ChatGPT’s Stranglehold Breaks as AI Search Fragments Three Ways

ChatGPT’s Stranglehold on AI Search Cracks Open as the Market Splinters Three Ways

For three and a half years, ChatGPT defined what generative AI search meant. That era ended quietly across twelve months of data that Similarweb has now made public.

ChatGPT’s share of worldwide generative AI web traffic slid from roughly 76% in mid-2025 to approximately 53% by May 2026, a drop of more than 20 percentage points in a year. Gemini absorbed the bulk of that decline, climbing from under 9% to roughly 27-28%, while Claude quadrupled its footprint from barely 2% to nearly 9%.

The category itself is still expanding at breakneck speed. Average monthly web visits across generative AI platforms grew 70% year-over-year, reaching 9.5 billion between mid-2025 and mid-2026, while unique visitors rose 57% to 655 million.

That gap between visit growth and audience growth tells its own story. Existing users are opening these tools far more often, not just new people arriving for the first time.

For marketers and GEO strategists, the practical implication is stark: optimizing for a single AI platform no longer covers the addressable audience. A strategy built around ChatGPT alone now reaches roughly half the market.

Inside the 2026 Data: Market Share Redistribution, Mobile Divergence, and the Rise of Passive AI

The Web Traffic Redistribution

According to Similarweb, ChatGPT’s absolute visit count stayed roughly flat over the twelve-month period. The denominator simply expanded fast enough to make a flat leader look like a declining one.

Gemini and Claude are winning through fundamentally different mechanisms. Gemini rides distribution it already owns through Search, Android, and Workspace, scaling as aggressively as Google chooses to push it.

Claude, by contrast, is winning through deliberate adoption by developers and power users. That pool grows more slowly but tends to stick around longer once committed.

The distinction matters enormously for prioritization. If reach is the goal, Gemini’s distribution advantage makes it the bigger near-term prize. If a technically sophisticated, high-intent audience is the target, Claude’s growth pattern points the other way.

Mobile Tells a Radically Different Story

On US iOS and Android, the platforms barely registering on the web share chart are the fastest movers. Meta AI and Claude posted the sharpest US app growth between mid-2025 and mid-2026.

Gemini, the platform actually consuming ChatGPT’s web share, grew the slowest of the challengers on mobile. ChatGPT itself still grew solidly on its own app even as its web share slid.

This suggests ChatGPT is defending the app it controls far better than it defends the open web, where a user can switch platforms mid-search without installing anything. Meta AI’s worldwide app growth runs nearly as high as Claude’s, and the mechanism is instructive.

Meta AI is not a destination people seek out the way they seek Claude or ChatGPT. It is embedded inside apps billions of people already had open, matching a standalone challenger’s growth rate without asking anyone to download anything new.

That embedded distribution model is a preview of where much of future AI usage is likely to originate. Claude’s app story also skews heavily international rather than US-led, with worldwide app monthly active user growth dramatically outpacing its domestic numbers.

Referral Patterns and Citation Economics

In May 2026, ChatGPT changed how it surfaces brand links, replacing footnote-style citations with clickable brand names inside its answers. The effect was not a one-week spike followed by decay.

Homepage referrals from ChatGPT rose from roughly 26-29% of all referral traffic before the update to approximately 62-63% by late May 2026. That level held rather than faded, a plateau that signals a permanent behavioral shift rather than a novelty effect.

Total referrals and homepage-specific referrals did not grow at the same rate after the update. Homepage referrals grew several times faster, which means the update did not simply push more of ChatGPT’s existing traffic through the same door.

It changed what a meaningful chunk of that traffic actually is. Users are increasingly treating the brand name inside the answer as the endpoint of their research, with the homepage serving as a formality before converting or committing the brand to memory.

This traffic behaves less like a search click and more like the tail end of an ad impression. Homepage spikes that used to get filed under ‘no clear source’ increasingly deserve a second look before being dismissed as unattributed.

On the citation front, the rate at which ChatGPT cites sources in its answers remains low in absolute terms but is climbing steeply. Citation presence in US ChatGPT prompts rose from roughly 1.6% in mid-2025 to approximately 6.8% by May 2026, more than quadrupling over eleven months.

The trajectory was not a straight line. Citations jumped sharply starting in October 2025, then eased back for roughly four months before resuming their climb into spring 2026.

