From Corpus to Machine-Readiness Score

The terrain has already moved

Start with the scale of it. DataReportal's Digital 2026 Global Overview counts 6.04 billion people online worldwide, a gain of 294 million in twelve months, and 5.66 billion active social media identities, 68.7 percent of the world's population. The average person now uses 6.75 different platforms every month. That single number, 6.75, is the clearest evidence that attention is not consolidating anywhere. It is distributing itself across a widening set of surfaces, and it is doing so while spending more raw time online than ever: over an hour a day on core social networks alone, 2.5 hours a day once video platforms are counted, and an average of 18 hours 36 minutes a week on social media, rising to 25 hours 45 minutes a week among women aged 16 to 24, per the same report.

This is the starting condition for any visibility question a business owner asks today. If a market's attention is genuinely spread across seven-plus platforms a month, a report on any single one of them, a search ranking, a follower count, an engagement rate, is a snapshot of one corner of a much larger terrain, not a picture of the whole thing.

Search is still the floor, but the floor is cracking at the edges

None of this means search has stopped mattering. DataReportal found that 80.3 percent of online adults use search engines at least monthly, that Google alone accounts for roughly 90 percent of search referral traffic, and that global search advertising spend was projected at 352 billion US dollars for 2025. Search remains the widest single surface in the terrain, and for most businesses it still carries more buying-stage traffic than any other channel.

But the erosion at the edges is real and worth naming directly. The Reuters Institute cited an analysis of referral traffic across more than 2,500 news sites showing that Google organic referrals to news sites fell 33 percent globally and 38 percent in the US between November 2024 and November 2025, alongside a 3-percentage-point decline in search as people's main news gateway among those under 35. That referral figure is cited data inside the Reuters Institute's report, drawn from a third-party analysis rather than the Institute's own survey, which is why it carries an emerging rather than established tier here. It should be read as a real and significant signal, not yet as a fully settled industry number.

Where the shift actually lands: social and video, unevenly

The audience leaving search and television isn't landing everywhere at once, it's concentrating on specific platforms at very different depths. Pew's breakdown of US adults shows Facebook at 38 percent regular news consumption, YouTube at 35 percent, Instagram and TikTok each at 20 percent, and X at 12 percent of the general population. But look inside each platform's own user base and the picture changes: 57 percent of X's users and 55 percent of TikTok's users say they regularly get news there, far above those platforms' share of all US adults.

The TikTok trend line is the sharper story. Among people who already use the platform, the share who regularly get news there climbed from 22 percent in 2020 to 55 percent in 2025, according to Pew. That is not new reach, a platform pulling in people who weren't there before, it is depth, the same audience spending more of its attention, and more of its trust, on one surface over five years. A visibility map that only counts new followers or new reach misses exactly this kind of shift.

TikTok's own audience didn't just grow. It converted: the share of TikTok users who regularly get news there climbed from 22 percent to 55 percent in five years, according to Pew Research Center. That is depth, not new reach.

The AI-answer layer: real, growing, and still unmeasured as a competitive market

This is the layer every business owner is now asking about, and the evidence here has to be handled with real care. The Reuters Institute found that 10 percent of people globally now use AI chatbots weekly for news, up from 7 percent a year earlier, with usage nearly double that, 16 percent, among people under 35. Separately, a peer-reviewed benchmark study (the GEO paper, evaluated at KDD 2024 on a purpose-built set of diverse queries) found that specific content-optimization techniques could increase a source's visibility inside generative-AI answers by up to 40 percent under lab conditions.

Both of those are genuine findings. Neither of them is a competitive visibility census. The Reuters Institute's figures measure how many people use chatbots, not which brands those chatbots cite or favor in a given market. The GEO paper measures an optimization technique's effect on a benchmark, not a real-world share of answer that a business could check itself against a competitor. As of this research pass, no independent, standardized measurement of who is actually winning inside those generative answers, market by market, brand by brand, was found to exist anywhere. That is not a gap in this study's research effort, it is the current state of the field, and it is precisely why a Machine-Readiness Score for this layer has to be built from primary capture rather than assembled from someone else's usage survey.

