How to Get Your SaaS Recommended by ChatGPT: A Founder's GEO Playbook
ChatGPT trusts what other people say about your product more than what you say about it. Here is the founder playbook for earning the mentions that get you recommended.
Bhargab KalitaFounder of Xoru··8 min read
Open ChatGPT and ask for the best tool in your category. Then ask Perplexity the same thing. If your product is missing from both answers, you are invisible to every buyer who now skips Google and asks an assistant instead. Figuring out how to get your SaaS recommended by ChatGPT comes down to one uncomfortable fact: the model trusts what other people say about you far more than anything on your own website.
I ran that test on my own category early on. Tools I knew were weaker showed up. We did not. They had no secret trick. They had a footprint. Threads, comparisons, reviews and answers spread across the places these models read. This is the playbook I wish someone had handed me then.
How does ChatGPT decide which SaaS products to recommend?
ChatGPT recommends products by blending what it learned during training with live web results it pulls when a question needs current information.
Two layers matter for a founder.
Training data. The base model has read a huge slice of the public web. If your brand was discussed in forums, docs and articles before the training cutoff, the model has some memory of you. If not, it has nothing.
Live retrieval. For comparison and "best tool for" questions, ChatGPT search fetches pages from the web and summarizes them with citations. Perplexity does this on nearly every query. This layer moves much faster, because a page published this month can be cited this month.
OpenAI has said its product results are not ads and that you cannot pay to appear in organic answers. So there is no bidding war to win. You earn your way in.
Here is what the retrieval layer tends to lean on:
Signal
What it looks like
Why AI engines use it
Third-party discussion
Reddit threads, Hacker News comments, forum answers
Real people comparing tools in plain language
Comparison content
"X vs Y" and "alternatives to X" pages
Matches how buyers phrase their prompts
Review platforms
G2, Capterra, Product Hunt
Aggregated sentiment from users
Your own site
Product pages, pricing, docs
Ground truth for features and price
Consistent entity data
Same name, one-liner and links everywhere
Lets the model connect scattered mentions to one brand
What is generative engine optimization for SaaS?
Generative engine optimization (GEO) is the work of making your product easy for AI answer engines to find, understand and cite when someone asks a question in your category.
You will also see it called LLM SEO, AI search visibility or answer engine optimization. The labels differ. The goal is the same. Classic SEO tries to rank one page in a list of ten links. GEO tries to get your product named inside a single synthesized answer, often next to two or three competitors.
That difference changes the work. A page on your own domain is one vote. A dozen honest threads where real people mention your product are a dozen votes, and they come from sources the model already treats as credible. We go deeper on the mechanics on our generative engine optimization page.
Step 1: Find the prompts your buyers actually type
Start by writing down the exact questions a buyer would ask an AI assistant right before choosing a tool like yours.
Think in prompts, not keywords. People talk to ChatGPT the way they talk to a smart friend.
"What is the best [category] tool for a solo founder?"
"Alternatives to [competitor] that are cheaper"
"How do I [job your product does] without [painful workaround]?"
"[Competitor A] vs [Competitor B] for small teams"
Run each prompt in ChatGPT, Perplexity, Gemini and Claude. Write down who gets named and, more important, which sources get cited. Those cited URLs are your target list. If the same Reddit thread or comparison article keeps appearing, that is where the model is forming its opinion of your category.
Step 2: Show up in the threads AI engines cite
The fastest way to get cited is to contribute helpful answers inside the discussions those engines already pull from.
Reddit is one of the most cited domains in AI answers, and in 2024 it signed data licensing deals with both Google and OpenAI. Hacker News, niche forums and LinkedIn posts show up too. When a thread asks "what do you use for X?", the replies become raw material for the next AI summary of X.
So answer those threads. Properly. Explain how you would solve the problem by hand. Name the trade-offs. Mention your product only when it fits, and say plainly that you built it. A comment that would still be useful with your product name deleted is the kind that survives moderation and gets upvoted. If you want a full method for this, read how to find customers on Reddit without spam and our sibling guide on Reddit SEO for SaaS.
Xoru drafts replies in your voice. You read, edit and approve every one before it goes live.
Why autopilot replies hurt your AI visibility
Automated posting destroys the exact footprint GEO depends on.
Removed comments are never cited. Banned accounts lose their history. A burst of near-identical promotional replies is easy for moderators and spam filters to spot, and once a subreddit flags your domain, every future mention is at risk. We wrote about this in detail in the real risks of auto-reply tools.
Discovery can be automated. The reply has to be yours. