The Macro Shift
The AI-Answer Adoption Curve
Buyers picked up ChatGPT, Gemini, and Copilot faster than they picked up the PC or the internet, but the data shows a wide, still-unresolved gap between how many people open the tool and how many actually let it decide.
Part of The Macro Shift in the Insights library.
Abstract
By every measure available, buyers in the US and worldwide adopted AI answer engines faster than they adopted the personal computer or the internet itself, with global adoption reaching an estimated 53% of the population within roughly three years of ChatGPT's November 2022 launch. But the data also shows real disagreement on the specifics: reputable US surveys differ by more than twenty percentage points depending on how the question is asked, and the distance between how many people use an AI answer and how many people trust or click through on it is wide and only partly measured. The commercial effect is real where anyone has actually measured it (AI-referred traffic to US retail sites grew 693% year over year during the 2025 holiday season with a real conversion lift), but no independently audited source yet tells a given business how much of its own market's buying decisions AI answers are actually swaying.
The surface is moving fast by any historical standard, and it is moving on more than one track at once. In the US, Pew Research Center's February 2026 survey put ChatGPT usage at 44% of adults, up from 18% just three years earlier, with 49% now using some AI chatbot and 24% using one daily. Globally, Stanford's 2026 AI Index Report estimated generative AI had reached 53% of the world's population within about three years of ChatGPT's launch, a pace the earlier NBER study by Bick, Blandin, and Deming had already flagged as faster than either the PC or the internet diffused. DataReportal puts the active global user count at roughly 2.42 billion people, against 6.12 billion internet users worldwide, meaning generative AI use is catching up to internet use itself rather than trailing it by a generation the way earlier technologies did. And the platform mix underneath that growth is not settling into a single winner: vendor trackers built on web-visit data show ChatGPT's share of AI-chatbot traffic eroding through 2026 as Google Gemini's share climbs, even as ChatGPT remains the platform US adults name first by a wide margin.
What the growth in usage does not automatically produce is trust, and the two need to be measured separately because they move at different speeds. The Reuters Institute's 2026 Digital News Report found weekly AI-chatbot use for news rose to 10% of respondents globally, yet trust in what those chatbots said stood at just 20%, well below the 37% trust placed in news overall. On the search side, a pre-registered field experiment found that when Google's AI Overviews appeared on a query, organic clicks fell 38% and the share of searches ending with no click at all rose from 54% to 72%, without any measurable gain in how satisfied people said they were with the answer. Put together, these findings describe buyers who sample an AI answer readily, often stop there without clicking through to verify it, and still hold real reservations about relying on it. That is a meaningfully different posture than either full replacement of traditional search or a passing curiosity, and it means visibility now has to work at the moment the answer is generated, not only at the moment a link is clicked.
For a business trying to understand where its own buyers' attention actually sits, the picture is one of redistribution rather than collapse. Gartner's February 2024 forecast that AI chatbots would cut traditional search volume 25% by 2026 has not played out as stated: Google still holds over 90% of the search market in 2026. What has genuinely shifted is the shape of engagement within that market: SparkToro's analysis of Similarweb clickstream data found 68.01% of Google searches now end without a click, up from 60.45% two years earlier, as AI Overviews grew to appear on more than a fifth of all searches. Meanwhile Adobe Analytics, tracking over a trillion visits to US retail sites, measured AI-referred traffic growing 693% year over year over the 2025 holiday season, with those visitors converting 31% more often than other traffic. Attention has not vanished from search; it has split across a widening set of surfaces, some of which never generate a click at all, and any map of where a market's attention actually sits now has to account for the answer itself as a destination, not just the pages behind it.
The data, in one read
Three Surveys, Three Different Answers
Ask how many Americans use AI chatbots and you will get three defensible answers that differ by twenty points, and every one of them is correct for what it measures. Pew Research Center's February 2026 survey of 5,119 US adults, run on its long-standing American Trends Panel, found 44% of US adults report using ChatGPT, up from 34% in 2025 and 18% in 2023, and that 49% now use some kind of AI chatbot, up from 33% in 2024. Stanford's Human-Centered AI Institute, using its own global adoption methodology in the 2026 AI Index Report, places US adoption at 28.3%, which ranks the country 24th worldwide by that measure. A National Bureau of Economic Research study by Bick, Blandin, and Deming, the first nationally representative US survey of generative AI use, found 39 to 40% of adults aged 18 to 64 had used generative AI as of August 2024.
