The Verification Funnel: How AI Search Is Rewiring Auto Repair Discovery

Two things we can say for certain

Start with what the evidence actually nails down, because the temptation in a story like this is to round every number toward the most dramatic version of itself. Two patterns here are established, meaning they come from methodologically transparent, non-vendor-conflicted sources and were independently spot-checked against the original page: AI chatbot adoption is climbing fast, and Google's own results pages are increasingly answering questions before anyone clicks through. Pew's 2026 survey puts general chatbot use at 49 percent of US adults, nearly half again what it was two years earlier. SparkToro's clickstream analysis puts the zero-click share of US Google searches at just over 68 percent for early 2026, a jump of more than seven points in two years, with AI Overviews responsible for a large piece of that shift.

What is just as important, and easy to lose in the excitement around either number, is that Bain's separate survey of 1,500 consumers found that 56 percent still default to a traditional search engine over a chatbot, compared with only 16 percent who default to AI. Chatbot adoption and search-engine dominance are both true at once. The headline is not \"AI has replaced search,\" it is that a meaningful and growing slice of every category's demand now resolves without a click, whether that resolution happens inside an AI Overview, a chatbot's answer, or the map pack itself.

Chatbot adoption and search-engine dominance are both true at once. The headline is not that AI has replaced search.

The citation gap, everywhere researchers have actually measured it

The more provocative finding in this research pass is not about how many people use a chatbot, it is about what happens when they do. SOCi's 2026 Local Visibility Index, analyzing roughly 350,000 locations across 2,751 multi-location retail, restaurant, and financial-services brands, found that ChatGPT recommended a given location just 1.2 percent of the time, Perplexity 7.4 percent, and Gemini 11 percent, against Google's local 3-pack surfacing the same businesses 35.9 percent of the time. The report frames AI visibility as three to thirty times harder to win than a map-pack ranking, depending on the engine.

A second, single-category study reinforces the same shape of gap without depending on SOCi's methodology. Local Falcon analyzed 189,905 ChatGPT results against Google's local results for the same restaurant queries and found 83 percent of restaurants completely invisible to ChatGPT on local dining questions, versus only 14 percent invisible on Google; only 17 percent of restaurants ever appeared in a ChatGPT answer at all. Both studies are produced by vendors that sell visibility or rank-tracking software, which is exactly why we treat them as contested rather than established, and neither one includes automotive in its sample. What survives the scrutiny is a consistent, cross-vendor, cross-category signal: wherever anyone has actually measured it, ranking well on Google and being named by an AI answer engine are two different games, and the second one is much harder to win.

Ranking well on Google and being named by an AI answer engine are two different games, and the second one is much harder to win.

Why auto repair is a strong candidate, not a confirmed case

This is the point where the study has to be most careful, because it is also the point where a lazier version of this piece would quietly swap \"restaurants and retail\" for \"auto repair\" and let the 1-to-11-percent-versus-36-percent figures do work they were never measured to do. They were not. No methodologically transparent study located for this research measured an AI-citation rate for repair shops specifically. Every vendor claim of an \"auto repair AI visibility score\" that surfaced during this research either omitted its methodology, excluded automotive from its sample outright, or rested on a sample too small to generalize from.

What can be said is that auto repair carries several of the same structural features that appear to predict a wide citation gap in the categories that have been measured: heavy dependence on a single Google Business Profile listing, a buying decision that is almost entirely mediated by reviews rather than the shop's own marketing, and, per AAA's 2016 survey, a baseline of consumer distrust so high that two-thirds of drivers said they distrusted the category in general even while 64 percent had one specific shop they personally trusted. That AAA figure is a decade old and unreplicated at scale since, so it cannot be cited as current-state evidence, only as a historical marker of how deep the trust deficit has run in this category. Combine that history with J.D. Power's far more current 2026 finding, drawn from 51,228 vehicle owners, that independent aftermarket shops already out-compete dealership service on speed, a documented driver of which shop a customer chooses (62 percent of aftermarket visits finish in under an hour, against 1.61 hours for mass-market dealership service and 2.46 hours for premium-brand dealership service), and you have a category where the competitive stakes of who gets recommended first are unusually high. That combination of dynamics is why this study treats auto repair as a strong candidate for the citation gap seen elsewhere, while being explicit that the gap itself has not been measured for the category yet.

