Insights

News

Plain-English thinking on AI visibility, GEO, and how brands get quoted.

This is where we write about generative engine optimisation — the practice of getting your brand quoted when someone asks ChatGPT, Gemini, Claude or Perplexity a question your business should be the answer to. It is a young field, most of the advice circulating in it is guesswork, and a lot of it is recycled search-engine folklore with the word "AI" pasted on top. We try to write the other kind of piece: what we can actually observe from running the same questions across the assistants week after week, and what we can't.

Expect four recurring themes. Measurement — how to tell whether an assistant cited you, how to separate a real movement from a noisy week, and why analytics alone won't show you any of it. Diagnosis — the page-level reasons an engine skips you: thin or buried answers, missing structure, content that reads well to a person but parses badly to a model. Structured content — FAQ and schema markup, question-led headings, and llms.txt, the emerging file that tells AI crawlers which pages on your site matter. And proof — verifying that a change actually went live, then measuring what happened either side of it.

Everything here is written in plain English, with the evidence and the limits stated. If something is still uncertain, we say so.