Make your organisation citable in AI answers, not just ranked

It is a different layer on the same foundation. Classic SEO earns a ranked link; GEO earns a citation inside a generated answer, which is a different unit of visibility. The overlap is real: crawlable, fast, well-structured pages help both. The differences are answer-first passage structure, consistent entity data, hand-written schema, llms.txt, AI crawler access, and measurement by share of answer rather than by position. We run the two together for that reason.

Share of answer is the headline number: across an agreed set of buyer prompts, how often an assistant names you, cites a page, links to you, or leaves you out. Each month you get that figure broken down by assistant, the specific pages cited, the competitors quoted instead of you, any factual errors an assistant repeated about you, and referral sessions from assistant traffic in GA4. We also report the classic signals underneath, because they move first.

No, and nobody honestly can. Assistants generate answers probabilistically, from sources they choose, using models that change without notice. The same prompt can return different answers on the same day. What we can do is remove every mechanical reason you are not eligible to be cited, publish the source material worth citing, and show you the measured change in share of answer over time. An agency that guarantees a citation is guaranteeing something it does not control.

You do, all of it. The copy, the structured data, the llms.txt file and the robots rules live in your repository or CMS and are yours on publication. Prompt sets, the full citation history, dashboards and audit findings are handed over in editable formats, usually a spreadsheet and a document, not a portal we control. Any tool accounts are opened under your organisation with OlDevs added as a user, so removing us removes nothing.

Weeks for the mechanical problems, longer for the rest. If a crawler was blocked or your entity data was contradictory, citations sometimes appear within weeks of the fix. Building genuine source authority in a competitive category takes more time, and we would look for a clear trend in share of answer over roughly three to six months. Assistants also cache and refresh at different rates, so one engine often moves well before another.

That is a business decision, and we present it as one rather than deciding for you. Blocking GPTBot or Google-Extended protects content from training use but also removes you from the answers your buyers read. Most organisations open the marketing site and keep proprietary material, gated research and customer data closed. We map every bot, show what each one is used for, and implement the rules your legal team approves.

It depends on the size of the site, how many prompts and assistants you want tracked, how much of the entity data needs reconciling, and how many source pages have to be written, so we do not publish a rate card. We run a short scoping call, do a baseline prompt run and a technical read of the site, then send a written proposal with named deliverables and a clear scope. Request a quote and we reply within one business day.