None of these numbers is wrong. They ask different questions of different populations at different moments with different instruments: Pew asks specifically about named chatbot products on a probability panel; Stanford's figure comes from a global methodology built to compare countries against each other, not to be the definitive US household number; NBER's figure is nearly two years older and narrower in age range. The practice this study follows is to name the source and the definition every time a percentage appears, rather than repeat a single number as though the question had one settled answer.
Three reputable surveys, three different numbers, twenty points apart. The gap is the starting point, not a flaw to explain away.
Faster Than the PC, Faster Than the Internet
The NBER study's headline conclusion, drawn from the first nationally representative US survey of its kind, was that workplace adoption of generative AI ran as fast as adoption of the personal computer did in its own era, and that overall adoption ran faster than either the PC or the internet. That comparison was made using data through August 2024, less than two years after ChatGPT's public launch.
Stanford's 2026 AI Index extends the picture globally: it estimates generative AI reached 53% of the world's population within roughly three years of that launch. By the NBER researchers' own comparison, both the PC and the internet needed considerably longer to reach comparable shares of their eras' potential users. The speed is genuine and it is corroborated across two independent research efforts using different methods.
What speed of reach does not tell you is depth of reliance. A population can sample a new tool quickly and still be years away from trusting it with a real decision, and the sections that follow show exactly that gap opening up in the data on trust, clicks, and purchase behavior.
The Leaderboard Behind ChatGPT
ChatGPT remains the name US adults reach for first, but it is no longer the only name that matters, and the mix beneath it is genuinely reshuffling. Among US AI-chatbot users, Pew found Gemini at 24%, Copilot at 17%, Meta AI at 14%, Grok at 8%, Claude at 6%, and Character.ai at 3%. Adoption also skews sharply by age: adults under 50 are twice as likely as those 50 and older to use ChatGPT, 57% versus 28%.
Vendor trackers built on Similarweb web-visit data tell a consistent directional story even though they disagree with each other on the exact figures: ChatGPT's share of global AI-chatbot traffic is estimated to have eroded through 2026, from a reported 85%-plus share in early 2025 down to somewhere in the mid-50s to high-60s percent by mid-year, as Google Gemini's share climbed into the high twenties and Perplexity held a low single-digit share despite fast growth from a small base. Different trackers disagree by double-digit percentage points on the exact current number, so this should be read as a direction rather than a settled figure.
The practical implication is that reaching "the AI answer" is no longer a single-platform problem. It is a moving mix that skews young, favors ChatGPT today but not by the margin it did a year ago, and is being actively contested by Google's own default distribution advantage through Gemini.
From Sampling to Spending: What Holiday Shopping Data Shows
The clearest evidence of a real commercial effect, not just a survey response, comes from Adobe Analytics, which tracked over a trillion visits to US retail sites. It found AI-referred traffic rose approximately 693% year over year during the November-December 2025 holiday season, 769% in November and 673% in December, and that shoppers arriving via generative AI assistants converted 31% more often than other traffic overall, roughly double the conversion advantage Adobe had measured the year before.
Salesforce separately reported that AI and AI agents were associated with about 20% of online retail sales and roughly $262 billion in revenue over the same holiday season. That figure could not be independently reconfirmed in this research pass and rests on Salesforce's own proprietary attribution methodology, so it is treated here as a company-reported estimate rather than an established fact, and cited accordingly.
The summary is that a real, large, measured commercial effect exists where anyone has actually gone looking for it, but the measurement is still thin and largely proprietary. Nobody, including the companies publishing these numbers, has produced an independently audited figure for what share of any given market's purchase decisions AI answers are actually swaying.
The clearest number in this study, Adobe's 693 percent, describes a rate of change. It does not describe a share of the market. Both matter, and they are not the same question.
The Zero-Click Search Was Already Happening
SparkToro's analysis of Similarweb clickstream data found that 68.01% of Google searches in early 2026 ended without any click at all, up from 60.45% two years earlier, a rise of more than seven and a half points, as AI Overviews grew to appear on more than a fifth of all Google searches. SparkToro itself notes that its underlying clickstream data provider has changed more than once since 2019, so exact year-over-year comparisons carry some methodological uncertainty, though the direction is consistent with other sources.
