ChatGPT Search vs Google in 2026: Google Still Wins 373:1

By Priya Shah 2026-07-31 16 min read

ChatGPT Search launched on October 31, 2024 as an open challenge to Google, and twenty-one months later the scoreboard is lopsided. Google still processes roughly 373 times more searches than ChatGPT, controls more than 90 percent of mobile search, and now serves AI Overviews to about 1.5 billion people every month. ChatGPT did not replace Google. It did something more consequential: it forced Google to rebuild its results page around generative answers, accelerated a zero-click economy that is draining referral traffic from publishers, and exposed how much of the AI search stack still quietly runs on Bing and Google underneath.

What ChatGPT Search actually shipped, and why Google noticed

On October 31, 2024, OpenAI turned ChatGPT into a search engine. The feature added live web browsing, inline source attribution, and real-time answers for timely queries like news, sports scores, weather, and stock prices, all inside the chat window rather than as a separate product. Bloomberg reported that the release ran on OpenAI's GPT-4o model and launched first for paying Plus and Team subscribers, with enterprise, education, and free tiers to follow. The framing was unambiguous. TechCrunch called it a Google challenger, and OpenAI shipped a Chrome extension that let users set ChatGPT as their default search engine, aiming directly at the box where Google makes most of its money.

The strategic logic was clear. Instead of ten blue links and a page of ads, ChatGPT returned a single synthesized answer with citations, collapsing the research task into one response. For a generation of users who had already learned to paste questions into a chatbot, the jump from asking ChatGPT to asking ChatGPT to search felt like no jump at all. Within weeks the rollout widened. By December 2024, OpenAI extended search to all logged-in accounts, and later removed the login wall entirely.

The features that became the industry template

What shipped in that first release set the pattern every AI search product has copied since:

• Live web results pulled at query time, not from a static training cutoff
• Inline numbered citations linking back to source publishers
• Purpose-built answer cards for weather, sports, stocks, and maps
• A conversational follow-up model that kept context across a session
• A Chrome extension built to hijack the default search slot
• Tiered access that used search as a reason to upgrade to Plus

Google had seen this coming. Five months earlier, in May 2024, it had launched AI Overviews, its own generative summaries stitched into the top of the results page. The race was already on before OpenAI fired its shot, and both companies knew the prize was the same query box that has funded Alphabet for two decades.

ChatGPT search vs Google by the numbers in 2026

Strip away the narrative and the raw volume gap is enormous. According to Advanced Web Ranking, Google fields roughly 14 billion searches a day against ChatGPT's estimated 37.5 million search-like prompts, a ratio of about 373 to one. On a pure search-engine-share basis, Advanced Web Ranking puts Google at 93.57 percent and ChatGPT at 0.25 percent. Trackers that count AI chat prompts as search activity are kinder to OpenAI, placing ChatGPT nearer 9 to 12 percent of all query volume, but even the generous readings leave Google in a different weight class.

The engagement picture, though, is where the story turns. ChatGPT crossed 900 million weekly active users in February 2026 and passed 1 billion monthly users in June 2026, and its search feature alone handles an estimated 250 to 500 million queries a week. A typical Google session lasts a little over five minutes; a typical ChatGPT session runs past fourteen. Those are not the same behavior, and the second one keeps expanding. The table below lays out the comparison as it stood in mid-2026.

Two figures deserve attention. The first is that Google's dominance is measured in volume, not engagement, and volume is the metric under the most pressure. The second is the last row: PPC Land found that ChatGPT sends roughly 190 times less referral traffic than Google despite handling around 12 percent of search-like volume, because a cited answer rarely produces a click. That gap between usage and traffic is the single most important number in this entire debate, and it reframes what winning even means.

How Google answered: AI Overviews took over the results page

OpenAI's real achievement was not stealing Google's traffic; it was changing what a search result looks like, and forcing Google to match it at global scale. Google's AI Overviews, launched in May 2024, expanded relentlessly through 2025 and into 2026. Advanced Web Ranking's monitored dataset tracked the climb quarter by quarter, and the trajectory tells you everything about how seriously Google took the threat. What began as an experiment appearing in fewer than one in five results became a fixture on roughly half of them.

