Ranking in AI answers: what actually moves the needle
AI-generated summaries now sit above organic results for a growing share of queries. What we have learned about being cited in them.
The share of searches that end without a click keeps rising. For informational queries in several of our clients' categories, it is now the majority. This is not a reason to abandon search — it is a reason to change what you optimise for.
Across the accounts we manage, the pages cited most often in AI answers share a few unglamorous traits.
They answer the question in the first hundred words, in plain declarative sentences, before the context and the caveats. Models extract claims; buried claims do not get extracted.
They contain specific, attributable facts — numbers, dates, named methods — rather than general advice. A page that says "most B2B sites convert between 1.8% and 3.2%" gets cited. A page that says "conversion rates vary" does not.
They are structurally clean: real headings that describe content, tables for comparisons, and schema markup that confirms what the page is about.
And they come from domains with corroborating signals elsewhere — the same entity, saying consistent things, across sources the model already trusts. Digital PR turns out to matter more here than it did for traditional rankings, not less.
Want this run on your account?
We will apply the same diagnostic to your business and show you the three changes we would make first — no charge, no obligation.
Book a Free Strategy Call