The Way Your Customers Find You Is Breaking – 3 Mar 2026
Five data points on AI user behaviour — and what they actually mean for your business in 2026.
Most AI coverage focuses on the wrong story
Adoption rates, investment figures, productivity gains – these are real, but they are lagging indicators.
The more urgent story is happening at the level of individual behaviour: how people search, who they trust, how they make decisions, and where your business fits in the new information landscape.
The data from the last three months (December 2025 through February 2026) reveals a set of interconnected shifts that are more significant, and in some cases more alarming, than the headline numbers suggest. This report covers five of them in depth.
Google’s Six-Month Cliff
For nearly two decades, Google held over 90% of the global search market share.
No competitor — not Bing, not Yahoo, not DuckDuckGo — managed to move that number meaningfully.
Then, in a single six-month window between February and August 2025, the first genuine market share event in the history of search happened quietly, without a single major news event to trigger it.
Google's share of general information searches fell in just six months — its biggest decline in years. In the same period, ChatGPT's share of general search queries tripled from 4.1% to 12.5%. OpenAI's CFO confirmed this to CNBC directly.
The data gets more specific. Daily AI tool usage more than doubled over the year. Among Americans under 30, 74% now use AI tools to search. Among Gen Z specifically, 53% say they go to TikTok, Reddit, or YouTube before Google when looking for something — and an increasing share bypasses Google entirely for research and decision-making queries.
This is not a slow erosion. The HigherVisibility study that captured the 73% to 66.9% shift noted that 34.8% of people reported actively changing their search behaviour by August 2025, up from 27.7% just six months earlier. The word “changing” is important here — these are not passive shifts but deliberate new habits.
Why this matters for business leads: SEO strategies built for a keyword-search world are becoming structurally less effective. The question is not whether to act but whether to act before or after your competitors do. The companies adapting now are building visibility inside AI systems — not just Google rankings.
One finding buried in the search data reveals the full scale of what is happening: ChatGPT is on pace to potentially overtake Google’s total search volume by 2027. Not in a niche. In aggregate. If that projection holds, the entire discovery infrastructure on which most businesses were built will have been replaced within two years.
A Billion Users, an Inch Deep
When AI-powered search launched at scale, most publishers and brands assumed it would behave like previous search innovations — sending traffic, just in new ways.
The data from 2025 destroyed that assumption. AI answers the question and keeps the user on the platform. Nobody clicks. Nobody arrives.
Zero-click searches in the twelve months following Google AI Overviews' launch. Over 600 million monthly visits disappeared from publisher websites in under a year. Organic traffic to news publishers fell from 2.3 billion to 1.7 billion monthly visits.
The specific publisher data is stark. The Washington Post lost 40% of its traffic. NBC News declined 42%. Forbes and HuffPost both recorded 50% losses. Business Insider cut 21% of its staff. The Wall Street Journal, The New York Times, CNN — none were spared. These are not niche content businesses or blogs that were poorly optimised. They are the most recognisable media brands in the English-speaking world.
The counterintuitive detail: Forbes is cited in AI Overviews over 44,000 times. It still lost half its traffic. Being mentioned in an AI answer and being visited are no longer the same thing. For brands that have spent years building content authority, this separation is the central strategic problem of 2026.
The replacement math does not work: ChatGPT referrals to news publishers grew 25x year-over-year — a significant number. But news-related AI prompts grew 212% in the same period. The demand is flowing into AI and staying there. All AI platforms combined account for just 1% of publisher traffic.
There is one data point that cuts through the pessimism, though it comes with a significant asterisk. Research from Microsoft Clarity found that visitors who arrive via AI platforms convert to sign-ups at 1.66% — compared to 0.15% from search. That is more than a 10x difference in conversion quality. The traffic that does come from AI is disproportionately valuable.
The asterisk: the volumes are so small that even with a 10x conversion advantage, AI referrals cannot offset search losses at the current scale. The strategic bet is that this changes — that as AI platforms mature, they develop more robust referral mechanisms. Whether that happens before publishers reach a financial breaking point is an open question.
The Web Traffic Black Hole
The adoption headlines are technically accurate and strategically misleading. Over a billion people have used standalone AI tools. AI features reach two billion people through Google search alone.
These numbers are real — and they obscure the most important question: how deeply has AI embedded itself into how people actually work and live?
The highest penetration rate for any single AI use case — writing emails. Not one activity sees more than 1 in 5 people relying on AI for it. This is despite more than half of US adults having tried AI at least once.
Menlo Ventures’ State of Consumer AI report puts this in plain terms: AI is useful everywhere, but there is still a long way to go to everyday adoption. The activities with the highest reported penetration — writing (51%), coding (47%), school assignments (43%) — are the percentage of people who have “ever used AI” for the task, not who rely on it. The gap between those two numbers is enormous.