A pre-registered field experiment with 1,065 US participants, fielded in January and February 2026 and registered with the AEA RCT Registry, found that when Google's AI Overviews appeared on a query, organic clicks fell 38% and the zero-click rate on those specific queries rose from 54% to 72%, without a measurable improvement in participants' reported satisfaction or sense of information quality. The paper is a working paper posted to SSRN and has not yet completed peer review.
Taken together, these two independent sources describe the same underlying behavior from different angles: search does not have to be replaced by a chatbot to stop sending a business traffic. It can simply stop generating a click, because the answer already arrived on the results page itself.
Usage Is Not Trust
The Reuters Institute's 2026 Digital News Report, drawn from roughly 48 markets and about 2,000 respondents per market, found weekly use of AI chatbots for news rose to 10% of respondents globally, up from 7% in 2025. But trust in what those chatbots said stood at just 20% globally, compared with 37% trust in news overall, a meaningful gap between reaching for the tool and believing what it produces.
How often readers click through from an AI-chatbot news answer back to the original source is a genuinely unsettled number. Secondary write-ups of the same 2026 Reuters Institute dataset cite figures anywhere from a low single-digit percentage to well over half, depending on which denominator is used, and a direct check of the primary report reproduced the same inconsistency. Until that figure is resolved against the report's own data tables, it should not be quoted as a single headline number.
The larger lesson holds regardless of that one unsettled figure: adoption curves measure how many people open the tool. They do not measure how many people let the tool decide for them, and the distance between those two things is where a business can badly misjudge how much weight an AI answer is actually carrying with its buyers.
Adoption measures how many people opened the tool. It does not measure how many people let it decide. That distance is the part most easily misread.
The World Is Moving Faster Than Any One Country
DataReportal's mid-2026 update estimates roughly 2.42 billion people, about 29.2% of the world's population, are active users of generative AI tools, against a base of 6.12 billion internet users worldwide, or 73.8% of the population. DataReportal cautions this user count may not represent unique human individuals, given likely overlap from bots and duplicate accounts, but the scale is large enough that generative AI use is now within striking distance of overall internet use globally rather than trailing it by years.
Stanford's 2026 AI Index found adoption correlates strongly with GDP per capita, with notable exceptions: Singapore and the UAE both exceed the adoption rate their income alone would predict, suggesting factors beyond wealth, likely platform availability and policy, are shaping the curve in specific markets.
For a US business, or for the many small and mid-size businesses worldwide watching the same shift, the practical read is that this is not a domestic trend that will eventually reach other markets. It is already a global one, moving at different speeds in different places for reasons that are only partly explained by income.
Why the 25 Percent Collapse Never Came
In February 2024, Gartner forecast that traditional search-engine volume would fall 25% by 2026 because of AI chatbots and virtual agents. That prediction has not played out as stated. By 2026, Google still commands over 90% of the search market, and search volume has not collapsed.
What did happen, over the same period, was a genuine and substantial shift in behavior within that still-dominant search market: the zero-click rate climbed, AI Overviews spread across a growing share of queries, and AI-answer usage grew faster than either the PC or the internet had grown before it. The right model for what is happening is not displacement of one channel by another. It is redistribution of attention within and around a search market that, on the surface, still looks remarkably stable.
That is precisely why a static measurement of "are we ranking" is no longer sufficient on its own. The attention that used to travel in a predictable, clickable path is now settling in places, the answer itself, a chat window, a zero-click summary, that a business has to be positioned to be found in before it ever shows up in a traffic report.
The evidence, in numbers
Key findings, dated and sourced
-
44% of US adults report using ChatGPT in 2026, up from 34% in 2025 and 18% in 2023; 49% of US adults now use an AI chatbot of any kind, up from 33% in 2024, and 24% use one daily.
established Pew Research Center, Americans and AI 2026: Chatbots, Smart Devices and Views on Impact (n=5,119 US adults, +/-1.6pp margin of error) (2026-06-17)
-
Among US AI-chatbot users, ChatGPT leads far ahead of every rival platform: Gemini 24%, Copilot 17%, Meta AI 14%, Grok 8%, Claude 6%, Character.ai 3%. Adults under 50 are twice as likely as those 50 and older to use ChatGPT (57% vs 28%).
established Pew Research Center, Americans and AI 2026 (same survey as above) (2026-06-17)
-
In the first nationally representative US survey of generative AI adoption, roughly 39-40% of adults aged 18-64 had used generative AI as of August 2024. The researchers concluded workplace adoption ran as fast as personal-computer adoption, and overall adoption ran faster than either the PC or the internet.
