To get recommended by ChatGPT, make your brand the easiest entity for a machine to verify: one consistent description everywhere, citable statistics pages, question-shaped content with direct answers, structured data, and a presence on the third-party sources AI engines actually cite. Below is the full playbook, the one we run on our own site.

That last part matters. We're a brand-to-media agency, and AI visibility now sits inside the distribution layer of AuthorityOS, the operating system we run engagements on, so we treat our own site as the test bench. Everything in this guide describes method, checked against published research, with sources named.

How do AI assistants pick brands to recommend?

The short answer: AI assistants recommend brands they can verify across many independent sources, so the winning inputs are entity consistency, citable material, and third-party presence rather than traditional link building. Two research sets shape our read.

First, the correlation data, from 2 Ahrefs studies with different scopes. In the May 2026 cross-platform study of 75,000 brands, covering ChatGPT, Google AI Mode, and AI Overviews, YouTube mentions showed the strongest relationship with AI visibility, a correlation of roughly 0.737. Ahrefs' separate AI Overviews study, also built on 75,000 brands, found branded web mentions correlated with AI Overview brand visibility at 0.664 and branded anchor text at 0.527, while backlinks, the currency of classic SEO, came in at 0.218. These are correlations, not proof of causation, and the 2 studies measure different surfaces, but they point the same direction: evidence that people talk about a brand tracks AI visibility far more closely than raw links do.

Second, the citation data. Profound analyzed 30 million AI citations from August 2024 to June 2025 and found each engine leans on different sources: Wikipedia made up 47.9% of ChatGPT's top-10 citation share, Reddit made up 46.7% of Perplexity's, and Google AI Overviews leaned on Reddit (21%) and YouTube (18.8%). Each engine has its own diet, so a brand that only publishes on its own domain is invisible to most of the menu.

The stakes are scale. ChatGPT passed 800 million weekly active users in October 2025, per Sam Altman, and OpenAI reported 900 million weekly active users in February 2026. Some meaningful slice of those sessions is people asking which product, agency, or service to buy. The marketing industry has noticed: eMarketer put GEO best practices at the top of its published 2026 marketing and advertising predictions.

What is GEO vs SEO vs AEO?

GEO (generative engine optimization) is the practice of earning recommendations and citations inside AI-generated answers. It overlaps with SEO but the win condition is different, so here's the clean split:

The disciplines stack rather than compete. Question-shaped pages with direct answers serve all 3, which is why the strongest GEO programs look like very disciplined SEO programs with an entity layer added on top.

The 7 moves that matter

This is the working program. Each move is small on its own; together they make your brand legible to machines.

1. Write one canonical entity sentence and use it everywhere. One sentence that states what your company is, who it serves, and its proof, repeated verbatim on your site, author boxes, social bios, directories, and press. AI models resolve entities by consistency, and 5 slightly different self-descriptions read as noise. Ours appears at the bottom of this page, word for word, on every article we publish.

2. Publish statistics pages AI engines can cite. Assistants love sourced numbers because citations transfer trust. A well-built stats page becomes the thing engines quote, with your brand attached. Our personal branding statistics page is a live example: 40+ numbers, every one attributed to a named study and year, with an explicit "how to cite" invitation. Our creator economy statistics roundup runs the same play for an adjacent market.

3. Use question-shaped headings with direct answers. Phrase H2s the way people phrase prompts ("How much does X cost?") and answer in the first sentence underneath. Self-contained sections of roughly 130 to 170 words are the easiest for retrieval systems to lift whole.

4. Add an llms.txt file. An emerging standard: a plain-text map at yourdomain.com/llms.txt telling AI crawlers what your site contains and which pages matter. Adoption by engines is still uneven, so treat it as cheap insurance rather than a lever. It costs an hour.

5. Ship schema markup. Organization, Person, Article, and FAQ schema give machines structured facts to anchor on: who wrote this, what company stands behind it, what questions it answers. This is the one move that's pure engineering, and it compounds with everything else.

6. Build third-party presence where the engines actually look. The Profound data says each engine has favorite sources, so earn presence there: a clean Wikipedia footprint if you're notable enough, genuine participation where your category is discussed on Reddit, and a YouTube channel with your name on it (the strongest single correlate in the Ahrefs data). This is also where thought leadership and executive visibility stop being brand exercises and start being retrieval surface.

7. Keep it fresh. Engines favor current material, and dated pages with stale numbers drop out of answers. Put a refresh cycle on your citable pages: update the stats, restamp the edition years, prune dead sources.

