How to rank in AI search and SEO

1 The bottom line

Get cited by showing up where AI already looks: Reddit threads, top 10 listicles and free tools. This matters because AI answers pull from those sources and send free traffic, while classic SEO alone is slowing.

2 Key takeaways

• Start by finding which sources AI cites for your prompts. There is no real prompt volume data, so you need a proxy and a source list.
  • A query with 15,000 Google impressions a month implies about 5,000 AI search impressions at a 30% rate Matt Espinoza, Clover Labs21:55.
• Get into the community threads and comments that AI already cites. One cited thread can rank you for many prompts at once.
  • One Reddit URL was cited for five search queries across three AI channels Matt Espinoza, Clover Labs32:30.
• Get included on "top 10" listicles, and publish your own. AI engines lean heavily on that format.
  • GPTZero shows up consistently when users ask ChatGPT for the best AI detector Edward Tian, GPTZero16:50.
• Build free tools that lead into your product. Tools rank well in AI search and turn visitors into users.
  • Tools tend to convert slightly better than blog posts Ethan Ethier, Built With Science36:45.
• Make the product good enough to keep users. Retention and quality decide what gets recommended.
  • OpenAI uses these signals to decide which plugins to surface to its 1.2 billion users Tibo Sottiaux, OpenAI13:53.

you essentially get automatically be ranking on that particular search query

Matt Espinoza, Clover Labs

2.1 Find the prompts and sources that matter

Measure demand with Google data, then see which sources AI cites for each prompt.

• Build a prompt list and estimate its volume from Google. Source: Matt Espinoza, Clover Labs21:55
  • Write a list of around 2,000 prompts.
  • Pull each query's Google volume from a tool like Ahrefs or Search Console.
  • Assume about 30% of that volume happens in AI search.
  • Clover generates about 150 million views from AI search across its customers.
• Check which sources get cited for each prompt. Source: Matt Espinoza, Clover Labs21:55
  • Use the cited sources to decide where to publish.
  • The mix shifts. Over three weeks, Gemini, Perplexity and Google AI overview cited communities such as Reddit, Quora and Pinterest more Matt Espinoza, Clover Labs32:00.

2.2 Get into community sources

Put your content inside the Reddit threads and comments that AI engines already trust.

• Find the Reddit URLs cited across many prompts and engines. Source: Matt Espinoza, Clover Labs32:30
  • Look for threads that show up across ChatGPT, Gemini and Google AI overview.
• Seed the thread with a better post or a comment. Source: Matt Espinoza, Clover Labs32:30
  • Post seeding means publishing a better version of the cited post.
  • Comment seeding means adding a comment with your own ranked list.
  • Increasingly the cited item is a single comment. A comment listing a personal top 20 became the source of truth.
• Cover both community posts and blogs. Source: Matt Espinoza, Clover Labs32:00
  • Reddit citations dipped a little in ChatGPT but remain strong in blogs.

2.3 Get onto listicles

AI engines like "top 10" articles, so be on them.

• Publish your own listicles and get included on others. Source: Edward Tian, GPTZero16:50
  • The GPTZero team found that GEO engines draw heavily on "top 10" articles.
  • They built more of them and got included on more of them.
  • This came on top of building the best product and reputation, not instead of it.

2.4 Build free tools and shareable output

Tools and product artifacts earn search and AI discovery without ad spend.

• Build free calculators that lead into your core flow. Source: Ethan Ethier, Built With Science36:45
  • They built calorie and macro tools that lead into the quiz.
  • They also hired an external consultant and optimized existing content.
  • Organic SEO was about 30 to 40% of their traffic and started declining as people used AI search.
• Let what users make with your product spread and get indexed. Source: Phil Carter, Elemental Growth32:30
  • Shareable artifacts create social loops plus search and LLM discovery.
  • Gamma's presentations and websites get shared, indexed and referenced in ChatGPT and Claude answers. Gamma grew with little paid acquisition and raised a Series B at a reported $2B+ valuation.

2.5 Back it with authority, quality and social

Citations last when the product is good and people can find you from every direction.

• Use press to earn domain authority, then target easy keywords. Source: Dmitri, YourMove AI18:50
  • Coverage from the Wall Street Journal, Washington Post and Fox gave strong backlinks without paying for or hunting them.
  • He then picked low-competition, high-volume keywords and wrote blog posts with ads for the product.
  • The SEO blog drove nearly half of all traffic. A Fox Business appearance got fewer views than a single decent Reddit post.
• Drive retention, because quality decides recommendations. Source: Tibo Sottiaux, OpenAI13:53
  • ChatGPT looks at retention numbers and plugin quality, not discovery wording.
  • A good plugin can be recommended to a significant slice of users. A weak one stops being recommended.
• Run AI search and organic social together. Source: Matt Espinoza, Clover Labs38:19
  • Viral videos send people to the app store, and a Reddit post, citation or blog still surfaces when they search your app name.
  • Show what your app does inside those posts.
  • For consumer apps, go hard on both at once if capital and resources allow.

2.6 Where they disagree

How much to invest in SEO, and how long Reddit stays a top source, depends on the product and the moment.

• SEO is a core channel for some and a side bet for others.
  • For a mobile-first app, web traffic is arriving but is not in the top five channels. Tanuj Chatterjee, Super Unlimited10:00 says they will not over index on SEO, because conversion was weak and AI overviews are eating search space.
  • Content-led products see it differently. YourMove AI got nearly half its traffic from its SEO blog Dmitri, YourMove AI18:50.
• Reddit's pull in AI answers is shifting.
  • Phil Carter, Elemental Growth32:30 says Reddit over-indexed as an LLM source, but less than 6 to 12 months ago.
  • Another view says Reddit dipped a little in ChatGPT while other engines cited communities more Matt Espinoza, Clover Labs32:00. Check your own prompts before betting on one platform.