Generative Engine Optimization (GEO): What It Is and What Works
Generative engine optimization (GEO) is optimising content to be cited in AI answers. What the research shows, how it differs from SEO and where links fit.
Generative engine optimization (GEO) is the practice of shaping your content and your brand's presence on the web so that AI systems such as Google's AI Overviews, ChatGPT and Perplexity cite or mention you in their answers. It's also called AI search optimisation, and it overlaps heavily with classic SEO.
The term comes from a 2023 research paper, "GEO: Generative Engine Optimization", by researchers including Pranjal Aggarwal and Karthik Narasimhan, later presented at the KDD 2024 conference. It tested which content changes made a source more visible in AI-generated answers.
What that research found, and what has been learned since, is more useful than most of the GEO advice now being sold.
What the original GEO research found
The paper built a benchmark of 10,000 queries (GEO-bench) and measured how much of an AI answer drew on a given source before and after edits to that source. Its headline finding was that the best methods improved visibility by up to 40%.
| Method tested | Reported improvement | What it means in practice |
|---|---|---|
| Quotation addition | About 41% | Including credible quotes from relevant people or sources |
| Statistics addition | About 40% | Adding specific, sourced numbers |
| Cite sources | About 30% | Referencing reliable sources in the text |
| Fluency optimisation | About 15 to 30% | Clearer, better-written prose |
| Keyword stuffing | Little or no improvement | Classic keyword tactics didn't help |
The authors also tested on Perplexity, a live AI search engine, and saw similar gains. The effect varied by topic, so the results are best read as direction, not a formula.
One caveat is worth stating plainly. The study measured how prominent a source was once it had already been retrieved. It didn't measure what gets a page retrieved in the first place, which is where rankings, links and brand reputation come in.
GEO vs SEO: what's different?
| Classic SEO | GEO | |
|---|---|---|
| The goal | Rank a page and win the click | Be cited or mentioned in the answer |
| The unit | A page for a keyword | A brand or fact across many related queries |
| How success shows up | Rankings, clicks, traffic | Citations, brand mentions, share of AI answers |
| Measurement | Search Console, rank trackers | Specialist AI visibility tools, still immature |
| Official guidance | Extensive Google documentation | Google says no special optimisation is needed for its AI features |
What Google says about optimising for AI search
Google's documentation on AI features is clear that there's no separate GEO rulebook for its products. It says there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary", and that you don't need new machine-readable files, AI text files or special schema markup.
A Google Search Central blog post from May 2025 gives the same advice it has always given: focus on unique, non-commodity content, a good page experience, technical accessibility, and structured data that matches visible content. Our guide to AI Overviews covers Google's side in detail.
Other AI systems publish less. OpenAI and Perplexity document their search crawlers and say you need to allow them for your site to appear in their search answers, but neither publishes how sources are chosen.
Where do links fit in GEO?
Links matter in two ways, one well established and one less so.
The established part is that AI search systems retrieve pages through search. Ahrefs found in July 2025 that 76% of pages cited in AI Overviews ranked in Google's top 10 for the same query, and although an update in March 2026 put that figure at 38%, most cited pages still ranked somewhere in Google's results. Rankings still depend partly on backlinks, so links help you become a candidate.
The less certain part is whether links directly influence what a model says. In Ahrefs' study of 75,000 brands, backlinks correlated with AI Overview brand visibility at 0.218, while branded web mentions correlated at 0.664. That's correlation, not causation, but it suggests that being talked about matters more than link counts alone. Our brand mentions guide goes deeper.
A practical GEO checklist
- Make sure AI crawlers can reach you. Allow Googlebot, OAI-SearchBot and PerplexityBot in robots.txt if you want to appear in those products
- Write quotable content. Specific facts, sourced statistics and clear definitions near the top of the page
- Cover the subtopics. AI systems run several related searches for one question, so answer the follow-ups too
- Earn mentions, not just links. Reviews, comparisons, interviews and digital PR all put your brand in the text AI systems read
- Keep building links that help you rank. Classic link building still feeds the retrieval step
- Measure carefully. AI answers vary between runs, so track trends across many prompts rather than single checks
What GEO can't do
There's a lot of confident GEO advice that runs ahead of the evidence. Nobody outside the AI companies knows exactly how sources are chosen, citation patterns change quickly (Semrush recorded big swings in which domains ChatGPT cited during 2025), and no tool can guarantee a mention.
The parts that are well supported, being findable, being useful and being talked about, are the same things that have always worked in SEO.
The short version
- Generative engine optimization means getting cited or mentioned in AI answers, and it overlaps heavily with SEO
- The original GEO research found quotes, statistics and citations boosted visibility by up to 40%, while keyword stuffing didn't help
- Google says no special optimisation is needed for AI Overviews or AI Mode
- Links help pages rank and get retrieved, but brand mentions correlate more strongly with AI visibility in published studies
- Treat GEO claims with care: much of how AI systems choose sources isn't public