How we measure whether AI search is actually working for you
AI search is new enough that a lot of what gets reported about it is noise. When ChatGPT, Google's AI Overviews, Perplexity, and Gemini answer a question, being cited or recommended is the goal — but measuring that honestly is harder than measuring a blue-link ranking, and it is easy to dress up vanity numbers as results. The metric that matters most is citation presence: when a buyer asks an assistant something like "who are the best family law attorneys in Alexandria" or "roof repair companies near Old Town," does your business get named, and in what position? We run these real-buyer prompts on a schedule, across the major assistants, and log whether you appear, who appears alongside you, and whether the AI links to your site as a source. That is the closest thing to a ranking in this world, and it is what we hold ourselves to.
The second metric is referral traffic and conversions from AI sources. Assistants increasingly send clicks, and those visitors tend to arrive further down the buying path than a cold searcher. We watch how much of your traffic now comes from AI referrers and — more importantly — whether those sessions turn into calls and form fills, because a citation that drives no business is a trophy, not a result. The third is the accuracy of what the AI says about you. If an assistant is confidently telling people you are closed on Saturdays when you are open, or quoting an old service area, that is a measurable problem we fix by correcting the sources the models pull from. We check what the machines believe about your Alexandria business and treat wrong answers as bugs.
- What we ignore: raw impression counts nobody acts on, "AI visibility scores" from tools with no methodology you can inspect, being mentioned for searches no real buyer runs, and any metric that goes up while your calls stay flat.
None of this works without the plumbing underneath it. Getting cited depends on clean, well-structured content that answers real questions directly, plus schema and entity signals that tell the models exactly what your business is and where it operates. That is why our AI search work is inseparable from the content and technical foundation — you cannot measure your way into an answer you never gave the models a reason to include, and the businesses that get cited are the ones that published the clearest answer first.
The honest summary is this: AI search measurement is still maturing, and we would rather show you a handful of prompts where you are named and a real referral trend than a dashboard full of numbers designed to look impressive. If a metric cannot be tied back to a buyer finding you and reaching out, we do not put it in your report. That discipline is what keeps this from becoming the kind of measurement theater the last decade of digital marketing was full of.