Measurements vary because every tracker uses a different keyword sample. Conductor's Q1 2026 benchmark across 21.9 million queries recorded 25.11 percent, while Similarweb data cited in July 2026 put AI Overviews in about 43 percent of searches, up from 15 percent a year earlier. The spread is wide, but the direction is not in dispute. Industry-level data shows AI Overviews on 88 percent of healthcare queries and 83 percent of education queries, against 37 percent for entertainment.

The pattern is a defensive one. Every percentage point of AI Overviews coverage is a query where Google answers in place rather than sending the user onward. Google is cannibalizing its own click economy to stop users from leaving for ChatGPT. Longer queries accelerate the shift: prompts of eight words or more are about seven times more likely to trigger an AI Overview, and long conversational phrasing is exactly what ChatGPT trained a billion users to write. Google is not just defending share; it is defending the habit.

The uncomfortable secret: ChatGPT search runs on Bing, and maybe Google

Here is the twist that reframes the whole rivalry. ChatGPT's search results are not built on an independent OpenAI index. They lean heavily on Microsoft Bing, and a series of 2025 studies suggested Google's own results are in the mix too. In February 2025, Seer Interactive analyzed more than 500 citations across 100 queries and found that over 87 percent of SearchGPT citations matched Bing's top organic results in positions one through twenty, while Google matched only 56 percent with a median rank of 17.

"It's not just Bing, it's Bing's top rankings driving what SearchGPT cites." Christina Blake and Alisa Scharf, Seer Interactive, February 6, 2025.

Then it got stranger. In July 2025, Abhishek Iyer, a former Google search infrastructure engineer, ran a sting operation modeled on Google's own 2011 investigation into Bing. He seeded a unique nonsense term and watched where it surfaced. PPC Land reported that Iyer found URLs carrying Google's distinctive "srsltid" tracking parameter inside ChatGPT responses, which he read as evidence of Google Search infrastructure being used behind the scenes despite OpenAI's documentation naming only Bing and Shopify. Backlinko's Brian Dean ran a parallel test in August 2025 with a fabricated term and reached the same conclusion.

The takeaway: index visibility is AI visibility

The commercial layer tells the same story. Search Engine Land documented that ChatGPT sources 83 percent of its shopping carousel products from Google Shopping data through query fan-outs. The strategic takeaway for marketers is blunt: if Google does not index you, many AI tools will not surface you either. The visibility playbook that wins in ChatGPT is not separate from search; it runs straight through the same crawlers, the same structured data, and the same top-of-page rankings that have always mattered. This is why generative engine optimization and classic technical SEO are converging rather than competing, and why teams that treat them as two budgets are paying twice for one outcome.

The zero-click reckoning both engines created

Whichever engine wins, the publisher loses clicks. That is the shared consequence of the AI answer race, and the data is now unambiguous. Pew Research found that when a Google AI summary appears, the share of searches producing a click to any website falls from 15 percent to 8 percent, and only 1 percent of users click a link inside the summary itself. Independent trackers report that around 60 percent of Google searches now end without a click at all, and on AI-heavy query sets that figure climbs to roughly 83 percent.

Position rank no longer protects you the way it did. Multiple 2026 analyses put the position-one organic click-through rate at about 1.6 percent when an AI Overview is present, down from roughly 7.3 percent two years earlier, close to a 78 percent collapse for the single most valuable slot in search. Aggregate click-throughs have fallen nearly 30 percent since AI Overviews launched in May 2024. Here is how the erosion breaks down:

• Clicks when an AI summary shows: 8 percent, down from 15 percent (Pew Research)
• Clicks on a link inside the AI summary: about 1 percent
• Zero-click share of all Google searches: roughly 60 percent
• Zero-click share on AI-heavy queries: about 83 percent
• Position-one CTR with an AI Overview present: about 1.6 percent, from 7.3 percent
• Overall CTR decline since May 2024: nearly 30 percent

ChatGPT is not the cleaner alternative here. Its 190x traffic deficit against Google means a cited answer sends even fewer visitors onward. The net effect is a two-front squeeze on referral traffic, and it explains why brands are shifting budget from click-based measurement toward citation and mention tracking. When the answer is the destination, being named in the answer is the only visibility that counts, which is the core premise behind answer engine optimization.

