ChatGPT Now Owns 92% of AI Referral Traffic — And Google Just Made Its AI Overviews Even Stickier

New data from Previsible and Google's AI image generation launch in AI Overviews confirm the two-surface reality every service business needs to plan around right now.

Ido Cohen · Published 2026-07-16 · SEO & Search

Two pieces of AI search news landed this week that, read together, tell service businesses exactly where their next customer is being won or lost. First, Previsible published its largest AI Traffic Study yet — 6.77 million real LLM-driven sessions analyzed across 166 websites — and found that ChatGPT commands a staggering 92.4% of all trackable standalone AI referral traffic, growing 12.8x in just 19 months. Second, on July 14, Google announced it is adding AI image generation directly inside AI Overviews, powered by its new Nano Banana model, giving its already-dominant search surface yet another reason for users to never leave the results page. If you run a plumbing company, dental practice, law firm, HVAC business, or any other local service operation, these two developments define the terrain you are competing on right now.

What the Previsible Data Actually Says

The headline number is 92.4%, and it is as stark as it sounds. Previsible's third AI Traffic Study, published July 6 and covered immediately by Search Engine Land, analyzed 6.77 million LLM-driven sessions across 166 GA4 properties — covering SaaS, finance, legal, health, insurance, education, and SMB sites — from November 2024 through May 2026. According to the study, ChatGPT commands 92.4% of all trackable LLM referral traffic, growing 12.8x over 19 months with no sign of slowing.

To be specific about what "standalone AI referral traffic" means: this measures sessions that arrive from a user clicking a link inside ChatGPT, Claude, Gemini, Perplexity, or Copilot — the AI chat tools people use directly. It does not include traffic routed through Google's own AI Overviews or AI Mode, which operate on a different measurement model and almost certainly represent an even larger volume of AI-influenced visits. The 92.4% figure is a ChatGPT number within standalone LLM traffic specifically. Google's surfaces are a separate and larger category entirely.

Here is what happened to everyone else in the race:

Perplexity and Copilot were both sold to businesses as the next big AI search investment 12 months ago. Keep that in mind next time a vendor promises you must be present on a specific AI platform. The landscape moves faster than contracts do.

Why Ranking #1 on Google No Longer Guarantees AI Visibility

This is the number that should change how you think about SEO investment. According to Ahrefs' 2026 AI citation analysis, the share of top-10 Google organic pages appearing in AI Overview citations has fallen from 76% to 38% in just eight months. Read that again: you can be the first blue link in Google Search and still get skipped over by the AI Overview sitting above you. AI Overviews are now pulling from long-tail, forum, structured, and answer-specific pages rather than defaulting to whoever ranks number one.

This matches what the Semrush 2026 AI Visibility Index found after analyzing 126 million U.S. AI search prompts across ChatGPT, Gemini, Google AI Overviews, and Perplexity between January and April 2026. According to Semrush, being mentioned in an AI-generated answer does not necessarily mean a brand's own website is cited as the supporting source. On Gemini specifically, the overlap between brands mentioned in answers and domains actually cited can be as low as 30%.

For a plumber, an HVAC contractor, or a family law attorney, this distinction matters enormously. You can appear in a ChatGPT recommendation for "emergency plumber in Austin" without ChatGPT linking back to your website. Getting mentioned and getting cited are two different jobs.

The Semrush data also found that organizations fully integrating SEO and AI visibility into a unified workflow reported 81% success in increasing traffic or leads from AI platforms. Among organizations managing the two areas separately, only 36% reported the same result. Treating AI citation as a separate project from SEO is leaving money on the table.

What Google's AI Image Generation in AI Overviews Means for Service Businesses

On July 14, Google announced it is adding AI image generation directly inside AI Overviews in Search, powered by its latest Nano Banana model — the same image system Google has been expanding across Search and Chrome in 2026. According to Google Senior Engineering Director Brad Kellet, the feature turns a simple text prompt into a high-quality custom visual generated from scratch, and rolls out over the coming weeks in English for all regions that already support image creation in AI Mode.

This is part of a broader Google Images redesign timed to the service's 25th anniversary, which also introduces a personalized, real-time gallery homepage for signed-in users. But the AI Overviews update is the more strategically significant move.

Here is the problem it creates for service businesses: AI Overviews already answer many queries without requiring a click. Adding generated images gives Google's answer layer one more thing it can produce on its own — without linking out to a home services website, an interior designer, or a contractor who has photos of their actual work. Search Engine Land noted plainly that this update may have an impact on traffic to publishers and photographers, as it adds AI-generated content to AI Overviews and potentially discourages clicks. Search Engine Roundtable's Barry Schwartz put it even more bluntly: he expects this feature will lead to fewer clicks on links and fewer people visiting Google Images from publisher sites.

For service businesses, the practical implication is straightforward: every upgrade Google makes to what the AI Overview can deliver on its own is one less reason a user has to click through to your website. The counter-move is not to panic — it is to make sure your website is the source that the AI Overview wants to cite when it does answer a local service query.

The Two Surfaces That Actually Matter for Service Business Owners

If you run a service business and you are trying to figure out where to put your AI search effort, the Previsible data gives you a clear answer backed by 6.77 million sessions: prioritize Google's AI surfaces first, then ChatGPT as the leading standalone alternative.

Previsible's study explicitly states that AI discovery happening inside Google — through AI Overviews and AI Mode — almost certainly represents more AI-influenced traffic than every standalone LLM platform combined, even though it operates on a different measurement model. Google's surfaces are the bigger pool. ChatGPT is the dominant standalone platform, by a margin so large that optimizing for "AI visibility" without prioritizing ChatGPT specifically means optimizing for an abstraction.

