Search rankings and AI answer citations are two different systems with two different competitor sets. We work both, measure both separately, and tell you which one moved.
The usual offer is content at volume. A model writes forty articles, they go live, and the report counts words published and keywords tracked. It is cheap to produce and it is easy to sell, because the output looks like progress before anything has been indexed.
The problem is not that a model wrote it. The problem is that nobody checked whether the search existed, or whether the page could ever rank for it. We have watched keyword tools return a two thousand a month term at a difficulty of zero, which reads as free money, and the live results page turn out to be a jobs board. Ranking there would deliver applicants, not customers.
Every keyword we commit a page to gets checked against the live results first. It takes about ninety seconds and it is the difference between a page that ranks and a page that ranks for the wrong people.
Being cited inside an AI answer is not a by-product of ranking well. We measured this directly on our own domain: comparing the set of sites that dominate Google organic results for a term against the set of sites actually cited in AI answers for the same buying question, only a handful of domains appeared in both.
The organic results were owned by niche specialists. The AI answers were owned by aggregators, directories and large established names. None of the specialists ranking on page one were being cited by the answer engines at all.
That has a practical consequence. Out-ranking your direct competitors will not, on its own, get you quoted inside ChatGPT or an AI Overview. Both are worth doing. Neither substitutes for the other, and an agency that reports one number for both is not measuring either.
Technical foundations first, because nothing else compounds without them. Crawlability, indexation, canonicals, sitemaps that carry honest modification dates, and structured data that validates rather than merely exists.
Then the pages themselves, targeted at terms that survived a live results check, written for a specific buyer rather than a word count, and marked up so an answer engine can lift a passage cleanly.
Then measurement, split properly. Rankings and impressions on one side. Citation presence in AI answers on the other. Reported separately, because they move for different reasons and averaging them hides both.
The gate applies here as it does to everything else we sell. We agree what the work has to move before it starts, establish where you stand today, and measure against that figure rather than against a dashboard of activity.
For a domain with no history, the honest baseline is often zero on every axis, and saying so is more useful than a report full of vanity metrics. A number that starts at zero at least tells you when something has changed.
Nothing here is sold as a silo. Most engagements start with one of these and pull in the next once the first is paying for itself.
The things owners actually ask on the first call.
It can, and that was never really the question. Google has been clear that it rewards helpful content regardless of how it was produced. What gets penalised in practice is content produced at volume with no demand behind it, because nobody checked whether the search existed. The model is not the problem. Skipping the research is.
SEO is about being ranked in a list of results. GEO, or generative engine optimisation, is about being cited inside an answer. We measured the two competitor sets against each other and found very little overlap: the sites winning organic results were largely not the sites being quoted in AI answers. They are different games and they need separate measurement.
On its own, no, and we would rather say that than sell it. Google's own guidance is that generative AI search runs on core search systems and does not require special files. An llms.txt is cheap and worth having, but the durable work is the same boring foundation: indexable pages, clear entity signals, and content clear enough to be quoted.
Longer than paid, and anyone who gives you a date without seeing your domain is guessing. Indexation on a new page can take days or weeks. Ranking movement on a competitive term takes months. We agree the leading indicators before starting so progress is visible before the rankings are.
Yes. The work runs in live businesses across the UK, Ireland, the United States and Australia. Search is market specific, so the research is run per market rather than assumed to carry across; a term that is soft in the UK is frequently three times harder in the US.
Half an hour with a founder, not a sales rep. We tell you straight whether a machine fixes this, what it would cost, and what it would return.
Book the call