Citation rates also vary enormously by category. Travel and hospitality prompts carry a citation roughly 23% of the time, and automotive hovers around 20%. Professional services sits under 4%.

Categories where users compare specific, checkable options get cited noticeably more often than categories built on general advice. Brands in travel and automotive have substantially less runway to establish citation visibility than those in advice-heavy verticals.

Advertising Enters the AI Response Layer

Roughly 26% of ChatGPT responses now contain an ad, a dimension of AI visibility that did not exist when the market share chart looked the way it used to. About a third of all ad impressions land in the very first response of a session, before the user has revealed much about their actual intent.

That share drops quickly with each subsequent message, falling to single digits by the third response. Yet a meaningful fifth of all ad impressions still appear ten or more messages into a conversation, meaning long exploratory sessions remain a live opportunity well after the obvious early real estate is gone.

For brands, this adds a genuinely new competitive layer to AI visibility. It is no longer only about being mentioned or cited; it is increasingly about who is buying placement around the answer a brand is trying to be found in.

Search Behavior Shifts Toward Passive Integration

Google’s AI rollout looks less like a single feature launch and more like two different bets running simultaneously, and one of them is clearly winning. AI Overviews, which appear automatically inside searches users already know how to run, have kept climbing without requiring any behavioral change.

AI Mode, which asks users to deliberately switch into a conversational interface, grew quickly early on and has since leveled off and started to soften. People appear far more willing to accept AI inside the tools they already use than to adopt a new tool built around AI.

Passive integration is beating deliberate reinvention, at least for now. Search queries have also grown noticeably longer since AI-native search became normalized, which only makes sense if users trust the system to interpret context rather than doing keyword engineering themselves.

Clipped, keyword-style searches were a habit built around a keyword-matching engine. Longer, conversational queries are a habit built around a system users believe can actually understand context, and that shift quietly redefines what ‘ranking for a keyword’ even means.

When AI Queries Replace Search Bar Habits: The 9% Erosion Reshaping Digital Discovery

The market share redistribution inside the AI category is only half the story. The other half is what AI search is doing to traditional search volume, and the data emerging from academic research paints a picture that should command the attention of every SEO and GEO practitioner.

Researchers at Bocconi University analyzed Comscore clickstream data to measure how broader ChatGPT access affected traditional search behavior, and the findings are sobering. Broader access reduced weekly traditional search queries by an average of 9.4%, with the decline reaching 17% after twenty weeks.

Among users who already used ChatGPT before broader access rolled out, the decline averaged 4.9%, rising to 8.2% after twenty weeks. The largest drops concentrated in informational searches, where academic research referrals fell 32.8% and reference queries dropped 26.5%.

Transactional and recreational searches, by contrast, were nearly unchanged. The pattern is clear: AI is not replacing all search behavior equally; it is disproportionately absorbing the informational layer that has historically driven enormous organic traffic volume.

The Bocconi researchers are careful to frame their findings precisely. As Search Engine Journal reports, the study measures only observable traffic reallocation.

The authors explicitly note the findings measure a change in observable traffic allocation, not consumer surplus, publisher revenue, or long-run content production, and emphasize that mobile, international, and revenue data are needed before drawing broader conclusions.

Yet the directional signal is impossible to ignore. A near-double-digit query decline tied directly to ChatGPT access, concentrated in informational intent, represents a structural shift rather than a temporary fluctuation.

There is competitive tension in the data worth surfacing. While the Bocconi study captures a 9.4% decline in traditional search, other research compiled across the industry suggests total combined search usage, traditional engines plus LLMs, grew roughly 26% worldwide.

AI is both cannibalizing traditional search and expanding the total search pie simultaneously. The brands capturing the net-new query volume are not necessarily the same ones losing the traditional volume, which means the reallocation creates winners and losers rather than a uniform tide.

Previsible’s third AI Traffic Study, covering 6.77 million LLM-driven sessions across 166 websites from late 2024 through mid-2026, adds further granularity. Total monthly standalone LLM-referred sessions grew 9.9x over the period, from roughly 65,000 in November 2024 to over 644,000 by May 2026.

ChatGPT held a 92.4% share of trackable standalone AI referrals in that dataset, up from roughly 84% in late 2025. Its referrals rose 12.8x, dwarfing Gemini’s 3.2x growth and Claude’s 64x climb from a tiny base.