Usage of AI chatbots is measured. Trust in what they say is measured. Which brands they actually favor, market by market, is not measured by anyone yet. That is the gap a Machine-Readiness Score has to close firsthand.

Reach and trust are not the same number

The most useful single contrast in this evidence pack is the Reuters Institute's finding that only 20 percent of people globally trust the answers AI chatbots give them, compared with 37 percent overall trust in news. A layer can be growing in usage, 10 percent weekly and rising, while carrying roughly half the trust that news carries overall. That is not a reason to ignore the AI-answer layer, its growth trajectory is real, but it is a reason not to weight it the same as a channel where reach and trust move together.

The same discipline applies across the whole terrain, not just to AI answers. Television and news websites have each lost roughly two percentage points a year in reach since 2020 (13 points and 12 points respectively), a slower erosion than the AI-answer layer's rise, but it is erosion in an incumbent, trusted channel, not just growth in a new one. A visibility score that only counts exposure, and never asks how much the audience believes what it sees on that surface, will consistently overrate the newest, loudest channel and underrate the slower, steadier one.

The terrain has moved like this before

None of this is the first time attention has migrated faster than the metrics built to track it. In May 2023, Google reported that its Lens visual search tool was handling 12 billion visual searches a month, a fourfold increase in two years, alongside a Shopping Graph of more than 35 billion product listings with 1.8 billion of them refreshed every hour. That data predates the current AI Overviews and AI Mode products by roughly two years and describes an earlier shift, search moving beyond typed text into images and structured product data, but it makes the same point the 2026 data makes: attention finds new surfaces faster than any single existing report is built to count.

The underlying idea that attention itself is a scarce, allocable resource is not new either. It traces to Herbert A. Simon's observation, in the 1970s, that a wealth of information creates a corresponding poverty of attention, a framing that helped earn Simon the 1978 Nobel Memorial Prize in Economic Sciences for his work on bounded rationality and decision-making under real constraints. Attention economics as a field, and the instinct to treat visibility as something that must be actively measured and allocated rather than assumed, rests on that fifty-year-old foundation, not on anything specific to search or social media.

Why a single-channel number can't carry the answer anymore

Put the pieces together and the case for a cross-channel model, rather than any one platform's report, is straightforward: attention is spread across close to seven surfaces per person a month, it is migrating between them at measurable rates (30 percent now naming social as the main news gateway, up from 22 percent; search referrals to news sites down double digits in a year), and it carries different amounts of trust on different surfaces (20 percent versus 37 percent). No single number from any one platform can answer where a given business actually stands.

Adapting the classic logic of media-mix modeling (cross-channel allocation, time-decay effects, causal calibration through controlled comparison) from its native use in sales and revenue outcomes to an attention or visibility outcome is not something this research found published or externally validated anywhere. It is a methodological bridge Raveneye is building deliberately, grounded in decades of established media-mix and attention-economics thinking on each side of it, but it should be presented as an emerging synthesis, not cited as an industry-standard practice that already exists elsewhere.

What this means for your business right now

Three things follow directly from the evidence. First, don't mistake a strong number on one platform for visibility overall, the market you're trying to reach is very likely splitting its attention across six or seven surfaces a month, not sitting on the one you're watching. Second, weight trust alongside reach when you decide where to invest, a channel that shows you to more people but where fewer of them believe what they see is not automatically the better bet. Third, be skeptical of any claim of AI-answer-layer visibility that isn't backed by measurement taken directly in your own market, because no independent, third-party scorecard for that layer exists yet for anyone to borrow from, including us.

Search is still where four out of five online adults show up every month, and it should still anchor a visibility plan. But the fastest-growing part of the terrain, the AI-answer layer, is also the part where the least trustworthy shortcuts are being sold right now, precisely because almost nobody has done the primary work of measuring it directly.