We are measuring AI visibility wrong, and the number looks fine
Every marketing team I speak to has added AI visibility to the reporting pack in the last six months. Good, it is the right thing to track. More than 900 million people use ChatGPT every week, according to OpenAI, and the recommendations it hands out are real commercial outcomes.
Most teams in multi-location & franchise brands though, are tracking a number that cannot tell them much that is useful.
The scores on the market are reporting at brand level. One figure for how often ChatGPT or Gemini names you when a customer asks for a recommendation. Yes its clean, simple, boardroom-ready, and for any brand with more than one location, close to meaningless.
Assistants do not answer nationally. Ask where to eat in Greenside and the model assembles an answer from what it can verify about businesses near Greenside. Ask in Umhlanga and you get a different answer, built from different data, judged against a different set of rivals. There is no national AI result to rank in. There are as many results as the brand has trading areas.
Which means a 200-store brand reporting one AI visibility score is reporting an average. And averages are where the problem hides.
The data is always the same, group level looks acceptable but underneath, a third of the locations is the top recommendation in its suburb, and a third does not appear at all. The invisible ones stay invisible to head office, because the reporting does not separate them.
Get Social Places’s stories in your inbox
Join Medium for free to get updates from this writer.
That should bother agencies more than it does. If you are advising a multi-location client on AI visibility and reporting a single brand score, you are handing them a metric that cannot survive the obvious question: which stores, and where.
What to do instead? Score per location, against the competitors that outlet actually competes with, not only the national keywords
Use the queries customers actually type. “Best burgers in Durbanville, Cape Town” is correct. “Best burgers near me” is better for organic and maps results.
Then aggregate per keyword so that you have two numbers — the score for each outlet, and one score for that keyword across the whole footprint. The first is a to-do list for your digital agency. The second tells the marketing director whether the brand owns that category nationally.
We built this into our platform because clients kept asking the question a brand-level number could not answer. Each location gets scored against its local competitive set, results roll up per keyword to a brand view. As far as we can establish it is the first time per-location AI visibility for multi-location brands has been tracked this way.
But the number is the easy half. A per-store score tells you a location is invisible in Bellville. It does not fix the wrong trading hours, the unanswered reviews or the missing structured data that put it there. The assistants are reading the same local data across all of the digital channels and declining to recommend what they cannot verify — its important you are listed correctly, answering reviews and postings localised content across the full digital eco system — Google Maps, Apple Maps, Bing, Facebook etc.
So if your AI visibility dashboard is green, ask the only question that matters: green for which stores, and where. If the report cannot answer that, it is not measuring your visibility. It is averaging your blind spots.
Ryan Haworth, CEO and Co-Founder, Social Places