established National Bureau of Economic Research / Harvard Kennedy School (Bick, Blandin, Deming), "The Rapid Adoption of Generative AI," NBER Working Paper 32966 (2024-09)
-
Stanford HAI's 2026 AI Index reports generative AI reached 53% adoption among the global population within roughly three years of ChatGPT's launch. Adoption correlates strongly with GDP per capita, with outliers like Singapore and the UAE exceeding their income-predicted rate. The report places US adoption at 28.3%, ranking the US 24th globally, notably lower than Pew's US-specific figures.
established Stanford Human-Centered Artificial Intelligence Institute, The 2026 AI Index Report (2026-04)
-
DataReportal's mid-2026 global update estimates roughly 2.42 billion people, about 29.2% of world population, are active users of generative AI tools, against a base of 6.12 billion internet users (73.8% of world population) as of April 2026. DataReportal cautions this figure may not represent unique human individuals due to bot and duplicate-account overlap.
established DataReportal (We Are Social / Meltwater), Digital 2026 Mid-Year Global Update Report (2026 (mid-year update))
-
Adobe Analytics, tracking over a trillion visits to US retail sites, found AI-referred traffic rose approximately 693% year over year during the November-December 2025 holiday season (769% in November, 673% in December). Shoppers arriving via generative AI assistants converted 31% more than other traffic, roughly double the conversion advantage seen the prior holiday season.
established "Generative AI shifts online holiday shopping traffic in 2025" (Digital Commerce 360 recap of Adobe Analytics data) (reported 2026-01-13 (covering Nov-Dec 2025))
-
Salesforce reported that AI and AI agents were associated with roughly 20% of online retail sales and about $262 billion in revenue during the 2025 US holiday shopping season. This rests on Salesforce's own proprietary, non-audited attribution methodology and could not be independently reconfirmed in this research pass, so it is treated as a company-attributed estimate rather than an established fact.
contested Salesforce, 2025 Holiday Shopping Data (reported January 2026)
-
Weekly use of AI chatbots for news rose to 10% of respondents globally in 2026, up from 7% in 2025, across roughly 48 markets (about 2,000 respondents per market). Trust in AI-chatbot news answers stood at just 20% globally, versus 37% trust in news overall.
established Reuters Institute for the Study of Journalism, University of Oxford, Digital News Report 2026 (2026-06-16)
-
Click-through behavior from AI-chatbot news answers back to original sources is reported inconsistently across write-ups of the same 2026 Reuters Institute dataset, with figures ranging from a low single-digit percentage to well over half depending on which denominator is used. A direct check of the primary report's summary page reproduced the same inconsistency, confirming this sub-figure is genuinely unsettled.
contested Reuters Institute for the Study of Journalism, Digital News Report 2026 (primary summary and secondary write-ups disagree on denominator) (2026-06-16)
-
A pre-registered field experiment (1,065 US participants recruited via Prolific, fielded January-February 2026, registered with the AEA RCT Registry) found that Google AI Overviews reduced organic clicks by 38% on queries where they appeared, and raised the zero-click search rate from 54% to 72%, without a measurable improvement in participants' reported satisfaction or information quality. The paper is a working paper, not yet peer-reviewed.
emerging Saharsh Agarwal (Indian School of Business) and Ananya Sen (Carnegie Mellon University), Working paper on the causal effect of Google AI Overviews on search behavior (via Search Engine Journal) (posted SSRN April 2026)
-
SparkToro's analysis of Similarweb US clickstream data (January-April 2026) found 68.01% of Google searches ended without any click, up from 60.45% in 2024, as AI Overviews grew to appear on over 20% of Google searches. SparkToro flags that its clickstream data provider has changed multiple times since 2019, so long-term year-over-year comparisons are not fully methodologically equivalent.
established SparkToro (based on Similarweb data), Google zero-click search analysis (2026-06-09)
-
Gartner's February 2024 forecast that traditional search-engine volume would fall 25% by 2026 due to AI chatbots has not played out as stated: by 2026, Google still commands over 90% of the search market and search volume has not collapsed, even though zero-click behavior and AI-answer usage have both grown substantially over the same period.
