_search_visibility

SEO, GEO and AIO: what businesses need to know about AI search

Three terms, one overlapping job. What has changed in search, what has not, what a sensible few months of work looks like for a UK business, and the promises to be sceptical of.

If someone has recently offered your business “GEO” or “AIO” as a new service, here is the short version. The three terms describe one job seen from three angles: making sure the systems people use to find things can find you, understand what you do and describe you accurately. Most of that work is the same work it has been for years. Some of it is new. None of it comes with a guarantee of appearing in any assistant, whatever the proposal says.

This piece sets out what the terms mean, what has changed and what has not, what a sensible few months of work looks like, and the claims we would walk away from.

The three terms, plainly

SEO, search engine optimisation, is the established discipline. Make pages that search engines can crawl and index, that load quickly, that match what people are actually searching for and that are useful when they arrive. It covers the technical state of the site, the content on it and how the rest of the web refers to it.

GEO, generative engine optimisation, is a newer term for improving how your expertise and information can be found and referenced when an answer is generated rather than listed. AI Overviews and AI Mode in Google, ChatGPT with web search, Perplexity and similar tools assemble a response from several sources. GEO is the effort to be one of the sources worth drawing on, and to be described correctly when you are.

AIO, AI optimisation, is the widest of the three and the least settled. People use it to mean making your digital information clear and useful across AI-assisted discovery generally: search, assistants and whatever arrives next.

The honest observation is that these overlap almost completely. A page that is crawlable, fast, specific and well structured is good for conventional search and is also the kind of page a generative system can quote. A business that is named and described consistently on its own site, its Google Business Profile, Companies House and its social profiles is easier for a search index and a language model alike to pin down. Anyone selling SEO, GEO and AIO as three separately priced products is selling one job three times. We offer them as one service for that reason.

What has not changed

This is worth stating plainly, because the noise suggests everything is new.

  • Crawlable, fast, clear pages. If a search engine cannot fetch and render a page, nothing else about it matters. Google’s own documentation on AI features in Search says there is nothing special to add for them and that the usual requirements apply. The work is unglamorous: a clean sitemap, sensible redirects, no important copy hidden behind scripts, pages that load quickly on a phone.
  • Useful original content. A page that answers a real question with specific detail from real experience is the thing every system is trying to surface. A page that restates what ten other pages say, in more words, is not.
  • Consistent business identity. The same name, address, description and contact details everywhere they appear, with the same account of what you do. Inconsistency was always a problem; it is now a bigger one, for reasons covered below.
  • Supported structured data, where it is accurate. Markup that describes what is visibly on the page helps search engines understand it. Google publishes the list of structured data types it supports; anything outside that list will not earn a rich result, and markup that claims something the page does not say is a liability rather than an asset.

What has changed

  • Answers are assembled from several sources. A conventional result is a list of pages and you choose one. A generated answer is a paragraph stitched together from several, sometimes with citations and sometimes without. Your page is competing to be a component of the answer rather than the destination.
  • Fewer clicks for some queries. When the answer to an informational question appears on the results page, a proportion of people will read it and leave. For the queries that matter to a business, the ones where someone needs to choose a supplier, the click is still the point. But it means measuring impressions alone tells you less than it did, and measuring qualified enquiries matters more.
  • Being clearly described and consistently named across the web matters more. A system assembling an answer about your company is drawing on how the whole web describes you, not only your site. Directories, professional profiles, supplier listings, press mentions and reviews all feed in. Where they disagree, the result is a muddled or wrong description, and there is no form to fill in to correct it.
  • Measuring AI referrals is rough. Some assistants send a referrer when a reader clicks through, some do not, and the tools you already use mostly do not separate this traffic out. You can sample how you appear for a set of questions, and you should, but it is a sample rather than a measurement, and anyone presenting a precise “AI visibility score” is presenting an estimate with a confident face.

A sensible plan for a UK business

Nothing here needs a year. Three to four months is enough to get the foundations right and start learning from evidence.

First month: health and identity. A technical audit of the site: what is indexed, what is slow, what redirects where, what is broken. Search Console and Bing Webmaster Tools set up and checked. Then the identity work: one agreed description of the business, one set of contact details, applied to the site, the Google Business Profile and every profile you control. Organisation markup on the site that points only to profiles that are real.

Second and third months: content worth quoting. Write down the questions customers actually ask before they buy, not the keywords a tool suggests. Pick a handful and answer each on a page that gets to the point in the first two paragraphs, carries a named author and a real date, and includes the specifics only you can supply. Add structured data for articles and services where it is accurate. Fix the pages that already rank but answer badly.

Third month onwards: measure and adjust. A baseline of qualified enquiries and where they came from. A monthly check of a fixed set of questions in the main assistants and in Google, noting whether you appear and how you are described. Search Console trends for the pages you have worked on. The next quarter’s work comes from what that evidence says, not from a package.

What to be sceptical of

  • Promises of inclusion. Nobody outside the companies that run these systems controls what they show, and those companies do not sell placement in the answers. A guarantee of appearing in an assistant is a guess with an invoice attached.
  • Mass-generated pages. Hundreds of thin pages produced by a model, each targeting a slight variation of a query, do not demonstrate expertise. They are the kind of content the systems are built to discount, and they dilute the pages that do deserve attention.
  • Invented citations and manufactured mentions. Fabricated reviews, fake directory entries and planted “as seen in” lines are a reputational risk, and they create exactly the inconsistency that muddles how you are described.
  • Markup sold as magic. FAQ schema on every page, a special file that supposedly tells assistants what to think of you, hidden text written for machines. Structured data is useful when it describes what is on the page. It is not a lever that makes an assistant prefer you.
  • Precise scores for things that cannot be measured precisely. A dashboard can sample; it cannot see inside a model.

What to do next

If your site is slow, thinly described or inconsistently named around the web, start there. That work pays off in conventional search today and in generated answers as they develop, and it is the same work either way. If you want to know what is realistic for your business in assistants specifically, we have written an honest account of what you can and cannot do about appearing in ChatGPT and AI Overviews. And if you would like us to look at where you stand, our search and AI visibility service begins with that audit, or you can get in touch and describe the situation.

Sources

_written_by Ryan Byrne Co-founder and lead developer

Ryan leads development at One Size Media. He designs the systems, writes and reviews the code, and looks after the platforms clients rely on every day.

Want to talk this through for your business?

Tell us what you're trying to achieve and we'll help work out the next step.