Why a 14-minute ChatGPT session worries Google more than share does

Market share is a lagging indicator. Behavior is the leading one, and the behavioral divergence between the two products is stark. A Google search session averages a little over five minutes and is optimized for speed: type, scan, click, leave. A ChatGPT session averages more than fourteen minutes and is optimized for depth: ask, refine, follow up, decide. Those are not the same activity, and the second one is eating into the queries that used to start on Google.

The queries migrating fastest are the high-intent, high-consideration ones: research, comparison, planning, coding help, and drafting. These are precisely the sessions where a user historically ran five or six separate Google searches, and where the advertising was most valuable. Google keeps the enormous long tail of navigational and quick-fact lookups. ChatGPT is capturing the deliberative middle, and the deliberative middle is where purchase decisions get made. That distinction matters more than the headline share number for anyone selling a considered product.

The usage split looks roughly like this:

• Navigational and quick facts: still overwhelmingly Google
• Local and maps queries: still overwhelmingly Google
• Research and multi-step comparison: shifting toward ChatGPT
• Coding, drafting, and analysis: dominated by ChatGPT
• Shopping discovery: contested, with ChatGPT leaning on Google Shopping data
• Breaking news and live scores: contested, both offer real-time cards

This is also why the AI chatbot category itself is consolidating around ChatGPT. Its share of chatbot usage has slipped from a peak above 87 percent to somewhere between 68 and 77 percent through 2026 as Gemini, Perplexity, Copilot, and Claude compete, but it still commands the majority of the category. For growth teams, the implication is that a single platform now shapes how a large share of your highest-intent prospects research their next purchase, which is why an AI visibility audit has moved from optional to baseline.

What the executives are actually saying

The leaders on both sides have been unusually candid, and their words map the strategy better than any analyst note. Sam Altman has repeatedly declined to claim outright victory over Google. Testifying before the U.S. Senate in May 2025, when asked whether ChatGPT would replace Google as the main search engine, he answered plainly and called his rival a ferocious competitor.

Asked if ChatGPT would replace Google as the top search engine, OpenAI CEO Sam Altman answered "probably not," while adding that some use cases "are definitely better done on a service like ChatGPT." Reported by GeekWire, May 2025.

A month later, at Y Combinator in June 2025, Altman reframed ChatGPT not as a search replacement but as an agent: "For a long time ChatGPT was like a Google replacement. It's like a very junior employee that can work on something for a short period of time." That shift, from answering queries to completing tasks, is where OpenAI now aims, and it explains the pivot toward agents and an OpenAI browser rather than a frontal assault on the ten blue links.

Google, for its part, has treated the threat as existential from the start. In December 2024, Sundar Pichai told employees the company had to internalize the urgency of the moment.

"I think 2025 will be critical. I think it's really important we internalize the urgency of this moment, and need to move faster as a company." Sundar Pichai, Google CEO, December 2024, as reported by Search Engine Land.

By December 2025, the urgency ran both ways. After Google shipped Gemini 3 in November 2025 and topped several benchmarks, Altman issued an internal code red, ordering teams to prioritize ChatGPT's quality, speed, and reliability. The competitor Altman called ferocious had bitten back, and the search war had become a two-way arms race.

What ChatGPT search vs Google means for B2B marketers

For growth, demand, and SEO teams, the practical conclusion is not to pick a side. It is to optimize for a results surface where a generative answer sits on top and a shrinking set of clicks sits below, on both Google and ChatGPT. The tactics that move the needle in this environment overlap far more than the vendor marketing suggests, because, as the srsltid and Bing studies showed, the AI answer layer is largely reading from the same organic index you already compete in.

The near-term playbook comes down to a handful of moves:

• Win the top of the organic index first, because ChatGPT and AI Overviews both cite from positions one through twenty disproportionately
• Structure content as clean, extractable answers with clear headings, definitions, and tables that a model can lift verbatim
• Publish original data, benchmarks, and named-expert quotes, the assets models cite when they need authority
• Ship comprehensive schema and structured data so shopping and entity fan-outs can find you
• Track citations and brand mentions inside AI answers, not just rankings and clicks
• Build entity authority so your brand is associated with your category in model training and retrieval
• Keep Google Shopping and merchant feeds pristine, since ChatGPT pulls 83 percent of carousel products from them

This is the discipline now sold under three overlapping labels: AI optimization, answer engine optimization, and generative engine optimization. The names differ; the underlying work is the convergence of technical SEO, structured content, and authority signals aimed at being the cited source rather than the tenth blue link. Teams that still measure success purely in sessions and rankings are optimizing for a page that fewer users read to the bottom of. The ones pulling ahead treat search visibility as presence inside the answer, wherever that answer is generated.