The monetisation picture tells the same story. Only 3% of AI users globally pay for premium services. ChatGPT, the fastest-adopted product in the history of consumer technology, converts just 5% of weekly active users into paying subscribers. A product used by 800 million people weekly is struggling to monetise at scale because use has not deepened into dependency.
The real opportunity: The window before daily AI habits lock in is open right now. Consumer behaviour research consistently shows that once a tool becomes part of someone's daily ritual, switching costs become prohibitive. Whoever converts casual AI use into genuine reliance — in your industry, for your customers — builds a structural moat. That conversion has not happened yet for most use cases.
The data also reveals which sectors are closest to that inflection. The Menlo Ventures study found that parents are among the most engaged AI users of any demographic — using it for childcare management, note-taking, and topic research — ahead of many professional groups. This is unexpected, and it signals that emotional utility (reducing decision fatigue, organising overwhelm) may be the vector through which deep AI habits form, not professional productivity.
Trust by Fluency
This is the most counterintuitive finding in recent AI behaviour research, and arguably the most commercially significant. The problem is not that AI gets things wrong — it is that users have developed a systematic bias toward not checking.
Of users who admit they do not check sources if the AI answer looks convincing. A further 65.9% say citations increase their trust in AI answers — yet only 27% ever click those citations.
Researchers named this pattern “trust by fluency.” The mechanism is straightforward: well-structured, confident prose feels authoritative. Citations provide a visual trust signal without requiring verification. The result is that a polished AI answer — even one that is factually wrong — receives a level of credence that an equivalent answer from a random website would not.
The Reuters Institute, in a study across six countries, found that trust in AI search answers sits at 50% among those who encounter them, moderate, not high. But crucially, that trust becomes selective. In high-stakes domains — health, legal questions, financial decisions — users verify more. For everyday queries, the default is passive acceptance.
The sector risk: Legal, finance, and health are the three fastest-growing categories in AI search traffic — and they are the three domains where acting on a wrong AI answer carries the highest consequences. Legal sector AI traffic grew 11.9x. That is not a niche trend.
There is a second layer to this finding. A KPMG global study of 48,000 people across 47 countries found that trust in AI has actually decreased as adoption has increased. More use has made people less trusting — but the verification behaviour data shows it has also made them less likely to check. The combination of declining trust and declining verification is not contradictory; it describes a population that is simultaneously sceptical and passive. That is a difficult state for regulators, product designers, and businesses alike.
For brands, this creates a specific challenge: in a world where AI answers carry implicit authority by virtue of their presentation, the accuracy and tone of how your brand appears inside AI systems matter in ways that your direct communications never did. You cannot edit an AI answer. You can only shape the information environment it draws from.
2027 – The Inflection Year
The four findings above are not independent trends. They are compounding toward a single structural shift: the moment AI becomes the primary interface between humans and information — not a supplement to existing channels, but the layer through which discovery, evaluation, and trust are built.
The data points to 2027 as the year this becomes undeniable. ChatGPT is on pace to match Google’s query volume by then. Google is simultaneously building AI into 60% of its own search results, accelerating the zero-click trend it is trying to own. The zero-click rate is already at 69% and rising. And user habits — particularly among the under-30 cohort that will represent the majority of purchasing decisions by the end of the decade — are solidifying around AI-first information behaviour.
What wins in this environment:
- Brands that understand AI visibility is not the same as search ranking — and invest accordingly.
- Products that convert casual AI use into daily reliance — by solving emotional problems, not just productivity ones.
- Businesses that own a direct relationship with their audience — newsletters, communities, subscriptions — that do not depend on search referral traffic to exist.
- Leaders who recognise that the trust crisis is a product problem, not just a PR one — and design accordingly.
The window to adapt is measured in months, not years. The businesses watching this shift from the sidelines are not waiting for clarity — they are ceding ground to the ones that are moving
Sources & Methodology
Data window: December 2025 – February 2026. All statistics sourced from primary research unless noted. Key sources include: HBR / Filtered Top-100 Gen AI Use Case Report (April 2025) · DataReportal Digital 2026 Global Overview (October 2025) · Similarweb AI Overviews & Zero-Click Analysis (July 2025) · Pew Research Center Search Behaviour Study (October 2025) · Menlo Ventures State of Consumer AI (November 2025) · KPMG Global AI Trust Study 47-Country Survey (November 2024 – January 2025) · McKinsey State of AI 2025 (November 2025) · HigherVisibility Search Behaviour Study (August 2025) · Microsoft AI Economy Institute Global Adoption Report (January 2026) · AdExchanger / Digiday Publisher Traffic Analysis (December 2025 – January 2026).