contested Gartner (original prediction); independent retrospective analyses, Gartner press release plus 2026 retrospective coverage (e.g., Future Factors, Search Engine Land) (prediction 2024-02-19; retrospective checks published through 2026)
-
Vendor market-share trackers built on Similarweb web-visit data show ChatGPT's share of global AI-chatbot traffic eroding through 2026 (estimates ranging roughly 54-68% by mid-2026, down from a reported 85%-plus share in early 2025) as Google Gemini's share rose into the high twenties percent, with Perplexity at a low single-digit share of visits despite fast year-over-year growth. Different trackers disagree by double digits, and no single primary Similarweb report page was independently verified.
contested Multiple secondary trackers citing Similarweb data, Aggregator market-share trackers (e.g., Vertu Lifestyle, ALM Corp) (figures dated through mid-2026, multiple snapshots)
Learning outcomes
What this study teaches
- The US adoption number depends entirely on which survey you're reading. Treat any single adoption percentage with suspicion until you know its source and its definition of use.
- Adoption moved faster than the PC or the internet did, but usage is not trust, and trust is not the same as being chosen at the moment someone actually decides to buy.
- The commercial effect is measurable and large where anyone has actually measured it, but no independently audited source yet tells any single business how much of its own market's purchase decisions AI answers are swaying.
- Traditional search did not collapse the way some 2024 forecasts warned. It changed shape: zero-click behavior kept rising even as Google held its dominant share, which means visibility now has to work in the answer itself, not only on the page behind it.
- Trust in AI answers still lags usage by a wide margin, so being present in the answer is necessary but not sufficient. Being verifiably credible at the moment a buyer checks the source is what closes that gap.
Honest limits
What this does not yet settle
- No source found gives a rigorous, methodologically transparent measure of AI-chatbot usage specifically for purchase decisions, as distinct from general chatbot usage or usage for news and work. The closest commerce-specific figures traced only to secondary aggregators and could not be confirmed against a primary source, so they were dropped from this study rather than cited on weak footing.
- US adoption estimates vary by more than 20 percentage points across authoritative sources depending on definition and survey design: Pew's 44% ChatGPT use and 49% any-chatbot use from a February 2026 probability panel, Stanford HAI's roughly 28% from its own global methodology, and NBER's 39% of 18-to-64-year-olds as of August 2024. No single number should be quoted as the US adoption rate without naming which survey and definition it comes from.
- The Reuters Institute's AI-chatbot news click-through figure is internally inconsistent across secondary write-ups of the same 2026 report, likely due to different denominators, and needs a direct check of the primary report's data tables before it can be cited as a headline number.
- Similarweb-based AI-chatbot market-share trackers disagree by double-digit percentage points across secondary sources and dates, and no primary Similarweb report page was independently verified. Any specific market-share figure should be read as directional, not exact.
- No peer-reviewed, causally identified study was found isolating the effect of AI-chatbot usage specifically on purchase decisions. The strongest causal evidence available, the Agarwal and Sen field experiment, measures Google AI Overviews' effect on search-click behavior generally, not chatbot use for shopping specifically, and remains an unpublished working paper.
- Country-level data on AI-answer usage for commercial or purchase-relevant tasks outside a handful of well-studied markets is thin. The best country-granular sources available cover news consumption and general adoption, not shopping behavior, so genuine adoption-curve comparisons for purchase decisions specifically remain largely US-centric.
- Salesforce's approximately 20%-of-holiday-sales figure and Adobe's 693% year-over-year AI-referral-traffic figure are correlational, influence-oriented measures built on proprietary methodologies, not independently audited causal estimates of how much AI answers actually change buying outcomes.
This is a synthesis of dated, attributed evidence, not a census. The AI-answer layer in particular has no independent, Nielsen-grade measurement yet, so readings of it are directional and named as a frontier, never presented as settled.
Straight answers
Frequently asked questions
Is AI-answer adoption actually growing faster than past technologies?
Yes, on the evidence available. An NBER study by Bick, Blandin, and Deming found generative AI adoption ran faster than the personal computer or the internet did in their own eras, and Stanford's 2026 AI Index estimated generative AI reached 53% of the world's population within roughly three years of ChatGPT's November 2022 launch. That speed shows up in two independent research efforts using different methods, so the pace itself is well established. What it doesn't tell you is how deeply people trust or rely on what they're using, which is a separate and much less settled question.
What percentage of Americans actually use ChatGPT or an AI chatbot?
