GEO: optimising for Google's AI Overview as a B2B brand

AI Overview now sits above the first organic result on roughly half of B2B queries we track. Click-through rates on the traditional blue links have dropped 30 to 50 percent on the queries it shows up for. If you sell B2B services or software and you’re still optimising only for position one, you are optimising for a SERP that no longer exists. Generative Engine Optimisation — GEO — is the new game, and the rules look different.

AI Overview cites differently than Google ranks

Google ranking is about link authority, on-page relevance, and user signals. AI Overview citations are about extractable answers. The model needs a sentence or paragraph it can lift, attribute, and present as authoritative. That means short, declarative sentences with concrete claims earn citations more reliably than long, flowing prose. The number-one ranked page often isn’t the cited page.

Structure beats prose

Posts that get cited share a pattern: a clear question as an H2, a one-paragraph direct answer underneath, then the supporting detail. We rewrote 14 B2B blog posts for a SaaS client into this structure and citation rate inside AI Overview tripled in eight weeks. The model wants to find the answer fast, not parse three paragraphs of context.

• Question H2. Phrased the way a buyer would actually ask it.
• Direct answer. 40-80 words, declarative, no hedging.
• Supporting evidence. Numbers, examples, founder quotes.

Original data is the unfair advantage

AI Overview cannot cite a fact that exists on twelve other sites without picking one source. Original research, internal benchmark numbers, founder-interview quotes — these have only one source, and that source gets cited. We push every B2B client to publish one piece of original data per quarter. A 200-respondent survey of their own customers is enough. That single asset will earn more citations than fifty derivative posts.

Brand mentions matter even without links

LLMs train and ground on the open web, and they pick up brand mentions even when there’s no link. A founder podcast appearance, a guest essay, a Reddit thread that names your product — these all feed into the model’s understanding of who you are. We’ve watched brands climb into AI Overview citations primarily through unlinked mentions in industry newsletters and podcast transcripts. PR is now an SEO channel again.

The schema layer most B2B sites are missing

Article, Organization, Person and FAQ schema are not optional in 2026. We’ve seen pages with identical content rank identically in classic search but get cited very differently in AI Overview because one had clean schema and the other did not. The model uses structured data to disambiguate authorship, organisation, and topical scope. It’s a one-time engineering effort with disproportionate return.

Track citations, not just rankings

Tools like Profound, Otterly, and even manual prompt audits now let you track which queries cite you in AI Overview, ChatGPT, and Perplexity. We run a monthly audit of 40 buyer-intent queries for each B2B client and report citation share alongside traditional rank tracking. Founders who only track rank are flying blind on the channel that’s actually growing.

How we help at The Nerdish Mic

We build GEO programs for B2B founder-led brands — the content restructure, the original-data calendar, the schema cleanup, and the citation tracking that proves it’s working. If your blog traffic dropped in 2025 and you suspect AI Overview is the reason, talk to us. We’ll audit your current citation share in two weeks and tell you honestly whether GEO is your highest-leverage channel.