This appears to contradict Similarweb’s web traffic picture of ChatGPT at 53% share, but the two datasets measure fundamentally different things. Similarweb tracks all web visits to AI platforms, while Previsible tracks only referral sessions that actually land on participating websites.

A platform can command enormous usage while sending relatively few referrals, and a platform with smaller overall traffic can punch above its weight in referral generation. The gap between usage share and referral share is itself a competitive signal worth monitoring.

Marketing Tech News reports that Google AI Overviews and AI Mode together now represent a larger volume of AI-influenced traffic than all standalone LLM platforms combined. For GEO strategists, this means Google’s AI surfaces remain the highest-volume prize, even as ChatGPT dominates the standalone referral landscape.

AI traffic penetration varies dramatically by industry. E-commerce saw a 37x increase in LLM-referred sessions, with product pages serving as the primary landing surface. Insurance climbed 18.9x to reach 1.51% of total sessions, while health was the only vertical to decline.

The conversion dynamics are equally uneven. AI-referred visitors demonstrate roughly double the engagement metrics, pages per visit and time on site, compared to standard visitors, and conversion rates from different LLM sources span a wide range.

In AI search environments, the first result recommended by the AI becomes the user’s choice roughly 74% of the time. Organic click-through rates are approximately 35% higher when a brand is cited inside an AI Overview, making citation presence a direct conversion multiplier rather than merely a vanity metric.

One finding that deserves particular attention: roughly 61.7% of LLM citations appear to be ‘ghost citations,’ where a domain is used as a source but the brand name never appears in the answer text. Brands investing heavily in citation-worthy content may be earning sourcing credit without earning visibility credit, a gap that traditional citation tracking would miss entirely.

Beyond Single-Platform GEO: Building Visibility Across a Fragmented AI Ecosystem

The AI platform market went from a single dominant player to a genuinely competitive three-way field in roughly twelve months, and the rules governing referral behavior, citations, and advertising are still being written in real time. The brands treating these shifts as an annual checkpoint rather than a monthly monitoring priority are already falling behind.

Coverage now means more than one platform. With ChatGPT below 55% of AI web traffic and both Gemini and Claude holding meaningful, growing shares, a GEO strategy built around a single assistant reaches a shrinking fraction of the AI-using audience. The platforms are winning through different mechanisms, distribution versus developer adoption, and the optimal visibility strategy differs accordingly.

Homepage and direct traffic deserve forensic investigation. ChatGPT’s referral behavior has settled into a pattern where roughly six in ten referred visits land on a homepage rather than a specific page, making this traffic behave less like organic search and more like brand-driven discovery. Traffic previously filed under ‘direct, no clear source’ may increasingly be AI-driven discovery that simply does not carry a clean referral tag.

The citation window remains open but is narrowing by category. A sub-7% overall citation rate means most content still does not appear in AI answers verbatim, but categories built on comparable, checkable facts are already seeing citation rates three to five times the average. Brands in those verticals have less time to establish structured, referenceable content footprints before the competitive field fills in.

Ads inside AI answers introduce a new variable that compounds every other visibility challenge. Appearing early in a session and being cited cleanly both carry more weight when a quarter of responses also carry paid placements competing for the same user attention.

The GEO discipline is maturing from a speculative edge into an operational necessity, and the infrastructure required to track brand presence across multiple AI platforms, monitor citation fidelity, and measure the conversion value of AI-driven discovery is fundamentally different from the tooling built for traditional search. The teams building that capability now, before the fragmentation accelerates further, are positioning themselves for whatever the field looks like when the 2027 data arrives.

For brands and practitioners navigating this fragmentation, the technical foundation underpinning AI discoverability is increasingly inseparable from site performance, structured data architecture, and the automation pipelines that make multi-platform visibility scalable rather than manually exhausting. Programmatic SEO and AI-driven content automation have moved from experimental tactics to core infrastructure for any organization serious about maintaining presence across a splintering AI search landscape. The same logic applies to the technical performance layer: AI crawlers and citation engines favor fast, cleanly architected sites, making speed optimization a direct input into AI visibility rather than merely a user experience concern.

For teams ready to build that foundation, programmatic SEO and AI automation services provide the pipeline infrastructure to scale GEO efforts across platforms, while WordPress speed engineering ensures the technical performance layer supports rather than undermines AI discoverability. To explore how these capabilities fit a specific GEO strategy, reach out to Andres directly or learn more about the methodology behind Andres SEO Expert.