What is GEO (generative engine optimization)? A practical guide for small teams

Generative engine optimization (GEO) is the work of making your brand easy for AI assistants to find, understand and name when someone asks a buying question. It builds on SEO rather than replacing it: the pages assistants retrieve still decide most of the answer. For a small team, GEO comes down to a short loop: measure which questions name you, fix the pages and third-party sources behind the gaps, and re-measure.

This post is the practical version. If you only want the definition, see our GEO guide and the glossary entry. Here we cover what a team of one to ten people can realistically do in a few hours a week, and what the evidence says about each step.

What does GEO actually optimize for?

Classic SEO optimizes for a position in a list of links. GEO optimizes for being named inside a generated answer, ideally with a link to your site as a cited source. Three outcomes matter:

• Mention: the assistant writes your brand name in its answer.
• Position: where you appear if it names several brands.
• Citation: your page is shown as a source for a statement.

The term comes from a research paper by Aggarwal et al., "GEO: Generative Engine Optimization", accepted at KDD 2024. The authors built a benchmark of diverse queries and tested content changes against generative engines, including Perplexity.ai. They reported that some changes raised visibility by up to 40 percent in their setup. That is a research result on a benchmark, not a promise for your site, but it showed that how a page is written affects whether an engine uses it.

Is GEO different from SEO?

Mostly it is SEO plus a new measurement problem. Google states that to appear in AI Overviews and AI Mode a page must be indexed and eligible to show with a snippet, and that there are no additional technical requirements or special optimizations. Other assistants run their own crawlers and search indexes, and OpenAI, Perplexity and Anthropic each document how theirs behave (links further down).

What does change:

Why can't I just ask ChatGPT once and see?

Because one answer is an anecdote. A January 2026 study by SparkToro and Gumshoe.ai ran 12 recommendation prompts nearly 3,000 times across ChatGPT, Claude and Google's AI features and found that the same list of brands came back less than 1 percent of the time (Search Engine Journal summary, original report). The authors note the work is not peer reviewed and that Gumshoe.ai sells tracking tools. Their own takeaway is more careful than the headline: exact rankings are mostly noise, while how often a brand shows up across many runs can still be informative.

So the unit to track is a rate over many runs, not a screenshot. That is also why weekly tracking beats a one-off test.

What should a small team do first?

A routine that fits into a few hours a week:

1. Write 15 to 25 buyer questions. Category ("best invoicing tool for freelancers"), problem ("how to stop sending invoices late") and comparison ("Brand A vs Brand B"). Use the language and country of your buyers.
2. Measure. Count in how many answers per assistant your brand appears, at what position, and which domains are cited. Our guide to tracking AI visibility walks through the manual way.
3. Make sure assistants can reach you. Check that robots.txt does not block the search crawlers you want (see below) and that your key pages render without JavaScript. Our free AI crawler checker reads your robots.txt in a minute; whether the text is in your HTML you can test with curl.
4. Put the answer first on your key pages. State what the product is, who it is for and how it compares, in plain sentences near the top. Answer-first pages are easier to quote.
5. Fix the third-party gap. Look at the sources cited for questions where competitors are named and you are not. Often these are roundups, comparisons and community threads, not your own site.
6. Re-measure weekly. Compare against the same questions and the same named competitors.

Which tactics have evidence behind them, and which don't?

Honest answer: the evidence is thin and mostly from vendors or small studies. What we can point to:

• Citing sources, adding quotations and adding statistics improved visibility by roughly 30 to 40 percent on one metric and 15 to 30 percent on another in the GEO paper's benchmark, while keyword stuffing offered little to no improvement there. In the same paper, a lower-ranked site gained more from citing sources than a top-ranked one did. Treat it as a hint about writing style, not a guarantee.
• Third-party coverage matters. In a May 2026 study by the agency DerivateX (40 B2B SaaS categories, one buyer-intent prompt each, run ten times in ChatGPT with web search), only 11.6 percent of recommendations cited the recommended tool's own site, and 81.9 percent of citations went to blogs and vendor content. It is a modest sample from a company that sells GEO services, so read it as direction (report).
• Schema and llms.txt: Google says it does not require special markup or AI text files for its AI features. For llms.txt, independent evidence of benefit is missing; see our llms.txt guide and the checklist post linked below.

No one can honestly promise you a rank inside an assistant. Anyone who does is selling something.

Which AI crawlers should I allow?

Search-oriented crawlers decide whether you can show up in answers; training crawlers are a separate choice. The vendors document this:

OpenAI says its settings are independent, so you can allow OAI-SearchBot and disallow GPTBot (OpenAI bots). Perplexity describes PerplexityBot as designed to surface and link sites in search results (Perplexity). Anthropic lists its three bots and says all honor robots.txt (Anthropic). Whether to allow training crawlers is a business decision; blocking the search crawlers keeps you out of those assistants' search answers.

What do we measure, and how can you check?

Nodpick asks the questions your buyers ask and records for each answer whether your brand was named, in which position and which domains were cited. Paid plans cover ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity and Claude (Grok, Mistral, DeepSeek and Kimi as an add-on or on Agency). A summary score gives 100 points for first place, 80 for second, 60 for third, 40 for later and 0 when absent. See the method page for details. Answers vary, so we report results over repeated runs and say plainly that they are estimates. We have no case studies to show yet; we will publish our own measurements when we have them.

The free check opens with the launch. It will need no sign-up, ask 5 questions to ChatGPT, Perplexity, Gemini and Claude, once each, and show your score and the first three recommendations; join the waitlist and we email you once, the day Nodpick opens. Weekly tracking, the Google engines and the full recommendation list are the paid layers.

FAQ

Is GEO the same as AEO? The terms overlap. AEO (answer engine optimization) is often used for answer boxes and voice, GEO for generative assistants. See AEO and GEO.

How long until I see results? Nobody can say. Crawlers need to re-fetch changed pages (OpenAI mentions about 24 hours for robots.txt changes to affect its search; Google says recrawling after preview-control changes can take days to months). Third-party sources move on their own schedule.

Do I need a tool? No. You can run the routine above by hand with a spreadsheet. A tool saves time and runs more repetitions than most people do manually.

Can I pay to be recommended? Not by buying a ranking from an assistant. Be skeptical of any service that guarantees placement.

Is GEO just SEO? Google's own guidance says optimizing for generative AI search is still SEO and that no special markup or AI text files are needed for its features. The GEO-specific part is measuring what assistants say and fixing the third-party sources they quote. More on what is and is not known: how AI assistants decide.

Sources

• Aggarwal et al., GEO: Generative Engine Optimization (arXiv, KDD 2024)
• Google Search Central: AI features and your website
• SparkToro / Gumshoe.ai study via Search Engine Journal
• DerivateX B2B SaaS AI citation study
• OpenAI crawler documentation
• Perplexity bots documentation
• Anthropic crawler documentation

Next: how AI assistants decide which brands to recommend.