post
What Is AI Search Traffic, and Why Home Service Businesses Should Start Tracking It
A new channel that didn't exist a couple of years ago
For most of the history of local service marketing, the source list looked the same: Google organic, Google Ads, Facebook, referrals, direct. Somewhere in the last two years, a new one showed up in analytics reports that didn't fit neatly into any of those buckets — visits and leads arriving from ChatGPT, Perplexity, Gemini, Claude, and similar AI assistants.
People are asking these tools questions like "who's a good plumber near me" or "which HVAC companies in my area do emergency repairs," and the assistant is answering with a short list — sometimes with a direct link. When someone clicks through and fills out a form, that's a real lead. Whether your systems can tell you it happened is a different question, and for most small businesses right now, the honest answer is no.
How big is this, really?
It's easy to either dismiss this as a fad or overstate how much of your business it already represents, so it's worth looking at actual numbers instead of vibes. According to Similarweb's 2026 analysis of generative AI referral traffic, AI platforms still represent a small slice of total web traffic overall — a fraction of a percent industry-wide — but that slice has grown enormously fast, and the visitors it sends convert at a noticeably higher rate than average organic traffic. Similarweb's research found AI-referred visitors converting at roughly double the rate of typical organic search visitors across the sites it analyzed, which tracks with the intuition that someone who asked an AI assistant a specific question and clicked through arrives already further along in deciding what they need.
The mix of which assistant sends the traffic is shifting quickly too. ChatGPT remains the dominant source of AI referral traffic by a wide margin, but its share of that traffic has been slipping even as its absolute volume keeps growing, because assistants like Gemini and Claude have been growing even faster off a smaller base. None of this is stable yet — the market is genuinely still forming — which is exactly why it's worth tracking as its own channel now rather than waiting until it's obviously large enough to matter. By the time it's obvious, competitors who started paying attention early will already have a head start on showing up in these answers.
Why this is worth tracking separately, not lumping into "organic"
If a lead from an AI assistant just gets bucketed into "organic search" or "direct" — which is what happens by default in most analytics setups — you lose the ability to answer an increasingly relevant question: is my business actually showing up when people ask AI tools for a recommendation? That's a distinct problem from classic SEO, and it responds to different things. We cover the broader mechanics of connecting a lead back to its real source, not just wherever it happens to land in a generic bucket, in our guide to marketing attribution.
Tracking AI search as its own source lets you see, concretely, whether that channel is producing any inquiries at all yet, and whether it's growing month over month. For most home service businesses right now it's a small slice of total leads — but small and growing is exactly the kind of trend worth watching before it becomes a channel you've ignored for two years while competitors figured it out.
How this typically gets detected
Attribution for AI-referred traffic generally works by looking at where the visit came from — referrer data and known patterns associated with these assistants — and tagging the resulting lead accordingly, the same way a Google Ads click gets tagged with its campaign. The lead shows up in your inbox like any other, just labeled by source, so your team follows up exactly the same way regardless of channel. What matters is that the label exists at all — most out-of-the-box website analytics and basic contact-form tools simply don't have a category for this yet, so the traffic either gets miscategorized as direct or disappears into an "other" bucket that nobody ever looks at.
What actually helps you show up in AI answers
This is where the practice usually called GEO — Generative Engine Optimization — comes in: the practice of making it easier for AI assistants to find, understand, and cite your business when someone asks a relevant question. It's still a moving target, but a few things consistently help.
Clear, accurate, plainly-written information about your services and service area. AI crawlers, like traditional search crawlers, do better with a well-structured page that states plainly "we do emergency water heater repair in [service area]" than with a page built entirely around flashy visuals and marketing language that a crawler has to work to parse.
Consistent listings and reviews across the places these tools tend to pull from. The same NAP (name, address, phone) consistency that's mattered for local SEO for over a decade still matters here, because inconsistent business information across directories makes it harder for any system — human or AI — to confidently confirm who you are and where you operate.
A published llms.txt file. This is the newest piece, and probably the least familiar one. Proposed in September 2024 by Jeremy Howard and maintained at llmstxt.org, llms.txt is a simple markdown file placed at the root of a website that gives AI crawlers a clean, structured summary of what a site is and what's on it — conceptually similar to how a robots.txt file or an XML sitemap helps traditional search engines, but written for language models with limited context windows instead of for crawlers indexing every page. As Search Engine Land explained when the standard was introduced, the core problem it solves is that a modern website's HTML — full of navigation menus, scripts, and layout markup — is inefficient and imprecise for an AI system to parse compared to a short, direct summary written specifically for that purpose. It's worth being clear-eyed that adoption is still uneven — not every AI platform has confirmed it uses llms.txt files today — but publishing one costs almost nothing and gives crawlers that do respect it a much cleaner signal than they'd otherwise get.
Actually answering the question, directly. Content written as a direct answer to a specific, plainly-phrased question — "how much does a water heater replacement cost," "how do I know if my AC compressor is failing" — tends to get pulled into AI-generated answers more often than content written primarily to rank in traditional search results. The two goals overlap but aren't identical, and increasingly it's worth writing with both in mind.
How to check whether you're already showing up
Before investing time in GEO specifically, it's worth spending fifteen minutes checking where you currently stand. Open ChatGPT, Perplexity, and Gemini separately and ask each one a question a real customer might ask — "who's a good [your trade] in [your city]," or a more specific version like "which companies in [your city] do emergency water heater repair." Note whether your business comes up at all, and if it does, note whether the information given about you is accurate — correct phone number, correct service area, correct services offered. It's common to discover that an assistant is citing outdated information from an old directory listing or a review site you haven't touched in years, which is itself useful to know, since it points to a specific listing worth updating rather than a vague sense that "AI search" is a problem to solve.
It's also worth repeating this check periodically rather than once. Because the underlying models and the sources they draw from keep changing, an answer that didn't mention you six months ago might now, and vice versa — treating this as a one-time audit rather than an ongoing, low-effort check is a common way businesses miss early signals of a channel worth paying attention to.
What this means for your marketing this year
None of this replaces the fundamentals — a fast website, accurate business listings, real reviews, and ads that reach people actively searching for your service. AI search traffic is still, for most home service businesses, a small percentage of total leads. But it's one of the only channels growing from a near-zero base at this speed, and the businesses that start showing up in these answers now — by publishing clear, accurate, well-structured information and giving AI crawlers a straightforward way to understand what they do — have a real head start over the ones who wait until it's an obvious, unavoidable line item in every marketing report.
If you want to see how AI search fits alongside every other channel in a single attribution report instead of disappearing into "other," Local Lead Feed's AI intelligence feature tracks it automatically, right alongside Google Ads, organic search, referrals, and everything else — see the full feature breakdown or check out how it ties into marketing attribution more broadly.