It depends entirely on which survey you read. Pew Research Center's February 2026 survey of US adults put ChatGPT use at 44%, with 49% using some kind of AI chatbot; the NBER study found 39 to 40% of adults aged 18 to 64 had used generative AI as of August 2024; and Stanford's 2026 AI Index, using a different global methodology, placed US adoption at just 28.3%. These figures differ by more than twenty percentage points because they ask different questions of different populations at different times, so any single adoption number should always be quoted with its source and its definition of use attached.
Is search traffic actually collapsing because of AI chatbots?
No, not in the way a widely cited 2024 forecast predicted. Gartner forecast in February 2024 that traditional search-engine volume would fall 25% by 2026 due to AI chatbots, but by 2026 Google still commands over 90% of the search market and search volume has not collapsed. What has genuinely changed is the shape of engagement within that still-dominant market: SparkToro's analysis of Similarweb data found 68.01% of Google searches now end without a click, up from 60.45% two years earlier, as AI Overviews grew to appear on more than a fifth of all searches.
If people are using AI answers more, does that mean they trust them and buy based on them?
Not necessarily, and the study treats usage and trust as two different things that move at different speeds. The Reuters Institute's 2026 Digital News Report found weekly AI-chatbot use for news rose to 10% globally, yet trust in what those chatbots said stood at just 20%, well below the 37% trust placed in news overall. There is a real commercial effect where it has actually been measured, Adobe Analytics found AI-referred traffic to US retail sites grew 693% year over year over the 2025 holiday season with a 31% conversion lift, but no independently audited source yet tells a given business how much of its own market's purchase decisions AI answers are actually swaying.
Which AI platform should a small business actually pay attention to, just ChatGPT?
ChatGPT is still the platform US adults name first by a wide margin, but the study found the mix beneath it is genuinely reshuffling and not settling on one winner. Among US AI-chatbot users, Pew found Gemini at 24%, Copilot at 17%, Meta AI at 14%, Grok at 8%, and Claude at 6%, and vendor trackers built on web-visit data show ChatGPT's share of AI-chatbot traffic eroding through 2026 as Google Gemini's share climbs, though different trackers disagree by double digits on the exact numbers. The read is that reaching the AI answer today is a moving, multi-platform problem rather than a single-platform one.
Provenance
References
- Pew Research Center, "Americans and AI 2026: Chatbots, Smart Devices and Views on Impact" https://www.pewresearch.org/internet/2026/06/17/americans-and-ai-2026-chatbots-smart-devices-and-views-on-impact/
- Bick, Blandin, and Deming, "The Rapid Adoption of Generative AI," NBER Working Paper 32966 https://www.nber.org/papers/w32966
- Stanford Human-Centered Artificial Intelligence Institute, "The 2026 AI Index Report" https://hai.stanford.edu/ai-index/2026-ai-index-report
- DataReportal, "Digital 2026 Mid-Year Global Update Report" https://datareportal.com/reports/digital-2026-mid-year-global-update-report
- Digital Commerce 360, "Generative AI shifts online holiday shopping traffic in 2025" (Adobe Analytics data) https://www.digitalcommerce360.com/2026/01/13/generative-ai-online-holiday-shopping-traffic-2025/
- Salesforce, "2025 Holiday Shopping Data" https://www.salesforce.com/news/stories/2025-holiday-shopping-data/
- Reuters Institute for the Study of Journalism, University of Oxford, "Digital News Report 2026" https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2026/dnr-executive-summary
- Search Engine Journal, "AI Overviews Cut Organic Clicks 38%, Field Study Finds" (Agarwal & Sen working paper) https://www.searchenginejournal.com/ai-overviews-cut-organic-clicks-38-field-study-finds/573145/
- Search Engine Land, "Google zero-click search analysis" (SparkToro / Similarweb data) https://searchengineland.com/google-zero-click-searches-2026-study-479717
- Gartner, "Gartner Predicts Search Engine Volume Will Drop 25 Percent by 2026" (2024-02-19 press release) https://www.gartner.com/en/newsroom/press-releases/2024-02-19-gartner-predicts-search-engine-volume-will-drop-25-percent-by-2026-due-to-ai-chatbots-and-other-virtual-agents
- Vertu Lifestyle, AI-chatbot market-share tracker (Similarweb-based) https://vertu.com/lifestyle/ai-chatbot-market-share-2026-chatgpt-drops-to-68-as-google-gemini-surges-to-18-2
Every measured figure is dated to its capture and tagged with an evidence tier. Every cited work is real and locatable. Where an engine could not be captured this round, it is named as uncaptured, not estimated. Small-sample readings are labelled as directional.