How AI search visibility shows up in pipeline
Written by
Odin Lund
Category
Playbook
Date
Share this article
Standing is not revenue
Three measures describe where a business stands in AI answers. Visibility is the share of relevant answers that name it. Share of voice is its share of all company mentions across those answers. Position is its rank among the businesses an answer names.
All three are leading indicators. None of them is money. A business can move share of voice from single digits to first place and still not know what it bought, which is the position most reporting in this category leaves you in.
The connecting step is to treat a mention as the start of the measurement rather than the end of it.
AI referrals behave differently from search traffic
Two things are true at once: there are fewer of these visitors, and they are worth more each.
Semrush put the value of a visit from AI search at 4.4 times a traditional organic visit in July 2025. Ahrefs found that AI search drove 0.5% of visits to its own site but 12.1% of signups in June 2025. Adobe Analytics measured AI-referred retail traffic converting 54% better than other channels by May 2026, alongside 53% more time on site.
The reason is structural rather than magical. The comparing already happened inside the answer, so the person arriving has done their shortlisting. They are later in the decision than a search visitor at the same moment.
What to connect, and in what order
Start from the answer. For every answer that names the business, record which question produced it, which model, and which sources it cited.
Then the visit. AI assistants pass a referrer, so sessions arriving from them can be separated from organic search in analytics rather than estimated.
Then the page. Which page did the visitor land on, and was it the page that answered the question they asked the model?
Then the outcome, which is the step most businesses skip: did that session become an enquiry, and did that enquiry become a customer. This requires joining analytics to whatever the business uses for intake, and it is the only step that turns the rest into a number anyone outside marketing cares about.
Reading it by stage, not in aggregate
An aggregate number hides where the value is. In our legal data set, decision-stage prompts named the client in 37.8% of answers while comparison prompts named it in 6.3%.
Those are not equally valuable positions. Being named when someone has decided what they want is closer to revenue than being named while they are still reading around the subject.
So we report every measure by model and by stage of the customer journey, and match referrals to leads on the same split. It is the difference between knowing that visibility went up and knowing which part of it paid.
Share Blog
