What AI search means for the traffic you already have.
In this article
If you’ve looked at your Google Search Console recently and noticed impressions holding steady while clicks drift down, you’re not imagining it. Something has changed in how people find businesses, and it’s worth understanding properly before you decide what to do about it.
The short version is that search engines have moved from being directories to being answer machines. The longer version is what this article is about.
What’s actually changed
For most of the last twenty years, search worked the same way. Someone typed a few words into Google, got ten blue links, clicked one or two and made up their mind. Your job was to be one of those links, ideally near the top.
That still happens. But a growing share of searches now end without a click at all. Google shows an AI Overview at the top of the page that answers the question directly. People ask ChatGPT or Perplexity instead of Google in the first place. Voice assistants read out a single answer. In each case, the person gets what they asked for without visiting anyone’s website.
This matters for two reasons. First, the traffic you’ve been earning from informational searches is being skimmed off before it reaches you. Second, and more importantly, when someone asks an AI tool “who’s a good signage company near Farnborough” or “which accountants specialise in ecommerce”, the tool draws up a shortlist. If your business isn’t structured in a way those systems can find, understand and trust, you don’t make the list. And nobody tells you that you’ve been ruled out.
Three kinds of search, one set of foundations
You’ll hear three terms used, often loosely. They describe different things, and it helps to keep them apart.
SEO (search engine optimisation) is the one you know. Earning a position in a list of results by having a site that search engines can crawl, content that matches what people search for and enough authority to be ranked above the competition.
AEO (answer engine optimisation) is about being the answer rather than one of the links. Featured snippets, People Also Ask boxes, AI Overviews and voice results all pull a direct answer from somewhere. AEO is structuring your content so that it’s the one chosen.
GEO (generative engine optimisation) is about how your business appears, or doesn’t, in answers generated by tools like ChatGPT, Perplexity and Gemini. When someone asks for a recommendation, is your business referenced? Is your expertise recognised? Or does the tool simply not know you exist?
Here’s how they compare in practice.
| SEO | AEO | GEO | |
|---|---|---|---|
| Where you appear | A ranked list of results | Inside a direct answer on the results page | Inside an AI-generated response or recommendation |
| What the user does | Clicks through to your site | Often gets the answer without clicking | Gets a shortlist or recommendation, may or may not click |
| What it rewards | Relevance, technical health, authority | Clear, directly structured answers to specific questions | Consistent facts about your business, credible mentions, depth of expertise |
| Main signals | Keywords, links, page quality, site speed | Question-led headings, concise answers, schema markup | Brand mentions across the web, reviews, expert authorship, clear positioning |
| What you measure | Rankings, organic traffic, enquiries | Snippets and overviews won, impressions, brand searches | How often and how accurately you’re named in AI answers |
| Can anyone guarantee it | No | No | Definitely not |
The thing to take from the table is that these aren’t three separate projects. They share the same foundations. A site that’s technically sound, clearly positioned, well structured and trusted across the web does well in all three. A site that’s weak in those areas does badly in all three. The work overlaps far more than the acronyms suggest.
What this means for the traffic you already have
It’s tempting to read all this as “SEO is dead” or “traffic is going to collapse”. Neither is true, but the shape of your traffic is changing, and it pays to know which parts are at risk.
Informational traffic is most exposed. If a lot of your organic visits come from people asking general questions (“how long does a van wrap last”, “what’s the difference between a brochure and a catalogue”), expect those clicks to keep falling. The answer is being given on the results page, and often that’s fine. Those visitors were rarely close to buying anyway.
Commercial and local traffic is more resilient. Someone searching for “van wrap quote Aldershot” or “ecommerce web agency Hampshire” still needs to visit a website, compare options and get in touch. AI tools can shortlist, but they can’t quote, so the click still happens. The question is whether you’re on the shortlist.
Brand searches become more important. If an AI tool names you, the next thing a lot of people do is search for your business by name. A rise in branded searches is often the first sign that GEO work is paying off, even before you see it in the tools themselves.
