Do AI Overviews Reduce HVAC Leads? The Honest Data

What the correlational studies found

Ahrefs' April 2025 study compared 150,000 keywords that trigger an AI Overview against 150,000 informational keywords that don't, using aggregated Google Search Console desktop data. Position-1 CTR on AI Overview keywords fell from 0.073 (March 2024, pre-rollout) to 0.026 (March 2025). After controlling for the general CTR decline on informational queries, Ahrefs attributed a 34.5% CTR reduction to the AI Overview itself.

Ahrefs re-ran the same methodology in February 2026 with December 2025 data. The attributed reduction had grown to 58% at position 1. Their per-position table shows the damage tapering, not disappearing, down the page:

Source: Ahrefs, February 2026. Aggregated GSC desktop CTR, December 2023 vs December 2025.

One methodological caveat we'd be dishonest to skip: both Ahrefs studies use desktop CTR data, while Ahrefs' own keyword research found AI Overview queries skew heavily mobile — 81% of their traffic (Ahrefs). The true mobile picture could be better or worse; the published data simply doesn't show it.

Pew Research Center's July 2025 analysis took a different route: real browsing data from 900 US adults covering 68,879 Google searches in March 2025. Users who saw an AI summary clicked a traditional result on 8% of visits, versus 15% without one — roughly half. Only 1% of visits produced a click on a source cited inside the summary. And users were more likely to end their browsing session entirely after an AI-summary page (26% vs 16%).

What the randomized experiment added

Correlational studies can't prove cause. The first study that can: a randomized field experiment by researchers at the Indian School of Business and Carnegie Mellon (reported by Search Engine Journal, April 2026). A Chrome extension randomly hid AI Overviews for some of 1,065 US participants over two weeks in early 2026. Results: AI Overviews appeared on 42% of queries, and removing them raised outbound clicks from 0.38 to 0.61 per search — a 38% reduction in outbound organic clicks attributable to the feature. Zero-click searches rose from 54% to 72% when AI Overviews were present. Caveat we'd want stated about our own work too: it's an SSRN working paper, not yet peer-reviewed, and the sample is active desktop Chrome users.

The honest summary of all three: the click loss is real, replicated, and growing. Ahrefs' update also compiles corroborating measurements — Seer Interactive at 49.4–65.2%, Kevin Indig at over 50%, Authoritas at 47.5% (as listed by Ahrefs). For a breakdown of how these summaries behave specifically on HVAC searches, see our guide to AI Overviews for HVAC companies.

Where AI Overviews actually trigger

Ahrefs' analysis of 150,000 AI Overview keywords found that 99.2% of keywords that trigger an AI Overview are informational in intent. Commercial keywords trigger one just 5.8% of the time; transactional keywords, 4.0%. The median AI Overview keyword is four words long, versus two for non-AIO keywords — these are "why is my furnace short cycling" searches, not "furnace repair [city]" searches.

Pew's browsing data shows the same shape from the user side: only 8% of one- and two-word searches produced an AI summary, against 53% of searches with ten or more words, and 60% of searches phrased as questions (Pew Research Center). The searcher typing a short, urgent, local query — the searcher closest to booking — is the one least likely to see an AI answer between them and your listing.

Worth adding: the local pack predates AI Overviews as a click-absorber. Ahrefs' update names Local Pack among the long line of SERP features that "resolve a user's query without leaving Google" (Ahrefs). If your leads come primarily through map-pack visibility and calls, zero-click search has been your operating environment for years — AI Overviews extended it to the informational layer above you.

The 68% local claim — and why we discount it

You will also find much bigger numbers in circulation. Contractor Magazine cites Whitespark Q2 2025 research claiming AI Overviews appear in 68% of local business searches. That figure sits far above what the intent-level datasets show for commercial and transactional queries, and the underlying methodology isn't visible in the article. We flag it rather than hide it — that's the deal on this site — but when vendor percentages and 300,000-keyword datasets disagree by an order of magnitude, we weight the dataset with the published methodology. Definitions matter here too: "local business searches" can include informational research queries, which would reconcile part of the gap.