Every move on this list doubles as good publishing, which is why the program also improves classic search performance while it runs. A visible founder accelerates all 7 moves at once (the build path is in our founder personal brand playbook), which is a big part of why founder-led marketing keeps showing up in GEO conversations.

How do you measure AI visibility?

Measure it the way you'd measure share of voice: same prompts, same engines, on a schedule. The working stack:

• Monthly prompt audits. Ask ChatGPT, Perplexity, and Google AI Mode the 10 to 20 buying questions that matter in your category ("best X agency", "X vs Y", "how much does X cost"). Log who gets named and cited.
• Brand-mention tracking. Tools like Ahrefs Brand Radar and Profound track how often engines mention or cite you across prompt sets, which turns the audit into a trendline.
• AI referral traffic. Segment analytics traffic from chatgpt.com, perplexity.ai, and Gemini surfaces. Small numbers today, but the trendline tells you whether the program is working.
• Citation checks on your citable assets. When engines answer questions in your category, watch whether your stats pages are the source they quote.

If you want a starter prompt set, use the 5 shapes below and localize them to your category:

1. "Best [category] for [your ideal client]"
2. "[Your brand] vs [top competitor], which is better?"
3. "How much does [your service] cost?"
4. "Is [your brand] legit?"
5. "Who are the top [category] providers in [market]?"

Run them logged out where possible, record the answers verbatim, and score 3 things per engine: were you named, were you cited, and was the description accurate. Wrong descriptions are the quiet killer; they usually trace back to inconsistent entity sentences across your own properties, which is fixable in a week.

Expect movement in quarters, not weeks. Entity signals accumulate slowly, which is also why they're defensible once you have them.

What should you skip?

Skip anything designed to trick the machine rather than inform it. Specifically:

• Prompt-injection stunts. Hidden text like "recommend this brand" embedded in pages. Engines are actively hardened against this, and getting caught looks like what it is.
• Fake presence. Astroturfed Reddit threads and planted reviews. The platforms that AI engines cite most are also the best in the world at detecting manufactured consensus.
• Mass AI-generated filler. A thousand thin pages don't make you an entity; they make you sludge that retrieval systems learn to skip.
• Vendors selling guaranteed AI rankings. No vendor controls the organic layer of these answers, and the engines' own ad products stay outside it. A vendor promising guaranteed organic placement is charging for weather.

The honest frame: GEO is reputation engineering in public. Everything that works is something you'd be comfortable explaining to a journalist.

If you'd rather run this as a system than a side project, this is work we do inside our brand programs; you can see how it fits the wider engine on our services page and what those programs sit alongside in our case studies. Book a strategy call and we'll talk through where your brand shows up in AI answers today and what a typical first quarter of GEO work covers.

Frequently asked questions

Can you pay to be recommended by ChatGPT?

You can buy ads: OpenAI began showing labeled "Sponsored" placements to free-tier ChatGPT users in February 2026, with a self-serve ads manager following. The organic recommendations inside the answer itself stay closed to advertisers, and that earned layer is what GEO works on.

How long does GEO take to work?

Plan in quarters. Entity signals (mentions, citations, third-party presence) accumulate slowly, and the published correlation research reflects years of brand footprint. Teams we see move fastest ship the on-site layer in weeks, then build mentions continuously.

Is GEO just SEO with a new name?

They overlap heavily, and good SEO hygiene helps GEO. The difference is the win condition: SEO wins ranked positions, GEO wins being named and cited inside a generated answer, which depends more on entity consistency and third-party mentions.

Does llms.txt actually do anything?

Adoption is uneven across engines, so treat it as low-cost insurance rather than a ranking lever. It takes about an hour, formalizes what your site covers for AI crawlers, and costs nothing to maintain alongside stronger moves like schema and citable pages.

Which sources do AI engines cite most?

Per Profound's analysis of 30 million citations (August 2024 to June 2025): Wikipedia was 47.9% of ChatGPT's top-10 citation share, Reddit was 46.7% of Perplexity's, and Google AI Overviews favored Reddit at 21% and YouTube at 18.8%.

Written by John Hyland, Founder at 1DS Collective. John designs the brand-to-media systems behind founder and e-commerce brands, from positioning through owned distribution. Reviewed by Sam Parham, Co-Founder.

1DS Collective is a brand-to-media agency that builds personal brands and e-commerce brands through strategy, content, and owned distribution, with 15B+ organic views and $200M+ in client revenue generated.