What decides whether an engine ever mentions your shop

If the citation gap does extend into auto repair the way it appears to in retail and restaurants, the next question is what actually earns a citation once an engine is deciding what to say. Yext's analysis of 6.8 million AI citations, drawn from 1.6 million queries per model across ChatGPT, Gemini, and Perplexity, found that 86 percent of citations trace back to sources the business itself controls: 44 percent to the business's own website, 42 percent to directory and listing data, and 8 percent to reviews and social content. Automotive was not part of Yext's sample either, so this is one more directionally useful but unconfirmed data point rather than a repair-specific finding.

Taken at face value, though, it argues against the instinct to treat AI visibility as some new, opaque discipline you have no control over. If close to nine in ten citations across the categories that have been studied come from information the business already publishes or maintains, the practical lever is closer to home than it might feel: an accurate, complete, consistently updated Google Business Profile and website, the same assets that have mattered to local search for years, appear to still be doing most of the work in an AI-mediated world. Semrush's separate 2026 AI Visibility Index, built from 126 million US AI search prompts across 22 industries, found AI-referred traffic to US retail sites up 1,324 percent between October 2024 and May 2026, and travel sites up 2,215 percent over the same window, evidence that the audience arriving through this channel is real and growing quickly in the categories where it has been tracked, even as 45 percent of the marketing leaders surveyed in that same report said they cannot yet accurately measure their own AI-answer visibility.

The zero-click funnel, one search at a time

Zero-click is not an abstraction when the search is \"transmission shop near me open now\" or \"why is my check engine light on.\" It means the answer the buyer sees, whether it is a Google-generated summary, an AI Overview, or a chatbot's response, may be the entire interaction that decides whether your shop's phone rings. SparkToro's 68.01 percent zero-click figure and the finding that AI Overviews cut click-through by nearly 60 percent when present describe the whole US search population, not auto-repair queries specifically; no study located for this research isolates how much of that zero-click rate is attributable to AI Overviews for service or \"near me\" searches as opposed to general informational queries, and that gap in the data should be acknowledged rather than papered over.

What can be said is directional and still useful: a category that already depends this heavily on being seen inside someone else's shortlist, whether that shortlist is a map pack or a chatbot's answer, has more to lose from an invisible or incomplete online presence than a category where buyers routinely click through several options before deciding. If a meaningful share of \"who do I call\" decisions for a car problem now resolve before a website is ever visited, the business's information has to be right, complete, and citeable wherever it lives, not just optimized for a click that may never come.

The gap is addressable, and that part is proven

The most encouraging finding in this research is also the most rigorously sourced one. A peer-reviewed paper accepted to KDD 2024, \"GEO: Generative Engine Optimization,\" by Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan, and Deshpande, introduced a benchmark called GEO-bench and demonstrated that applying the framework's citation-optimization methods increased measured content visibility in generative-engine responses by up to 40 percent. This is not a vendor claim; it is an academic, peer-reviewed result, and it establishes something the rest of this study depends on: the AI-citation gap, wherever it exists, is not a fixed property of a business or a category, it is addressable through deliberate work on how information is structured and presented.

That matters because this study cannot promise that fixing a shop's website or Google Business Profile will close a gap that has not even been measured for the category yet. What it can say, grounded in a source that has cleared academic peer review rather than a vendor's product marketing, is that the underlying mechanism, engines choosing what to cite based on how information is presented to them, responds to deliberate optimization. That is a meaningfully different, and more useful, claim than \"AI visibility is out of your control.\"

What this means for the next twelve months

Put the pieces together and the picture is neither alarmist nor complacent. AI chatbot use is genuinely rising (49 percent of US adults, established), traditional search is not disappearing (56 percent still default to it, established), and Google's results pages are answering more questions without a click than they did two years ago (68.01 percent, established). Separately, in every category anyone has rigorously measured, being visible in Google's map pack has stopped guaranteeing visibility inside an AI answer, sometimes by a wide margin. Auto repair has not been measured directly, but it shares the exact structural features, review-dependence, single-listing reliance, and a buying decision built on distrust of the category, that appear to predict that gap elsewhere.

The practical takeaway for a shop owner is not to chase a number that does not exist yet for this category, but to treat the underlying asset work as worth doing regardless of exactly how wide the auto-repair gap turns out to be: a complete, accurate, consistently maintained Google Business Profile and website, the same assets Yext's cross-category data ties to 86 percent of citations, are the foundation either way. What is missing, and what this study cannot supply on its own, is a repair-specific measurement of where a given shop's own audience is actually looking today, across the map pack, the review platforms, and the AI engines all at once.