Skip to main content
Search increasingly answers questions directly instead of listing links. Ranking first is worth less when the answer sits above you and the user never scrolls. The AI section measures whether you appear in those answers.

​
The six tools

Google AI Overviews is the only one that runs off your tracked keywords automatically. The rest are lookups you run per query.

​
Two metrics that matter

Everything here reduces to two rates, and they’re often confused: Trigger rate — of your tracked keywords, how many produce an AI answer at all. You don’t control this. It tells you how exposed your market is: a high trigger rate means the traditional blue links matter less for your keywords than they used to. Citation rate — of the queries that trigger an AI answer, how many cite your domain. This is the number you can actually move, and the goal of everything in this section. Both appear on Tracking → Overview and on the project dashboard.
Citation rate is a percentage of triggered queries, not of all queries. 50% citation on a 4% trigger rate means you’re cited in half of a very small number of answers.

​
Where to start

  1. Google AI Overviews — it needs no input and covers keywords you already track. Read the trigger rate first: if it’s near zero, AI search isn’t yet affecting your market and the rest can wait.
  2. Citation gap — if competitors are cited where you aren’t, that’s a concrete, addressable list.
  3. LLM benchmarking on your most important queries, to see whether the picture differs by platform.

​
Being cited

Citation behaves differently from ranking. Three things follow from how these systems work:
  • You can be cited without ranking well. AI answers draw on sources that answer the question directly, not necessarily the top result. A clear, direct answer on page two can be cited over a longer page at position one.
  • Being cited elsewhere matters. AI systems cite the sources they consider authoritative on a topic, which includes third-party sites that mention you. Citation gap surfaces those sources.
  • Structure helps. Content that answers a question plainly, near the top, in language matching how people ask, is easier to extract than the same information buried in a long narrative.

​
How this data is collected

GEO snapshots run on a daily schedule by default and track topic-level share of voice. The AI search tools on these pages are run on demand, apart from the initial collection when a project is created. AI Overview data for tracked keywords is captured as part of each rank check, which is why that page needs no input. See How data collection works.

​
AI search vs. GEO

Two related sections that answer different questions: Use AI search for specific queries, GEO for tracking a subject area over time.