The questions worth asking before you pay for SEO or GEO.

SEO is about appearing in ordinary search results. GEO, generative engine optimisation, is about being named or cited inside an AI-generated answer. AEO, answer engine optimisation, was largely the same idea under an older name and the two terms are now used interchangeably by most of the market. None of the three has an agreed definition, which is worth knowing when a vendor quotes one at you. The foundations overlap heavily, because a page still has to be findable and worth trusting. What differs is the measurement: search position is a stable number, while an AI answer is probabilistic and has to be measured as a rate across repeated runs.

It depends which system has to notice, and there are four of them running at different speeds. Titles and snippets on a page that is already eligible can show a click-through change in four to six weeks. Relevance and content work sits on a model retraining cycle of roughly ten weeks. Click history accumulates over a rolling thirteen months, which is why a bad title left up is expensive. Anything touching site-level quality usually takes months. Google says improvements do not have to wait for a major core update, but if nothing has moved after a few months, the next one is often when it shows. We tag every recommendation with its clock so you know what you are waiting for.

No, and nobody can. Google does not sell placement in organic results and AI answers are probabilistic, so the same prompt run twice can return different companies. What we can commit to is method: a dated baseline, work sequenced by mechanism, and the same measurement re-run afterwards whether or not the result flatters us. Anyone guaranteeing a citation is selling you a coin flip without mentioning the coin.

That is the most useful thing you can tell us, and there is usually a structural reason for it. Positive signals appear to add together, but demotions behave like multipliers applied afterwards. If something is capping the site, everything added underneath it is multiplied by close to zero, and good work produces nothing. Most of the time the previous work was not bad, it was just filed behind a cap nobody had looked for. So we look for the cap first, before recommending any content or any links.

Start where your buyers already are. If demand comes through Google, start with SEO. If your buyers are asking assistants for recommendations, start with AI visibility. If you are not sure, that is itself a finding worth checking, and the free check is designed to answer it. In practice most businesses need the Google side working first, because Google AI Overviews and AI Mode run on the ordinary Google index and a page has to be indexed and eligible for a normal snippet before it can support one.

A dated baseline, the findings with the mechanism attached to each one, screenshots or fetched evidence rather than assertions, what is being recommended instead of you, and a sequence rather than a list. Each finding says what it is, which system it affects, what the evidence is, which clock it sits on, and how you will know it is done. The point is not to hand you more data. It is to make the next decision obvious.

Yes. Implementation can include technical fixes, content, internal structure, entity and business-information consistency, and authority work. The audit decides the scope rather than a prebuilt package, and if the scope turns out to be small we will tell you that instead of padding it.

Probably not, and the usual pitch here is overstated. We pulled the robots.txt of 48 agency websites in September 2026. Of the 44 that returned a valid file, none blocked the crawler that lets ChatGPT surface a site. But only four had written any rule about it, so the rest were open by default. A robots file cannot tell you whether that was a decision. The real problem is not that you are blocked, it is that the setting is easy to leave unchecked, and robots.txt is not the only gate because a CDN can challenge a bot that robots.txt permits.

No, and we would tell you not to buy one as a deliverable. We publish one on this site and describe it as documentation hygiene rather than a tactic. In the largest server-log study of the format, 97 per cent of published files were never fetched in the month measured, and Google states plainly that it ignores them. If someone is charging you for one as an AI visibility service, that is a useful signal about the rest of their offer.

No. Structured data is genuinely useful for rich results in ordinary search, and we implement it for that. It is not what gets a page quoted inside a generated answer. Google says so in its own documentation: you do not need new machine-readable files, AI text files, or special schema.org markup to appear in its AI features.

Three things, and they are written into the scope as your deliverables. Access to Search Console and analytics, read-only is fine. Someone who can deploy changes, or permission for us to. And about forty minutes a month to extract facts only your business can state, such as prices, outcomes, timelines and real examples. That last one is the input engagements most often drop, and it is the one that decides whether the work compounds, because facts nobody else holds are the one thing a language model cannot generate for your competitors.

Then we say so. Sometimes the visibility problem is not a search problem at all, and sometimes the honest answer is an afternoon of work rather than a retainer. We would rather lose the project than sell a plan we do not believe in, partly because we have to write the follow-up report either way.

Service businesses, specialist firms, startups and local companies with a clear offer and something true to say about it. The method depends on having facts only you can state, so businesses with real outcomes, real prices or real process detail get more out of it than businesses that need generic content produced at volume. We do not produce content at volume.

Yes. Research, audits and implementation support can be provided as a specialist partner or as a white-label resource, including the AI visibility measurement, which most agencies have not built a method for yet.