How to get your brand cited in AI Overviews, ChatGPT and Perplexity
What generative engine optimisation (GEO) really involves: AI crawler access, citable content, entity consistency and off-site mentions, and how to measure it.
A growing share of searches now end with an AI-generated answer: Google’s AI Overviews and AI Mode, ChatGPT search, Perplexity and Microsoft Copilot. These answers cite sources. Being one of those sources is the new version of ranking on page one.
The good news is that generative engine optimisation (GEO) isn’t a separate discipline. It builds on good SEO. But it does shift the emphasis in four important ways.
1. Make sure AI systems can actually read your site
AI search tools rely on crawlers, and many sites block them without realising it, through robots.txt rules, CDN bot protection or firewall settings. Check access for the crawlers that power AI search experiences specifically, and treat training crawlers as a separate decision.
Search crawlers and training crawlers
| Crawler | What it feeds | Our default |
|---|---|---|
| Googlebot | Google Search, including AI Overviews and AI Mode | Allow |
| Bingbot | Bing search and Microsoft Copilot | Allow |
| OAI-SearchBot | The sources shown in ChatGPT search | Allow |
| PerplexityBot | Perplexity’s search index | Allow |
| GPTBot | Training OpenAI’s models | Your call |
| Google-Extended | A robots.txt token for Gemini models; it doesn’t affect Search | Your call |
You can allow search crawlers while restricting training crawlers: they are separate user agents with separate robots.txt rules.
Access also means content the crawler can see. Many AI crawlers fetch the HTML without running JavaScript, so text that only appears after scripts run may be invisible to them. Our guide to what crawlers actually see on JavaScript sites explains how to check.
Quick check: open your robots.txt, then ask your hosting or CDN provider whether “AI bot” blocking is switched on. It often is by default.
2. Write passages that can be quoted on their own
AI systems pull out passages, not whole pages. A paragraph that only makes sense after reading three others is unlikely to be cited. Make every key passage self-contained, answer-first, specific and well structured.
Hard to quote vs easy to quote
Hard to quote
- The answer arrives in paragraph four
- “It” and “this” instead of the subject’s name
- Vague claims such as “competitive” or “industry-leading”
- Key facts locked in images, tabs or PDFs
Easy to quote
- The direct answer in the first sentence
- The subject named in every key passage
- Specific numbers, thresholds and dates
- Descriptive headings, short lists and tables
For example, “Our pricing is competitive” will never be quoted. “SEO retainers at Serpevo start from £950 per month for local businesses” can be.
3. Be consistent about who you are
AI models build an understanding of your brand from everything written about it. Inconsistent descriptions, outdated addresses or vague positioning lead to vague or wrong answers.
- Use
Organizationstructured data with your official name, logo, contact details andsameAslinks to your official profiles. - Describe your company the same way on your website, LinkedIn, directories and review platforms.
- Keep an up-to-date About page that clearly states what you do, for whom and where.
Start by asking ChatGPT, Perplexity and Copilot to describe your company, and note anything wrong or outdated. Errors usually trace back to an old directory listing, a stale press release or an inconsistent profile, and those are things you can fix.
The trust signals that help human visitors help here too: named experts, verifiable company details and real reviews. We covered them in E-E-A-T for service businesses.
4. Earn mentions where AI answers look
For commercial questions such as “best SEO agency for e-commerce in the UK”, AI answers lean heavily on third-party sources: review platforms, industry publications, comparison articles and community discussions. Your own website alone is rarely enough.
This is where digital PR and reputation management meet GEO. Being mentioned, and ideally linked, by trusted publications in your category increases the chance that AI systems include you. Original data helps most, because a specific new fact is exactly what an answer needs to quote.
How to measure AI visibility
Rankings don’t tell you whether you’re cited. We track AI visibility like this:
- Define a fixed set of 50–200 prompts your buyers actually ask, across awareness, comparison and purchase stages.
- Run them regularly in each AI platform and record whether your brand is mentioned, cited or linked, and which competitors appear.
- Track share of voice over time, alongside referral traffic from AI platforms in GA4.
For Google specifically, Search Console’s Search Generative AI performance report now shows which of your pages earn impressions in AI Overviews and AI Mode. It counts impressions only, not clicks, and those impressions are already part of your normal Web totals, so don’t add the two together.
Share of voice across a fixed prompt set
AI answers vary from one run to the next, so a single check proves little. Keep the prompt set stable for at least a quarter before you change it. The trend across a stable prompt set is what tells you whether the work is paying off. For the analytics side, see measuring SEO in GA4 and Search Console.
What about llms.txt?
llms.txt is a proposed file for summarising your site for language models. It’s cheap to add, but no major AI search engine currently relies on it and Google has said it doesn’t use it. Treat it as optional and focus on the fundamentals above.
Curious how often AI tools mention your brand today? Our free SEO audit includes an AI visibility snapshot, and our AI search optimisation service takes it from there.