GEO: How to Get Recommended by ChatGPT and AI Search
A step-by-step GEO guide: find buyer questions, get into the pages AI assistants cite, build comparison pages, add structured data and measure results.
The Toby team··8 min read
On this page
- How AI assistants decide who to recommend
- Step 1: List your buyer-intent questions
- Step 2: Measure your starting point
- Step 3: Get into the sources AI assistants cite
- "Best [category]" articles and listicles
- Review platforms and comparison sites
- Forums and Reddit
- Media, podcasts and partner content
- Step 4: Make your own pages easy to quote
- Step 5: Add structured data that matches the page
- Step 6: Check that assistants can read your site
- Step 7: Measure what changes
- What not to do
- A 30-day GEO checklist
- How Toby does this for you
More and more buyers start with a question to an AI assistant: "What's the best invoicing app for freelancers?", "Alternatives to [leader] for a small team?", "Which candle brands use natural wax?". The assistant answers with a short list of brands. If you're not on that list, you're not in the running.
GEO (generative engine optimization) is the work of getting your brand into those answers. It overlaps heavily with SEO, but the emphasis shifts: from ranking one page to being named across the sources AI assistants read, and from keywords to the exact questions buyers ask.
This guide gives you a process you can run yourself, with templates.
How AI assistants decide who to recommend
When ChatGPT search, Perplexity, Claude with web search, Microsoft Copilot or Google's AI Overviews and AI Mode answer a buying question, they generally search the web, read a set of pages, and summarize what those pages say, often with links to their sources. Three consequences follow:
- Classic search visibility still matters. Google states there are no additional requirements or special optimizations needed to appear in AI Overviews or AI Mode beyond the usual SEO fundamentals (Google Search Central).
- Third-party pages matter as much as yours. If the pages an assistant reads ("best X" lists, comparison sites, review platforms, forum threads) don't mention you, the answer probably won't either.
- Quotable content helps. In a 2023 research paper that coined the term GEO, researchers found that adding citations, quotations and statistics to content improved its visibility in generative engine answers by up to 40% on their benchmark (Aggarwal et al., arXiv). That's a lab result, and effects varied by domain, but the direction is useful: clear facts beat vague prose.
Step 1: List your buyer-intent questions
Write 15 to 25 questions a buyer would ask an assistant without knowing your brand. Use these six patterns:
| Pattern | Example (invoicing app) |
|---|---|
| Best [category] for [audience] | Best invoicing app for freelance designers |
| [Category] for [specific need] | Invoicing software that adds sales tax for clients in other states |
| Alternative to [leader] | Cheaper alternative to [Leader] for a one-person business |
| [You or rival] vs [rival] | [Rival A] vs [Rival B] for freelancers |
| How to solve [problem] | How do I get clients to pay invoices on time? |
| Is [category/product] worth it | Is paid invoicing software worth it for a freelancer? |
Where to find the real wording: sales call notes, support tickets, "how did you hear about us" answers, Search Console queries (filter for queries starting with how, what, best, which), Reddit threads in your niche, and the "People also ask" boxes on Google.
Step 2: Measure your starting point
For each question, ask each assistant your buyers use, and record what comes back. Use a fresh, logged-out session where possible, and run important questions more than once: answers vary between runs.
Copy this tracking table:
| Question | Assistant | Brands named (in order) | Are we named? | Sources cited |
|---|---|---|---|---|
| Best invoicing app for freelancers | ChatGPT | |||
| Best invoicing app for freelancers | Perplexity | |||
| Best invoicing app for freelancers | Google AI Mode |
After 15 questions across 3–4 assistants, two things become clear: which competitors keep coming up, and which sources the assistants keep citing. That second column is your to-do list.
Step 3: Get into the sources AI assistants cite
Sort the cited sources into types, then work through each type.
"Best [category]" articles and listicles
These are often the backbone of AI answers to "best X" questions.
- List the articles that rank in the top 10 for your main "best [category]" searches and that are cited by assistants.
- Check if you're mentioned. If not, find the author and whether the page is updated regularly (a "last updated" date is a good sign).
- Pitch with something useful for their readers, not a request for a favor.
Pitch template:
Subject: A [category] option for [audience] your list doesn't cover yet
Hi [Name], I read your "[article title]" (updated [month]). It's one of the clearest comparisons I've seen for [audience].
One gap your readers might care about: none of the options listed [does specific thing, e.g. "adds the right sales tax for out-of-state clients automatically"]. That's what we built [Product] for. Happy to give you a free account to test it, plus [screenshot / pricing / a customer you can talk to].
If it doesn't fit your list, no problem at all.
[Your name], founder of [Product]
Some publishers charge for inclusion. If you pay, the listing must be disclosed as sponsored. Prefer the honest editorial route.
Review platforms and comparison sites
G2, Capterra, Trustpilot, app marketplaces, Google Business Profiles, and niche directories in your industry are frequently cited. Ask real customers for honest reviews at the right moment (after a success, not after a support ticket). Don't offer rewards conditioned on a positive review, and check each platform's rules: Google, for example, doesn't allow businesses to offer incentives in exchange for reviews (Google's policy).
Forums and Reddit
Reddit threads are often cited for "what do you recommend" questions. The only approach that works long term is being genuinely helpful:
- Answer the actual question first, including options that aren't yours when they fit better.
- Disclose your affiliation every time ("I'm the founder of X, so I'm biased").
- Read and follow each subreddit's rules on self-promotion. Never use fake accounts or coordinate votes.
