Craft · How the work gets made
SEO, GEO and AEO: three acronyms, one job
Neon Frame Creative
10 min read
A client forwarded us a proposal recently. Their existing SEO agency had written to say that SEO was no longer sufficient and that a separate GEO retainer would be required, at additional monthly cost, to protect their visibility in AI search.
Attached was a twelve page deck. Somewhere around slide five it explained that GEO required authoritative, well structured, clearly written content on a fast, technically sound website.
Which is what the SEO retainer had been buying for four years.
We are not going to pretend nothing has changed. Plenty has, and some of it matters. But the gap between what these three acronyms actually require of you and what the market is currently charging for them separately has become quite wide, and it is worth being straight about.
What each one means, without the deck
SEO. Search engine optimisation. Getting your pages to rank in a list of results, mostly Google, so that someone clicks yours. Keywords, technical health, links, authority. The goal is a visit.
AEO. Answer engine optimisation. Getting your content lifted out and used as the answer itself. Featured snippets, People Also Ask, AI Overviews, voice assistants. The goal is to be the thing that gets read out, whether or not anyone clicks.
GEO. Generative engine optimisation. Getting your brand named and cited inside answers written by ChatGPT, Gemini, Perplexity and Claude. The goal is to be in the answer when someone asks a question that has nothing to do with your name.
Three surfaces, three goals. So far so tidy.
Now the part the deck skips
These are not three separate disciplines. They are one discipline with three exits.
The clearest evidence is in the overlap. Ahrefs analysed four million AI Overview citations in March 2026 and found that 37.9% of the pages cited already rank in the top ten organic results, and 69.1% sit somewhere in the top hundred. Being good at the old thing turns out to be most of being good at the new thing.
The overlap is falling, though, and the direction matters more than the number. Eight months earlier the same analysis put the top-ten figure at around 76%. Ahrefs puts the drop down to query fan-out: the engine now answers a question by running several related searches rather than the one you typed, so it cites pages that rank for the query next door.
That is easy to read as an argument against everything above. It is not. Seven in ten citations still come from pages that rank somewhere. What has stopped working is ranking top ten for your one obvious keyword and assuming it covers you, because the engine is no longer only asking your one obvious keyword. Necessary but not sufficient. A real change, and a much narrower one than a second retainer.
Which makes sense once you stop and think about what an AI answer engine is doing. It has to find sources, judge whether they are credible, and pull out the bit that answers the question. Every one of those steps leans on the same signals search engines have used for years. Is the site crawlable. Is the page fast. Does anyone else reference this brand. Is the content specific enough to be useful. Is there a real organisation behind it with a track record.
Nobody built a separate ranking system from scratch. They built an answer layer on top of the one that already existed.
So when an agency tells you GEO is a fundamentally new discipline requiring a fundamentally new budget line, the honest translation is usually that they want to charge you twice for improving the same four things.
The nuances that are actually real
Having said all that, treating the three as identical is lazy in the other direction. There are differences. They are smaller than the pitch decks suggest and they are mostly at the edges, but they change what you make.
SEO is still the floor
Everything else sits on top of it. If your site is slow, badly structured or thin, no amount of AI-era cleverness rescues it, because the engines cannot use what they cannot crawl and will not cite what they cannot verify.
What has changed is that ranking is no longer the finish line. Click-through rate for the top organic position has fallen sharply as AI Overviews and other result features have expanded, with one study of more than 200,000 keywords putting the drop at around 32% year on year, from 28% to 19%. You can hold position one and watch your traffic fall. That is not an SEO failure. It is the surface changing shape underneath a metric that used to be a fair proxy for it.
AEO is mostly about structure
This is the smallest genuine difference and also the easiest to act on.
Answer engines extract passages, not pages. So the question is whether any single chunk of your content can be lifted out and stand up on its own.
In practice that means answering the question in the first two sentences before you explain anything. Writing headings as the questions people actually ask. Keeping one idea per section. Using proper schema markup so the machine knows what it is looking at. Putting the definition, the number or the direct answer near the top rather than building to it over four paragraphs like a piece of journalism.
None of that is new. It is the same advice good editors have given for a century, which is to say the thing you are actually about, at the start, in plain words. It just carries more weight now because a machine is doing the skimming.
GEO is mostly about what happens off your website
This is the biggest real difference, and it is the one most in-house teams underestimate.
Generative engines do not just read your site. They build a picture of your brand from everywhere your brand appears, and then decide whether that picture is coherent enough to repeat. Third-party mentions, directories, industry press, Reddit threads, YouTube, review sites, other people's articles about your sector.
Which means the work stops looking like SEO and starts looking like PR and brand. Are the basic facts about you consistent everywhere they appear. Is there anyone outside your own website saying what you do. Is your data fresh and dated, because a model choosing between your 2026 figures and someone's 2021 figures will take yours. Have you published anything worth quoting, or only marketing copy that says nothing a machine could repeat with confidence.
If nobody else is talking about you, generative engines have nothing to work with. You cannot fix that from inside your own CMS, which is exactly why it is the part that gets skipped.