What is worth watching: Claude grew 64x over 19 months and overtook Perplexity in March 2026, with particular strength among developers, technical buyers, and professional services. If your service business targets lawyers, financial advisors, architects, or other professionals, Claude is already an emerging discovery surface for your prospective referral network.

One more finding from the Previsible study that almost nobody is talking about: ChatGPT sends 28.8% of its referral traffic to internal search results pages — not to the specific service page that answers the query. According to the analysis, the model trusts your domain enough to send the visitor, but cannot always identify the right page, so it routes users to your site search box instead. For most service businesses, the website's internal search function is an afterthought. That has to change. A high-intent visitor from ChatGPT who lands on a blank search bar and gets zero relevant results is a lost lead that you will never see in your Google Analytics.

The Third-Party Source Problem No Service Business Has Fixed

Here is a structural challenge the Semrush data surfaces that is specific to service businesses: AI platforms do not only pull from your owned website to form their answers. According to the 2026 AI Visibility Index, AI platforms rely on customer reviews, community discussions, independent publishers, retailers, and industry sources to understand and recommend brands. Your Google Business Profile reviews, your Yelp ratings, your HomeAdvisor profile, local news coverage, and even Reddit threads about your category all feed into how AI systems describe and recommend you.

This is different from traditional SEO, where backlinks and on-page content were the primary inputs. AI systems are synthesizing your reputation from dozens of sources simultaneously. If your reviews on third-party platforms are thin, outdated, or contradicted by negative content on forums, the AI answer about your business will reflect that — even if your website is technically perfect.

The Ahrefs finding reinforces this: only 38% of AI Overview citations now come from top-10 organic pages. The other 62% come from long-tail content, forums, structured data, and answer-specific pages — exactly the kind of third-party content where your reviews and local directory profiles live.

What to Do This Week

If you have 3–5 hours this week, here is where to put them:

1. Audit your Google Business Profile and top three review platforms. AI systems pull heavily from third-party sources. Your GBP description, service categories, and reviews are source material for AI answers. Update anything inaccurate, and make sure your service descriptions match the exact language your customers use when searching.

2. Check whether your website is legible to AI crawlers. According to Previsible's data, 46% of ChatGPT bot visits begin in "reading mode" — a plain HTML version of your page with no images, CSS, or JavaScript. If your service pages rely entirely on JavaScript-rendered content, ChatGPT may be reading a blank page. Test your key service pages in a browser with JavaScript disabled and confirm the core text is visible.

3. Fix your internal site search. ChatGPT sends nearly 29% of its referral traffic to internal search pages. Go to your website right now and search for "plumber," "HVAC repair," "dental cleaning," or whatever your core service is. Does it return a useful result? Most service business websites return nothing or an error. That is a conversion leak on high-intent AI-referred traffic.

4. Publish one citation-worthy piece of content on your website this week. Not a blog post about "5 tips." A page that directly answers a specific local service question with real data, real specifics, and your first-hand expertise. According to Previsible's recommendations, citation-worthy evidence is the top lever for earning AI mention. Think: "How much does it cost to replace a water heater in [City] in 2026" with actual price ranges from your own jobs.

5. Open Google Search Console and check your AI Impressions report. Google launched this report in June 2026. It shows how often your pages appear in AI Overview results, by page and query. If you have not looked at it yet, you are flying blind on the surface that is already your biggest AI traffic source.

Frequently Asked Questions

What is AI referral traffic and why does it matter for my service business?

AI referral traffic refers to website visits that originate from a user clicking a link inside an AI platform like ChatGPT, Claude, or Gemini — as opposed to traditional search clicks from Google's blue links. It matters because users arriving via AI recommendations come with highly refined intent; they have already asked an AI assistant a specific question and received a recommendation. According to research cited by JumpFly from Reuters and Adobe, shoppers arriving via AI sources convert at rates approximately 42% higher than average web visitors and spend 48% more time on-site.

Should I create separate content to rank in ChatGPT versus Google's AI Overviews?

No. The Previsible study and Google's own guidance both point to the same conclusion: the content signals that earn Google AI Overview citations are largely the same signals that earn ChatGPT citations. Build clear, structured, factually specific service pages with direct answers to questions your customers ask, ensure your website is technically accessible to crawlers, and maintain consistent information across third-party platforms. You do not need two strategies — you need one strong content foundation.

How does Google's new AI image generation in AI Overviews affect service businesses?

The main risk is reduced click-through on visual queries. If someone searches "what does a good bathroom remodel look like" and Google generates an image on the spot, they have less reason to visit your before-and-after portfolio. The offset is that service businesses with strong local signals and citation-worthy content will be the ones AI Overviews recommend alongside those generated visuals. Focus on being the cited business, not the clicked image.

Why did Perplexity and Copilot collapse as AI referral traffic sources?

Perplexity dropped 61% from its early-2025 peak largely because it shifted toward keeping users inside its own browser and agent tools rather than referring traffic outward. Copilot collapsed 96% from its peak, likely due to product changes and Microsoft's strategic repositioning. The lesson for service businesses is to be skeptical of any vendor telling you to invest heavily in a specific AI platform other than ChatGPT or Google's surfaces — the landscape reshuffles faster than most marketing contracts.

What is GEO and do I need to hire someone to do it?

GEO stands for Generative Engine Optimization — the practice of structuring your content and online presence so that AI systems like ChatGPT and Google's AI Overviews cite your business when answering relevant queries. It is not a separate discipline from SEO; according to Semrush, organizations that integrate SEO and AI visibility into a unified workflow see 81% success rates, versus 36% for those treating them separately. You do not need a separate agency for GEO — you need your existing SEO or marketing effort to account for AI citation signals, including third-party reviews, structured content, and technical accessibility.

Sources: