It's a Tuesday morning in 2026. A small business owner in Hertfordshire — let's call her Fiona — wants to find a brand designer for her new coaching business. She picks up her phone.

In 2019, Fiona would have typed "best brand designers Hertfordshire" into Google. She'd see ten blue links. She'd click two or three. The first one would have a polished portfolio. She'd email them.

In 2024, the journey would have looked similar — but with Google's AI Overview at the top of the page, summarising the answer before the links. Fiona might still scroll. She might not.

In 2026, Fiona doesn't open Google at all. She opens ChatGPT and types "I'm a coaching business in Hertfordshire, who's good for branding under £2,000?" ChatGPT writes back a paragraph. Three names. One of them gets her email.

The other two web designers — the ones who used to rank #2 and #3 in Google for "best brand designers Hertfordshire" — never appeared on Fiona's screen. Their SEO budget is doing exactly what it always did. The query just doesn't pass through Google anymore.

That's the headline. The rest of this piece is the practical breakdown: where SEO still wins, where GEO is taking over, and how to invest if you only have time and budget for one of them.

The 70% they share

It's tempting to frame SEO and GEO as competitors. They aren't. They share most of their underlying tactics — the things you do for one benefit the other.

The shared 70% looks like:

  • Fast, well-structured site. Both Google and AI engines penalise slow, broken, or render-blocked pages. A well-built site is the foundation for both.
  • Clear titles and meta descriptions. Both reward titles that name what's on the page concretely.
  • Original, substantive content. Both reward long-form content that actually says something. Both penalise thin pages.
  • Internal linking and clean URLs. Both use site structure as a signal of topical authority.
  • External authority signals. Both treat directory listings, press mentions, podcast appearances, and credible backlinks as trust signals.
  • Mobile-friendly, accessible. Both penalise sites that don't work on phones.

If you're already doing these well, you're 70% of the way to GEO without lifting a finger.

The 30% where they diverge

The other 30% is where it gets interesting. SEO optimises for ranking on a list. GEO optimises for being the answer. They reward different things.

What SEO weights that GEO doesn't

  • Keyword density and exact-match phrasing. Less critical for GEO. AI engines understand intent better than keyword-matching.
  • Backlink quantity. Google's PageRank-derived signals still weight raw link counts. AI engines weight citation patterns — being mentioned credibly across many sources matters more than the number of inbound links.
  • Click-through rate from search results. Google watches this signal. AI engines mostly don't have it because there's no SERP to click from.

What GEO weights that SEO doesn't

  • Definitive, declarative language. "X is Y" gets cited. "X may, in some contexts, be considered Y" doesn't. AI engines look for unambiguous source material they can quote.
  • Structured data that explicitly answers questions. FAQPage schema is the highest-value addition. Article schema with proper headline and description fields. Service schema with offers and prices. AI engines literally read these.
  • An llms.txt file. A markdown sitemap-equivalent for AI crawlers. Ours is here. SEO doesn't care about it. GEO does.
  • The first paragraph. AI engines weight the top of the page heavily — it's the easiest material to extract and cite. The first paragraph of every page should be a clean, definitive statement of what the page is about.
  • Topical clustering. A pillar page plus 4-6 spoke articles all internally linked outranks 10 disconnected articles. AI engines build topical authority more aggressively than Google does.

Query-by-query: where each one wins

The clearest way to see the divergence is by query type. Here's how the same Hertfordshire small business owner might be served differently across query patterns:

"Best brand designer for coaches in Hertfordshire"

Where the user goes: ChatGPT, Perplexity, or Google AI Overviews.
What they get: A paragraph naming three to five agencies, often with one-line descriptions and approximate prices.
Who wins: The agencies whose pages are clearly structured around this exact intent — service page named correctly, FAQs covering price and audience, location declared in schema.
SEO/GEO weighting: 80% GEO, 20% SEO.

"Pentagram portfolio"

Where the user goes: Google.
What they get: A direct link to pentagram.com. Branded navigational queries don't get AI-summarised — there's nothing to summarise.
Who wins: The brand whose domain matches their name.
SEO/GEO weighting: 100% SEO.

"Why is my website not ranking on Google"

Where the user goes: ChatGPT or Perplexity. This is exactly the kind of advice query AI engines now own.
What they get: A list of likely causes, with citations to authoritative sources.
Who wins: The agency with a well-structured article answering this exact question, with FAQPage schema and a clear author byline. Service pages don't show up here. Articles show up.
SEO/GEO weighting: 90% GEO, 10% SEO.

"Compare Sort The Clicks vs ClickSlice"

Where the user goes: Both. Mostly ChatGPT and Perplexity for the summary, then Google for both sites.
What they get: An AI-generated comparison plus links to both agency homepages.
Who wins: Whoever has both (a) clean comparison-friendly pages and (b) a well-known brand. AI engines need source material to compare. If neither agency has a good "vs" page, AI engines synthesise from whatever they can find — which favours larger or older brands.
SEO/GEO weighting: 50/50.

"Coaching website inspiration"

Where the user goes: Pinterest, Dribbble, then Google Images.
What they get: Visual inspiration. AI engines don't dominate visual-discovery queries (yet).
Who wins: Sites with strong image SEO and Pinterest presence.
SEO/GEO weighting: 90% SEO (visual), 10% GEO.

The investment question: where to put your time

Here's the practical rubric we'd use for a UK small business in 2026:

  • If you're starting from zero on both: do them together. The shared 70% covers most of the work. Add the GEO-specific 10-15% on top — llms.txt, FAQPage schema everywhere, definitive language, content that answers questions explicitly.
  • If you're strong on SEO already: the missing GEO layer is small but high-leverage. Audit your schema. Add llms.txt. Rewrite hero copy to put your services in the first paragraph. Build out informational articles in your topic clusters. Most of the work is mechanical.
  • If you're strong on neither: do the GEO-friendly version first. SEO takes 6-12 months to compound through backlinks. GEO can show measurable lift in 4-8 weeks for well-structured pages. The early signal lets you justify the longer SEO investment.

Two things almost no UK small business is doing well in May 2026:

  • FAQPage schema on every service page.
  • An llms.txt file at the site root.

Both take a couple of hours. Both move the needle. If you do nothing else this quarter, do those two.

The 12-month forecast

Three things will likely be true by mid-2027:

  1. "AI search visibility" will be a line item in marketing budgets the way "SEO" is now. Currently most UK small businesses don't have one. By next year, most will.
  2. The early movers will compound. The agencies and businesses that get cited regularly in 2026 build a citation memory in AI training corpora that's hard to displace. Late-comers will need to spend more to break in.
  3. SEO won't die. It'll become one channel among several, sized roughly proportionate to the queries that still happen on Google. Likely smaller, but not zero.

The strategic question isn't whether to do GEO. It's whether to do it now while it's still cheap, or to do it in 18 months while playing catch-up.

If you'd rather not figure this out yourself — that's what our AI search & GEO service exists for. Three tiers: Insights £250/month (we monitor, you implement), Managed £600/month (we do the work, including one article per month), Premium £1,200/month (we go further — two articles, external citations, and competitor displacement). No setup fees on any tier.