Your analytics are quietly lying to you
Here is where most of the confusion actually comes from, and it is worth a minute because it changes how you judge whether any of this is working.
Look at your referral traffic from AI platforms and it will look like a rounding error. Conductor's 2026 benchmarking report, built on 3.3 billion sessions across 13,770 domains, put AI referrals at 1.08% of total website visits. Easy to dismiss.
Then look at what those visits do. Semrush's study of 500 high-value topics found the average AI search visitor worth roughly 4.4 times the average visit from traditional organic. Adobe's retail data showed AI-referred shoppers converting 42% better than non-AI traffic in March 2026, having converted 38% worse in March 2025. The whole channel flipped inside twelve months.
The reason is not mysterious. Someone arriving from an AI answer has already had their question answered, compared three options and been pointed at you. Someone arriving from a search result is at the beginning of that process. They are not the same person at the same moment.
And the referral number understates it anyway. Similarweb followed what actually happens after a recommendation and found that brands named by ChatGPT were about two and a half times more likely to get a site visit within a week than brands that were not. Of that traffic, 55.9% arrived through a branded search. Somebody heard the name and went and looked it up rather than clicking anything. In your analytics that is branded organic, or direct. No fingerprints on it at all.
And that is only the standalone assistants. Conductor found an AI Overview on a quarter of the 21.9 million Google searches it sampled, and an AI Overview that answers the question outright generates no referral of any kind. Same invisibility, far bigger surface.
So the dashboard says 1%. The reality is a channel that shapes decisions long before anyone reaches you and then quietly hands you the credit under a different label.
What this means for how you buy it
If the three overlap by roughly ninety percent, buying them as three separate services is a fairly expensive way to do one job badly.
Split the work across a technical SEO agency, a GEO specialist and a content team and you get three roadmaps, three reporting formats and three sets of priorities that quietly contradict each other. The GEO specialist recommends more authoritative content. The content team writes it. Nobody notices the template renders it in a way the crawler struggles with. Everyone reports success against their own metric and the client cannot work out why the numbers have not moved.
The work is one thing. A site that is fast and structured properly. Content that answers real questions clearly and early. A brand that exists visibly outside its own domain. Do those three well and you are optimised for search, for answers and for generative engines simultaneously, because they all want the same evidence.
That is how we run it. One team covering strategy, the website, the content, the brand presence around it and the paid media behind it, with a single view of what is working. Not three retainers pointing at the same problem.
Where paid comes in, and why we would not run organic without it
This is the part we would push hardest on, and it is where a lot of otherwise sensible organic strategies stall.
Everything above is about being found by people who are already looking. That is valuable and it converts well, and it is also inherently limited by how many people happen to be looking this month. Organic has a ceiling and the ceiling is demand.
Paid does the opposite job. It puts you in front of people who were not searching, could not have told you they had a problem, and would never have typed your category into anything. That is where the top of the funnel actually gets filled.
The two work far better together than either does alone.
Warming. Paid social gets the brand in front of the right audience before they need you. When they eventually do search, you are not a stranger competing on price with four other results. You are the name they half recognise, and that changes the click before anyone has spent a penny on SEO.
Nurturing. Your best organic content, the guide that took three weeks to write, gets read by a fraction of the people it deserves. Put budget behind it and it reaches the audience it was built for. Content and paid distribution are not separate departments. One makes the thing, the other makes sure someone sees it.
Remarketing. Almost nobody converts on the first visit. The person who found your article at nine in the evening, read the whole thing and closed the tab is not lost, they are unfinished. Remarketing is what brings them back a few days later, at a moment when they have time to act. Without it, your organic investment is a leaky bucket with good content in it.
Branded search. If AI recommendations are pushing people to search your name, somebody needs to own that result. Often that somebody is a competitor bidding on it. Owning your own branded terms in paid is one of the cheapest defensive positions in marketing and one of the most commonly ignored.
Speed. Organic tells you what works in six months. Paid tells you in six days. Test the message, the angle and the creative with budget behind it, find out which version people actually respond to, then build the organic content around the answer instead of guessing and waiting two quarters to find out.
Run them separately and you get a content team optimising for rankings while a media team optimises for cost per click, and no shared view of whether any of it produced a customer. Run them together and each one makes the other cheaper.
That is the strategy we build with clients, and we run both halves of it. Organic as the engine, paid as the pressure behind it, one set of messaging across both, and creative made once to work everywhere it needs to. The person planning the content and the person spending the media budget are in the same studio, looking at the same numbers, which removes the single most common reason this stops working.
The short version
The acronyms are new. Most of the work is not. Build a fast, well structured site, publish content that answers real questions properly and early, and make sure your brand exists somewhere other than your own website. That covers all three surfaces at once.
Then put budget behind it, or you have built something excellent and left it waiting to be found.
We run organic and paid as one strategy, in house, from the content and creative through to the media spend behind it. Tell us what you are trying to grow and we will come back with what we would do and what it costs.