Are Google AI Overviews Killing Your Contractor Leads?

Not your leads — your blog traffic. AI Overviews significantly reduce clicks on informational searches, but the commercial and local queries that actually produce booked jobs are far less affected. A homeowner still has to choose a company and call it, and an AI summary cannot do the work.

That distinction is the entire article. If your traffic is down and your phone is fine, you have a content problem, not a business problem. If your phone is genuinely quieter, AI Overviews are probably not the main cause and you should look at your local pack position, your review velocity, seasonality, and your paid channels before blaming Google's summaries.

Enough to matter on informational queries. Ahrefs' 2025 study of 300,000 keywords found a significant CTR decline for top-ranking organic results when an overview appeared, and Search Engine Land's reporting found desktop CTR roughly halved while mobile clicks fell about a third.

The picture is not uniformly bleak, though, and the honest version includes the counter-evidence. Seer Interactive's 2026 update found the steep 2025 decline did not simply continue into 2026 — impressions more than doubled while clicks stayed flat, and organic CTR on queries without an overview actually improved, climbing from around 2.8% in January 2025 to roughly 3.8% by February 2026. In other words: the queries Google doesn't summarize became more valuable, not less.

The question-shaped ones. Seer's 2026 analysis of roughly 49,000 tracked queries shows AI Overview prevalence varies enormously by query type — comparison queries triggered one 95.4% of the time, while single-word queries triggered one just 27.3% of the time.

Because one has an answer and the other has an action. An AI summary can fully satisfy 'what does a roof cost' — the searcher wanted a number and got one. It cannot satisfy 'plumber near me', because the searcher wants a person at their house, and that requires clicking, calling, or booking.

This is the structural reason local service businesses are better insulated from AI summarization than publishers. Your money queries end in a transaction that only a human business can complete. What you lose is the top-of-funnel research audience you were capturing with cost guides — real value, but not the same thing as losing leads.

• Resilient: '[trade] near me', '[trade] in [city]', 'emergency [trade]', branded searches, 'book/schedule [service]'.
• Exposed: 'how much does X cost', 'X vs Y', 'how long does X take', 'do I need a permit for X', 'best X for Y'.
• Mixed: 'best [trade] in [city]' — an overview usually appears, but it frequently names specific local businesses, which is an opportunity rather than a loss if the named business is you.

Compare clicks against impressions by query type in Search Console. The AI Overview signature is impressions holding steady or rising while clicks fall on informational queries, with commercial and branded queries roughly unchanged. If clicks and impressions both collapse, something else is wrong.

A 20-minute Search Console diagnostic

• Open Performance and compare the last 6 months against the same period last year. Note total clicks, impressions, and average CTR.
• Filter queries containing 'how', 'what', 'cost', 'vs', and 'best'. This is your exposed bucket — check CTR here specifically.
• Filter queries containing 'near me' and your city names. This is your resilient bucket. Compare its trend against the exposed bucket.
• Filter branded queries (your company name). If branded impressions are rising, AI-assisted discovery may be sending people to look you up by name — that's a win hiding in your data.
• Check the pages report: is the decline concentrated in blog and guide URLs while service and location pages hold? That pattern is the AI Overview signature.

Stop optimizing informational content for clicks and start optimizing it for citation — while defending the commercial pages that still convert. The goal for a cost guide is no longer a session; it's being the source the summary quotes and the name the reader remembers.

For exposed informational pages

• Rewrite answer-first: question heading, complete answer in 40-60 words, then the detail. Quotable passages get cited; meandering ones get skipped.
• Add specifics an AI can't paraphrase away — local price ranges, permit realities in your county, real timelines, named materials and brands.
• Put your brand and locality inside the answer text itself, so a citation carries your name rather than a generic fact.
• Add a genuine reason to click beyond the answer: a calculator, a checklist, a local cost table, photos of your actual work.
• Stop publishing generic national-average content. It is exactly what a model can synthesize without you.

For commercial and local pages

• Defend the local pack: Google Business Profile completeness, review velocity, correct categories, service-area accuracy.
• Keep city and service pages substantive — real local proof, not templated copies with the city name swapped.
• Make booking frictionless. A larger share of your remaining clicks now arrives further down the funnel and ready to act.
• Track calls and form fills, not sessions. Session count is now a worse proxy for business health than it was two years ago.

It raises the value of channels that don't depend on the SERP. If a slice of your organic research traffic is now being answered before the click, paid social and paid search become a larger share of predictable lead flow — not because ads got better, but because one substitute got less reliable.

The honest caveat: this is a reason to keep demand capture funded, not a reason to panic-buy more of it. Ad spend that wasn't profitable before AI Overviews is not profitable now. Run the same cost-per-booked-job math you always should, and let that decide the number. The strategic shift is in reliability and mix, not in the arithmetic.

Yes — but for different reasons than three years ago. Content now earns citation, trust, and sales-cycle leverage more than it earns sessions. A cost guide that gets quoted by an assistant and read by a homeowner before your estimate call is doing real work, even if analytics never records the visit.

• Citation value: being the source an AI quotes puts your name in front of buyers you'll never see in analytics.
• Sales enablement: linking a prospect to your own permit or process explainer shortens estimate calls and pre-frames objections.
• Entity building: depth across your trade and your service area is what makes models confident enough to name you specifically.
• Conversion support: a homeowner comparing three quotes reads all three companies' content. Yours should be the clearest.
• The thing to stop doing is volume-for-volume's-sake publishing on national-average topics with nothing local or specific in them.