Measuring AI VisibilityPart 1 of 9

From clicks to citations: the metrics that replace rankings

As buyers move to AI answers, rankings and clicks stop telling the whole story. The new scoreboard is presence, citations, share of voice, and sentiment. Measured inside the answers themselves.

Buffy Editorial2026-06-09 · 2 min read

For two decades the scoreboard was simple: rankings and clicks. In AI search, both are going quiet. Answers are increasingly zero-click, and the buyer forms an impression inside the answer, often without ever visiting your site. If you only measure traffic, you're blind to where the decision now happens.

The metrics have to move with the behaviour: from clicks to citations.

What stops working

  • Rank alone. Being #3 in the blue links matters less when the AI answer above them names two brands and you're not one.
  • Clicks/CTR as the whole story. A zero-click answer can win or lose you a customer with no click to count.
  • Single-engine checks. ChatGPT, Gemini, Claude, and Google's AI surfaces disagree; one of them is a third of the picture.

The new scoreboard

Track these inside the answers, across engines, over time:

Metric What it tells you
Share of Voice Do you appear, and how often vs competitors?
Citation Coverage Is your own site the cited source, or someone else's?
Brand Perception How are you described. Premium, budget, dated?
Sub-query recall Across the fan-out, how many branches contain you?
Consistency How reproducible is all of the above (AI is non-deterministic)?

Why this isn't just "SEO with new words"

These aren't rankings relabelled. Citation and presence behave differently: they vary run-to-run, differ by engine, and are driven by entity strength and corroboration as much as on-page work. You can't infer them from a rank tracker. You have to ask the engines the questions your customers ask and log who gets named and cited.

The brand cited in a zero-click answer can win the sale without ever appearing in your traffic report. If your dashboard only counts visits, you're measuring the old game.

The practical shift

Keep your SEO metrics. They still feed AI Overviews. But add an answer-level layer: presence, citations, share of voice, and sentiment, sampled across many prompts and every engine, tracked as a trend. That's the gap measuring AI visibility closes, and exactly what Buffy Intel is built to report.

Frequently asked

Do rankings and clicks still matter in AI search?

They still matter. Especially for AI Overviews, which lean on Google's index, but they no longer tell the whole story. With answers increasingly zero-click, a brand can shape a buyer's shortlist by being cited in the answer even when no visit shows up in analytics. You need answer-level metrics alongside the traffic ones.

What should I measure instead?

Presence (do you appear?), share of voice (vs competitors), citation coverage (is your own site the source?), sentiment (how you're framed), and sub-query recall (across the fan-out, how often you show up). Tracked across every engine, over time.