Google's position: stable clicks, higher quality

Google's head of Search, Liz Reid, wrote in August 2025 that "total organic click volume from Google Search to websites has been relatively stable year-over-year," and that average click quality has increased — defining "quality clicks" as visits where users don't quickly bounce back to the results. Her argument: an AI answer satisfies the quick lookup, and the users who do click are further along and more engaged. For a business that sells service calls rather than pageviews, that mechanism — fewer but warmer visits — would genuinely be fine if it held.

The problem is verification. As Search Engine Journal documented, Reid has made the same "bounce clicks" argument across three public appearances without releasing supporting data, and Search Console still gives site owners no way to separate AI Overview clicks and impressions from the rest. Both things can be true at once: Google's aggregate click volume can be stable while individual top-ranking pages lose half their CTR — aggregates hide redistribution. We treat Google's claim as a claim, not a finding, and we'd say exactly the same about any agency's unverifiable number, including ours.

The strongest honest point: nobody measured leads

Every dataset in this article measures clicks, impressions, or session behavior. Not calls. Not booked jobs. Not revenue. The 58% figure is position-1 CTR on a keyword set that is overwhelmingly informational; the 38% experiment measured outbound clicks per search across all query types; Pew measured click rates on a browsing panel. A missing click on "how often should I service my AC" is not a lost service call — most of those readers were never hiring anyone this month. The chain from informational click to HVAC lead is long and leaky, and no published study has yet traced AI Overviews' effect through to contractor lead volume. Anyone who tells you "AI Overviews cost you X% of your leads" is deriving a number no study contains.

[OWNER: once client or portfolio GSC + call-tracking data exists, publish our own before/after AIO-exposure vs lead-volume chart here — this is currently the missing evidence in the entire industry conversation.]

There is also real evidence the demand didn't vanish — it moved. Scorpion's 2026 national study of 2,000 US homeowners found 22% now use AI tools such as ChatGPT to research or find recommendations when they need a contractor; Contractor Magazine's coverage adds BrightLocal survey data showing consumer use of AI tools for finding local services jumping from 6% to 45% in a year. Fewer blue-link clicks doesn't mean fewer homeowners hiring — it means the path changed. We map that path in detail in how HVAC customers find contractors in AI search.

Read the table honestly and the picture is asymmetric: your blog and guide traffic is heavily exposed; your booking-intent queries are lightly exposed — on Google. The caveat in row three is the one to watch. The threat to commercial HVAC queries isn't primarily an AI Overview squatting on the SERP; it's the homeowner who never opened Google, which is a different problem with different economics — and the subject of our AI Overviews for HVAC breakdown and the behavior-shift data linked above.

1. Separate your traffic mentally into "informational" and "booking-intent" before reacting. The studies justify concern about the first bucket only. If your lead reports conflate blog sessions with service-page calls, split them before you judge any channel.
2. Expect informational content to earn fewer clicks per ranking — and judge it on different metrics. A cost guide that ranks #1 under an AI Overview sits in the −58% CTR class (Ahrefs). It still does work: AI summaries cite sources — 88% of the summaries Pew examined cited three or more — and being the cited source is the new version of being the ranking page. That's a visibility outcome you measure, not a guarantee anyone can sell you.
3. Protect the surfaces where booking happens — your service pages, local pack presence, and review profiles — because that's where the 4–6% trigger rates say the click economics remain closest to normal.
4. Check your actual exposure before spending a dollar. Ahrefs' Site Explorer has an AI Overview SERP-feature filter that shows which of your rankings sit under one (Ahrefs). Ten minutes of looking beats any agency's generic slide deck, including ours.
5. Treat AI answers as a surface to be counted in, measured monthly. That is the premise of our AI-search visibility program: baseline which engines mention you, work the evidence-backed inputs, re-measure, report honestly — no "we'll get you into ChatGPT" guarantees, because nobody controls these engines and the buyer who's been burned before is right to distrust anyone claiming otherwise. The same content discipline that survives AI Overviews is what durable HVAC SEO has always required.

This page deliberately stops short of the how: getting recommended by ChatGPT and ranking inside AI Overviews are their own playbooks, covered in their own guides.