Five predictions for AI search in 2026 and 2027

The next eighteen months will be defined by consolidation of the answer layer, not a single winner. Based on the trajectory of the data above, here are five specific calls for the rest of 2026 and into 2027.

1. AI Overviews will stabilize above 45 percent of tracked results by the end of 2026. The growth curve from 18.55 percent in Q3 2024 to the mid-40s in 2026 has slowed, not stopped, and Google will push coverage on commercial queries it can monetize while dialing it back on sensitive health and finance topics under regulatory pressure.
2. OpenAI will lean into browsing and agents, not a standalone index. Altman's junior employee framing signals that OpenAI wants to complete tasks, so expect the browser and agent products to absorb search rather than a Google-style crawler emerging to rival Bing.
3. Citation tracking replaces rank tracking as the primary AI-search KPI. With position-one CTR near 1.6 percent under an AI Overview, the industry will standardize on share-of-citation metrics, and the toolmakers who own that measurement will consolidate the category.
4. The Bing dependency stays load-bearing through 2027. Nothing in the 2025 studies suggests OpenAI is close to an independent index, so Microsoft's infrastructure remains the quiet backbone of ChatGPT search, keeping Bing strategically relevant despite its tiny consumer share.
5. Google keeps its volume crown but loses high-intent share. The 373-to-one volume gap will not close, yet ChatGPT will keep taking research and comparison sessions, the exact queries advertisers pay the most for, pressuring Google's revenue per query even as total queries hold.

The common thread is that search is fragmenting into a fast navigational layer that stays with Google and a deliberative answer layer that both companies are racing to own. Betting on one platform to win outright misreads the shape of the shift.

What to do Monday morning

The gap between reading this and acting on it is small, and the first moves are concrete. Start by measuring where you actually stand inside AI answers, because you cannot improve a surface you are not tracking. Then close the structured-content and authority gaps that keep you out of citations on both Google and ChatGPT.

A practical first week looks like this:

• Run your ten highest-value queries through ChatGPT Search and Google with AI Overviews on, and record who gets cited and who does not
• Audit your top commercial pages for extractable structure: clear H2 questions, direct-answer opening sentences, comparison tables, and definitions
• Verify your schema markup and merchant feeds are clean, since shopping fan-outs and entity retrieval depend on them
• Publish or refresh one original-data or expert-quote asset per priority topic, the kind of source models cite
• Switch at least one dashboard from rank-and-session tracking to citation-and-mention tracking
• Confirm you are indexed and ranking in the top twenty organically for every query you want an AI engine to cite

The strategic reframe is simpler than the tooling. ChatGPT Search did not dethrone Google, and Google did not stop the shift by shipping AI Overviews. Both moves accelerated the same outcome: a search experience where the answer is the product and the click is optional. Your job is to be the source that answer is built from. If you want a structured starting point, our AI visibility audit maps exactly which queries cite you today across both engines, and our generative engine optimization program is built to close the gaps it finds. The teams that treat AI search as a measurement and authority problem, not a channel to fear, are the ones compounding visibility while everyone else watches their click-through rates fall.

Priya Shah

AI Engineering Lead

Priya runs Skitrate's marketing automation and AI agent stack. Builds the workflows, schemas and data pipelines that move clients from manual ops to autonomous campaign management. Background in distributed systems before pivoting into ad-tech and martech in 2021. Holds patents on agentic campaign orchestration.

• Builds and operates the Skitrate AI automation stack (n8n + LangChain + Postgres + pgvector)
• Designed the agent workflows now running enrichment, scoring and reporting for 40+ clients
• Previously SRE-turned-platform-engineer at a Tier-1 ad-tech firm
• Speaker on AI in marketing operations at SaaStr and MarTech Conference

More posts by Priya Shah →