Media, podcasts and partner content
Industry newsletters, podcasts with show notes, integration partners' blogs and "made with [tool]" showcases all create mentions that assistants can find.
Step 4: Make your own pages easy to quote
Assistants need pages that answer the question directly. Build these first:
- "[Leader] alternative" pages for each major competitor buyers compare you with.
- "[Category] for [segment]" pages for each main audience (freelance designers, accountants, agencies…).
- Comparison pages ("[You] vs [Rival]") that are fair: state where the rival is better.
- A clear facts block on your homepage and pricing page: who it's for, what it does, price, key integrations, where you operate.
- FAQ sections where each answer starts with a one-sentence direct answer, then the detail.
Alternative page structure (copy it):
- H1: "The [Leader] alternative for [audience]"
- Two-sentence summary: who should switch, and who should stay with [Leader]
- Comparison table: price, key features, support, contract, data export
- Three reasons customers switch (with a customer quote if you have one, with permission)
- Migration steps: how long it takes, what's imported
- FAQ: price, cancellation, data, the top objection
- Last updated date
Write facts that can be checked: "starts at $12/month", "imports from [Leader] CSV in about 10 minutes" (if true), not "affordable and seamless".
Step 5: Add structured data that matches the page
Structured data (schema.org, in JSON-LD) helps search engines understand who you are and what you sell. The most useful types for most businesses:
- Organization (name, logo, website, social profiles) on the homepage
- Product with Offer for e-commerce, or SoftwareApplication for software
- Article with author and dates for guides
- BreadcrumbList for site structure
- Review / AggregateRating only for real reviews shown on the page
A minimal Organization example:
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Example Co",
"url": "https://www.example.com",
"logo": "https://www.example.com/logo.png",
"sameAs": ["https://www.linkedin.com/company/example-co"]
}
Two rules. First, markup must describe content that's visible on the page (Google's structured data guidelines). Second, don't expect FAQ markup to produce rich results: Google restricted FAQ rich results to a small set of authoritative government and health sites in 2023 (Google Search Central blog). The FAQ content is still worth having, because it answers buyer questions in plain text.
Step 6: Check that assistants can read your site
- Don't block the search crawlers of AI assistants in your robots.txt if you want to be cited. OpenAI says sites that block OAI-SearchBot won't appear in ChatGPT search answers (OpenAI publishers FAQ). Anthropic documents Claude-SearchBot and Claude-User, and says restricting them can reduce your visibility in Claude's search results (Anthropic).
- Training and search are separate. You can block training crawlers (such as GPTBot or ClaudeBot) and still allow search crawlers. For Google, the Google-Extended token governs Gemini training and grounding, not inclusion in Google Search.
- Put key content in the HTML. If your pricing or product descriptions only appear after JavaScript runs, some crawlers may miss them.
- Optional extras aren't shortcuts. Files like llms.txt are a proposed convention; Google says its AI features need no special files or markup. Spend your time on steps 3 and 4 first.
Step 7: Measure what changes
GEO results are slower and noisier than paid ads. Measure monthly:
- Share of answers: re-run your question set. Track "named in X of Y answers" per assistant, and which competitors appear.
- Referral traffic: in GA4, look at sessions from chatgpt.com, perplexity.ai, claude.ai, gemini.google.com and copilot.microsoft.com. ChatGPT adds
utm_source=chatgpt.comto links it sends (OpenAI). - Google: clicks from AI Mode are counted in the Search Console Performance report, and since June 2026 Search Console also has dedicated views of impressions in generative AI features (Google Search Central blog).
- Bing and Copilot: Bing Webmaster Tools has an AI Performance report showing which of your pages are cited in Copilot and Bing's AI answers.
- Ask your customers: add "AI assistant (ChatGPT, Perplexity…)" as an option to your "How did you hear about us?" question. Many AI-driven visits arrive later as direct or branded searches.
What not to do
- Publish fake reviews or pay for positive ones.
- Post on Reddit from accounts that hide who you are.
- Hide text on pages "for AI" that humans can't see.
- Publish a "best [category]" list on your own site that ranks you first without saying it's yours.
- Change everything at once: you won't know what worked.
A 30-day GEO checklist
- Week 1: write 20 buyer questions; run the baseline table on 3–4 assistants; list the 15 most-cited sources.
- Week 2: publish or fix your top "alternative" page and one "[category] for [segment]" page; add a facts block and FAQ to pricing.
- Week 3: pitch the 10 most-cited list authors; ask 10–20 happy customers for honest reviews on the platform assistants cite most.
- Week 4: add Organization and Product/SoftwareApplication markup; check robots.txt; set up the GA4 referral view; re-run the question set.
GEO and SEO share the same foundations, so the Search Console quick wins are a good companion to this list.
How Toby does this for you
- AI visibility test: Toby's growth audit writes the three questions a buyer in your market would ask an AI assistant, searches the web for each, lists the brands that come up, and tells you whether you're mentioned and which competitors appear instead. It also flags missing structured data.
- Leo, the SEO and GEO agent, prepares the pages: alternative and comparison pages, segment pages, facts blocks and FAQ sections, ready for you to review. With WordPress connected, he can publish them once you approve.
- Noah (affiliation and partnerships) lists the "best X" articles you're missing from and drafts a personalized pitch for each author.
- Ray (Reddit) drafts helpful answers that disclose your affiliation, for the threads where your buyers ask for recommendations. You review and post them yourself.
- Iris (data) reads your Google Analytics data (visits, conversions and revenue by channel) once it's connected, so you can see whether the work moves anything.
Run the free growth audit to see which AI answers you're in today.



