, "Bayesian Methods for Media Mix Modeling with Carryover and Shape Effects," Google Inc., 2017 (established)research.google

Every figure above is attributed to a real, dated source and tagged with its evidence tier. Where a claim could not be verified to a primary source, it is not stated as fact.