There is a new acronym doing the rounds, and it comes with a price tag. GEO, or Generative Engine Optimisation, is being sold as the skill you need to show up inside ChatGPT, Perplexity and Google's AI answers. Agencies have built packages around it. Some are charging a premium for it. So it is worth saying plainly what we tell our own clients: GEO is mostly SEO wearing a new hat, and that is good news for you.
I am not being cynical for the sake of it. There are one or two genuine differences, and I will get to them. But the core claim, that this is a brand new discipline requiring a brand new budget line, does not hold up when you look at what actually drives a citation.
What LLMs actually cite, and why
When an AI system answers a question with sources, it is not performing magic. It has retrieved a set of web pages, decided which ones are relevant and trustworthy, and summarised them. The pages it reaches for tend to share a few traits.
- They are structured clearly, so the model can find a clean answer to lift.
- They carry strong entity signals, meaning it is obvious who wrote them and what the business is.
- They are mentioned and referenced elsewhere, which is how trust gets established.
- They are technically healthy, crawlable and fast enough to be indexed properly.
Read that list again. There is nothing on it that a competent SEO has not been chasing for the last decade. Structured content, entity clarity, authoritative mentions, technical health. The generative engine is drawing on the same signals a search engine always has, because it is largely sitting on the same index. If you want the fuller version, we have written about what actually changes when AI enters search and what does not.
The overlap is almost total
Here is a way to think about it. Ranking factors and citation factors are two circles that overlap by perhaps ninety percent. The page that ranks well for a question is usually the page an AI reaches for when answering it, because both systems are asking the same underlying thing: is this source relevant, clear and trustworthy?
That means the work does not fork into two projects. Fixing your site structure helps both. Earning a mention in an industry publication helps both. Writing a genuinely useful answer to a real question helps both. You are not buying a second engine. You are tuning the one you already have so it performs on more surfaces.
This is why I get twitchy when I see a "GEO package" priced as a separate premium service. In a lot of cases it is the SEO checklist reprinted with the word "generative" added, sold at a markup to a business owner who has been told the rules changed. The rules did not change. The delivery surface did.
A fair warning about the packages
If someone pitches you GEO as an entirely new discipline that replaces your SEO, ask them one question. What are you doing that a good SEO plan would not already cover? A straight answer will name a couple of specifics. A vague answer, heavy on the acronym and light on the mechanics, is usually the tell that you are being sold repackaged work at a premium.
You do not need two invoices. You need one team doing the fundamentals well and paying attention to how AI surfaces pick up their output.
The honest differences
I promised I would not pretend the overlap is total, so here are the real distinctions, because they do exist and they matter at the margins.
First, query phrasing. People talk to AI tools in full, conversational questions rather than clipped keywords. Someone who typed "best accountant Norwich" into Google might ask an assistant "who is a good small business accountant near me who is used to dealing with contractors." Content that answers the fuller, more natural question tends to get pulled in. Writing for how people actually speak is a genuine shift in emphasis.
Second, being the quotable source. Ranking is about being clickable. Citation is about being liftable. A model wants a clean, self-contained sentence or short passage it can drop into an answer with attribution. If your key point is buried in a rambling paragraph, it is harder to quote than a competitor who stated the same thing crisply. So the craft of writing tight, standalone answers gets a little more valuable than it used to be.
Those two things are real. They are also refinements of good SEO, not a replacement for it.
What to actually do
Keep doing the fundamentals, and do them properly. Structured, well-written content. Clear signals about who you are. Mentions on sites that matter in your field. A technically sound site. Then adjust the emphasis: answer full conversational questions, and write your key points so they can be quoted cleanly.
If you would like a straight opinion on whether your current SEO already covers the AI surfaces, or whether there are genuine gaps worth closing, that is a conversation we are always happy to have without the acronym markup.