Generative Engine Optimisation - what is it?

Every marketing conference this year has had a session on GEO, generative engine optimisation, usually pitched as a completely new discipline requiring completely new specialists. Some of that is genuine. A lot of it is traditional SEO with a new acronym stapled on, sold to businesses who are anxious about AI and don't have the background to spot the difference. Knowing which parts are actually new is the difference between spending your budget well and paying twice for the same work.

What GEO actually means

Generative engine optimisation is the practice of increasing the likelihood that AI tools like ChatGPT, Google's AI Overviews, Perplexity and Copilot name, cite or recommend your business when answering a relevant question. That's a genuinely different outcome from ranking a page in a list of blue links, because the user often never visits a website at all. The model gives them an answer directly, and either your business is part of that answer or it isn't. There's no page two to fall back on, and no impression count to point to as a consolation.

What hasn't changed at all

Strip away the branding and most of what GEO requires is work that's been standard SEO practice for years:

• Crawlable, fast, well-structured websites
• Content that directly and clearly answers a real question
• Authority built through genuine backlinks and industry coverage
• Clear site architecture that makes it obvious what a business does and who it serves
• Technical fundamentals like clean metadata, working sitemaps and no crawl errors

None of that is new. If an agency is pitching GEO as a replacement for these fundamentals rather than an addition to them, that's a red flag, not a differentiator. It's also worth noticing how much of the GEO conversation online is itself just old SEO advice dressed in new terminology: "build authoritative content" and "earn quality backlinks" have been standard guidance for well over a decade, and repeating them under a new acronym doesn't make them a new discipline.

What genuinely is different

A few things about how AI models select and cite sources are real departures from how Google's classic ranking algorithm works:

1. 01Entities matter more than exact keyword matches. Models track how firmly your brand, services and sector are connected across everything they've read, not just what phrase appears on your page
2. 02Unlinked brand mentions carry weight. A trade publication naming your business without a hyperlink used to do very little for classic SEO. To a language model, that's still evidence worth weighing
3. 03Consensus across sources outweighs your own website. Models place significant trust in what independent sources say about you collectively, more than what you say about yourself
4. 04Answer-shaped content gets lifted and quoted. A direct, well-structured answer to a specific question is far more likely to be cited than a long, meandering article that eventually gets to the point
5. 05AI crawler access is a new technical gate. If your robots.txt blocks GPTBot and similar crawlers, you've opted out of being cited before any content work even starts

Why this distinction actually matters for your budget

Businesses are increasingly being sold two separate retainers: one for SEO and a second for GEO, often from two different agencies, doing largely the same underlying work with different reporting dashboards. That's an expensive way to buy overlapping services. The more sensible approach is one programme that covers the shared foundation properly and layers the genuinely new work, entity building, AI crawler access, citation tracking, on top of it, rather than duplicating the basics twice.

This is worth raising directly with any agency pitching GEO as a separate line item. Ask them to show, specifically, which parts of the proposed GEO work aren't already covered under your existing SEO retainer. A confident, specific answer naming entity consistency, AI crawler access and citation tracking suggests they've genuinely thought it through. A vague answer about "AI optimisation" repeated back at you is a sign the pitch is following a trend rather than solving a defined problem.

How to tell if an agency actually understands the difference

Ask what specifically changes in their process for GEO work compared to their standard SEO process. A credible answer names entity consistency, structured data for AI parsing, citation and mention tracking, and AI crawler access. A vague answer talks about "training the AI" or "optimising for ChatGPT's algorithm", both of which misunderstand how these models actually retrieve and cite information. There's no algorithm to game the way there once was with early Google, because the model isn't ranking your page, it's deciding whether to trust and repeat what it already knows or finds about you.

The honest answer is that GEO is mostly SEO done properly, with a genuinely new layer of entity and citation work on top. Anyone selling it as a total departure from what came before either doesn't understand the mechanics or is counting on you not to ask.

Common objections, answered

If it's mostly the same work, why does our SEO agency not already do it?

Plenty don't, simply because the agency hasn't updated its process yet. Entity building, citation tracking and AI crawler access are specific additions that need to be built into a workflow deliberately. An agency that's been running the same SEO playbook unchanged for several years is unlikely to have added them without being asked directly.

How do we know if GEO is even worth prioritising for our business?

Run the manual check first. Ask ChatGPT and Google's AI Overviews the actual questions your buyers ask before choosing a supplier in your category. If competitors are already being named and you're not, that's a clear signal it's worth prioritising now rather than waiting for it to become obvious in your enquiry numbers.

Does GEO replace the need for a content strategy?

No, it depends on one. Entity building and citation work need something to point to: pages that genuinely answer a question well. Businesses hoping to skip content and jump straight to "AI optimisation" are missing the part that actually gives a model something worth citing in the first place.

What a properly combined programme looks like

In practice, the sensible version of this isn't two separate workstreams running in parallel. It's one programme where technical SEO work also covers AI crawler access, content briefs are written to be both genuinely useful to a human reader and structured for a model to lift cleanly, and authority building tracks both classic backlinks and unlinked brand mentions on trusted sites. Reporting then covers organic rankings and AI citations side by side, since a business that's growing in one but stagnant in the other still has a real gap in how it's found.