Share of Attention: How a Broadcast-Era Metric Is Being Reinvented for the Feed-and-Answer-Engine Era

The same question, asked for a century

Every generation of media measurement has, at bottom, tried to answer one question: of a fixed amount of attention a population has to give, what fraction is any one channel actually capturing? "Share of attention," now being discussed as a distinct, formal metric in advertising and media circles, is simply the newest name for that question, generalized beyond the broadcast channels it was first invented to measure.

Treating it as a brand-new invention misreads its lineage. The vocabulary, and the underlying accounting logic, both trace back to the earliest days of audience measurement, decades before anyone used the phrase attention economy.

Crossley's 'rating' and Nielsen's 'share': the direct ancestors

The word "rating" itself was coined by Archibald Crossley, who formed the Cooperative Analysis of Broadcasting in 1930, the first national radio ratings service, using telephone-recall methodology. A rating measured what fraction of a defined population was tuned to a given broadcast. "Share" followed as a closely related concept in the audience-measurement tradition that Nielsen formalized from the 1940s onward: the fraction of the currently-tuned-in audience captured by one station or network, as distinct from the fraction of the total population.

Both concepts share the same structural logic that "share of attention" now claims: take a fixed, measurable pool, everyone listening, or everyone with attention to give, and calculate what fraction one thing captured. What has changed since 1930 is not the logic. It is the pool being measured.

The pool used to be a channel. Now the pool is a person.

Broadcast-era rating and share measured a defined, closed medium: how many households were tuned to any radio broadcast, and of those, what share went to one station. The unit being divided was total broadcast tuning within that medium. Modern share of attention proposes to divide something much larger and harder to bound: a person's total finite attention across every competing surface, television, streaming, social feeds, search, and increasingly AI answer engines, not just within one medium.

That is a genuine expansion of scope, not a cosmetic rebrand. It requires knowing, or estimating, total attention across surfaces that were never designed to be measured together, which is precisely why the metric is still being formalized rather than already standardized.

The evidence the pool has stopped growing

The practical trigger for formalizing share of attention as a distinct metric is that total daily media time appears to have hit a ceiling. eMarketer projects US adults will spend 12 hours and 44 minutes a day with media in 2025, with total daily time growing by only about 2 minutes in 2025 and a projected 1 minute in 2026. eMarketer's own analysts read this as attention reaching a saturation point, not a data artifact of how the figure was compiled.

This figure should be read with its evidentiary tier attached. eMarketer's underlying statistical methodology for turning survey and panel inputs into a published hours-per-day number is largely proprietary and not fully disclosed in public materials, which places it in the emerging rather than established tier, useful as a directional signal, not as a precise, independently verifiable constant.

If the pool of total available attention has genuinely stopped expanding, then the only way to grow is to take share from something else, exactly the accounting logic Crossley and Nielsen built for a single broadcast medium, now being asked to operate across an entire, fragmented media environment.

Why the metric is contested, not settled

Share of attention is not yet an agreed industry standard. As of 2025 and 2026, the IAB and various agencies are actively proposing competing systems for attention measurement, and there is no single agreed definition or measurement methodology yet. The category is real and the industry conversation around it is active, but any specific "share of attention" figure a vendor publishes today should be understood as one proposed system's output, not a ratified industry benchmark the way, for example, a Nielsen rating eventually became.

This is worth stating plainly rather than picking one system and presenting it as settled. The underlying problem, allocate a finite resource, attention, across competing claims on it, is a century-old and well-understood accounting logic; the specific metric that will formalize it for the current media environment is still being built.

Generalizing from broadcast channels to every surface a buyer is on

The useful move is not to wait for the industry to finish standardizing the metric before acting on its logic. A business does not need a ratified IAB definition to ask the Crossley-and-Nielsen question about its own buyers: of the finite attention our buyers have to give, what fraction lands on us, across every surface they actually use, classic search, the local map pack, AI answers, and reputation, rather than any single channel in isolation?

That is the same generalization the share-of-attention conversation is reaching for at industry scale, applied at the scale of one business and its actual buyers. It requires the same discipline the original broadcast measurement pioneers insisted on: a defined population, a defined measurement window, and a clear account of what the resulting number can and cannot claim.

What the frontier gap actually is

No source in the current measurement field, not Nielsen, not Ofcom or BARB, not the Reuters Institute, not eMarketer, not GWI, not DataReportal, not the International Telecommunication Union, currently publishes a rigorous method for measuring attention specifically to AI answer engines and generative search. That is the genuine frontier gap this whole lineage points toward, and it is the gap any credible share-of-attention claim for the current media environment has to confront directly rather than borrow a number from an unrelated method.

A metric that claims certainty about AI-answer attention today, without disclosing that no established instrument yet measures it directly, is making a stronger claim than the evidence supports. The right posture, matching the century-long pattern this article traces, is to build a measured, disclosed method for the new surface rather than assert a number the industry has not yet agreed how to produce.