How we measure whether AI search is actually working
AI search is new enough that the metrics around it are a mess, and plenty of agencies will hand you a screenshot that looks impressive and means nothing. Since this is where a growing share of Patrick County searches now start — someone asking ChatGPT or Google's AI overview for a contractor instead of scrolling links — it is worth being precise about what we track and what we deliberately ignore.
The measurement that matters is simple to state and harder to earn: does the AI name your business when a real customer asks a real question? So we test the exact prompts your customers use. "Who is a good HVAC company near Stuart, Virginia?" "Best plumber in Patrick County." "Reliable roofer close to Meadows of Dan." We run those across the tools people actually use, log whether you get named, and watch that answer improve over time as your presence and consistency strengthen. Getting mentioned when it counts is the whole game.
We also watch a couple of signals underneath that. One is whether the AI describes you accurately — right services, right service area, right specialty — because being named wrong is barely better than not being named. The other is referral traffic that arrives from AI assistants and AI overviews, which shows up in your analytics as a distinct source and tells us the mentions are turning into actual visits and not just theoretical wins.
Here is what we do not chase. We ignore raw "mention volume" pulled from prompts no customer would ever type, because it is trivial to inflate and it does not ring your phone. We ignore rankings on made-up "AI visibility scores" that no buyer sees. And we do not celebrate being cited in an answer to a question that has nothing to do with hiring you. A mention in a generic listicle about the region is not a lead; a mention in the answer to "who should I call" is.
A word of honesty about the ceiling. No one controls what a model outputs, and there is no dashboard that reports your exact standing the way Google Search Console reports keyword rankings — the tools are not built for it yet. So we measure with repeated, structured testing rather than a single number, and we are upfront about the uncertainty. What works in your favor in a market like Stuart is scarcity: there is very little clean, consistent information out there for the AI to draw on, so a business that supplies it clearly tends to become the default answer fast.
You get a plain report showing which questions now name you and which still do not, so you can see the work paying off in the terms that matter. If you want to see how this connects to the rest of your search presence, our SEO work feeds the same foundation the AI reads from.