By James Richardson, Founder, Optimising
Every few weeks lately, someone shows me a dashboard with a single number on it and a lot of hope pinned to that number. Eighteen years in SEO, and I know that look. It is an industry falling in love with a metric, falling in love with something to point at, or something to charge for, and we are doing it again.
The metric this time is AI visibility. Your share of voice in ChatGPT, your citations in Perplexity, and whether Google’s AI Overview bothers to mention you at all. There is a dashboard for it, a new tool every week, and a LinkedIn post every morning telling you that if you are not tracking it you are already behind.
As of this month, at least part of the raw version of it is free. Google started reporting AI visibility inside Search Console on 3 June, and Bing followed on 17 June, initially in the UK and then globally. The thing agencies were charging a premium to “uncover” is now a standard report on at least one of the biggest surfaces that matter.
When a number goes free to everyone overnight, it is fair to ask how much of an edge it ever was. I, for one, have watched this film before, and I know how it ends.
For a decade, SEO chased rankings the same way. “We’re number one”, went at the top of the report, the client nodded, and revenue sat exactly where it had been the month before.
We were polishing a scoreboard.
Ranking first for a term nobody valuable searched for, or one that never turned into a sale, the scoreboard read well, but it was stacked in our favour, quietly distracting us from the only contest that paid the bills.
The best people in this industry learned that the hard way and stopped confusing position with profit. Sadly, AI visibility is walking us straight back to the bottom of that hill.
Here is the part nobody selling an AI-visibility dashboard wants to say out loud: you can be cited all over AI answers and sell nothing. A citation means a machine name-checked you. It does not mean a person chose you, and it certainly does not mean anyone bought anything. For all the noise and traffic these platforms generate, the traffic they actually send is still, by and large, tiny. SparkToro’s latest numbers put ChatGPT at roughly 0.2 per cent of referral traffic.
Visibility is not a demand, and right now, it is barely even traffic.
The figure does not hold up on its own terms either. Mike King, who runs iPullRank and is about as respected as technical SEOs get, has shown that in the newer “agentic” search, where one question quietly fans out into a dozen smaller ones, citation counts miss most of what is actually going on, under-counting a brand’s real footprint several times over. So the thing everyone is rushing to optimise toward is both disconnected from revenue and, a lot of the time, simply wrong.
Believe it or not, even the measurement people have flagged it.
AMEC, the international body for communications measurement, used its new AI guidance to warn against exactly this: vanity scores, numbers nobody can audit, presence wearing a suit and calling itself proof. When the people whose whole job is rigorous measurement tell you the popular metric is hollow, that is worth more than a pause.
So what do we do instead?
Honestly, the same unglamorous thing the good operators landed on with rankings years ago. Use the number to point you somewhere. Do not mistake it for the destination.
AI visibility is a decent leading indicator. It is not a result. The test I would want any agency, mine included, to hold to is simple. When a brand’s AI visibility climbs, do the commercial signals climb with it in the same window? Branded search, direct traffic, and the assisted conversions that show up when someone who first met you in an AI answer comes back to buy. If those move together, you have found something real. If the visibility chart is soaring while branded search flatlines, you have not won anything. You have a chart that may be nice to look at, but ultimately one that will not move the dial for a CFO.
In our own tracking across four AI platforms and more than a hundred Australian brands, the thing that separates signal from noise is never who gets cited the most. Let that sit with you for a second. Not once was the brand cited the most, the one winning the most sales from AI.
So who is winning?
It is whose citations drag a commercial result along behind them. Sometimes the brand winning the visibility race is the one quietly going backwards on revenue, and you only catch it if you watch both at once. I am happy to show anyone how we set that up.
None of this means AI search does not matter. It matters enormously, and it will matter more over time. But the winners over the next few years will not be the brands with the biggest visibility number – that is a vanity metric.
They will be the ones who worked out early, which slice of that visibility actually moves money, and had the discipline to let the rest go.
We took the long way round on rankings once. It cost the industry the best part of a decade, and a lot of clients’ budgets, to accept that ranking first is not the same as getting paid.
I would rather we not do it twice, and I would not let an agency talk you into it either.