So the right response isn’t to panic about volume. It’s to protect the traffic that leads to revenue, accept that some informational traffic will go, and make sure you’re being recognised in the places where decisions are now being made.
How to build for AEO
Most businesses don’t structure their content to be an answer. They write about themselves, their services, their history. They don’t think about the specific questions a prospect is asking when they’re trying to solve a problem.
Three things make the biggest difference.
Structure around questions. Work out the actual questions your prospects ask (your sales team and your inbox are the best sources) and build content that answers each one directly. Use the question as a heading. Put the answer in the first sentence or two underneath it, then add the detail. Search engines and AI tools can lift that cleanly.
Add schema markup. Structured data tells search engines what each piece of content is. FAQ schema, article schema, product schema, local business schema. It’s not glamorous, but it’s the difference between a system understanding your page and guessing at it.
Prioritise clarity. Short paragraphs, plain language, bullet points where they help. If a paragraph needs reading twice, an AI system will skip it for one that doesn’t.
How to build for GEO
GEO is newer and less predictable, but the signals that AI models rely on are becoming clear.
Authority signals. Models look for credibility. Expert authorship (a real person with a real name and relevant experience), high-quality links from relevant sites and genuine depth in your subject. A site with twelve thin pages about everything loses to a site with forty detailed pages about one thing.
Brand citations. Your business needs to be mentioned consistently across directories, industry sites, review platforms and forums, with the same name, the same description and the same details everywhere. AI tools cross-reference. Inconsistency reads as unreliability.
Clear positioning. If your messaging is sharp, a model can categorise what you do and who you’re for. “A commercially focused marketing agency in Hampshire for SMEs and B2B suppliers” is easy to file. “A full-service creative partner for ambitious brands” could be anyone, so it becomes no one.
Monitoring. Pick a fixed set of questions your buyers would ask, run them across the main AI tools every month and record who gets named and how you’re described. It’s the only way to know whether anything is changing.
Fix the foundations before chasing the acronyms
One honest caution. There’s a lot of noise about AEO and GEO at the moment, and some of it is agencies selling the same SEO work under a new name. The reality is that if your site is slow, your pages are thin, your reviews are sparse and your positioning is vague, no amount of schema markup will get you into an AI answer. The modern layer sits on top of the foundations. It doesn’t replace them.
So the order of work is roughly this:
- Technical health. Can search engines crawl, read and load your site properly? If not, nothing else works.
- Positioning and messaging. Is it obvious what you do and who you do it for? This feeds everything downstream.
- Content that answers questions. Rework your key pages around what buyers actually ask. Add FAQs with real answers.
- Trust and citations. Reviews, case studies, consistent listings, mentions on relevant sites.
- Structured data and monitoring. Mark it all up and start tracking how you appear.
Most of that is good SEO practice anyway. The difference is the emphasis. Fewer, deeper pages. Questions as headings. Facts about your business that are identical everywhere they appear.
Frequently asked questions
Is SEO still worth doing if AI answers are taking the clicks?
Yes. AI tools and answer engines draw on the same signals that good SEO builds. A site that ranks well is far more likely to be cited in an answer than one that doesn’t. SEO is the foundation for both.
Can you guarantee we’ll appear in ChatGPT or Google’s AI Overviews?
No, and be wary of anyone who says they can. Nobody controls what an AI model says. What you can do is improve the factors that make inclusion more likely and track the change over time.
How do we measure GEO?
Run a fixed set of buyer questions through the main AI tools on a schedule and record how often and how accurately your business appears. Watch branded search volume alongside it.
Does local SEO help with AI answers?
A lot. When someone asks an AI tool for a recommendation in an area, it draws on Google Business Profiles, reviews and local listings. Keeping those complete and accurate does double duty.
Should we rewrite the whole website?
Rarely. Most of the gain comes from restructuring a handful of key pages, adding focused answer content and fixing the technical issues underneath.
Want to know how visible you are?
If you’d like a straight view of where you stand, our free 5-minute marketing audit scores your business across seven foundations, Google visibility among them, and tells you what to fix first. Or talk to us about search and we’ll look at your market together.




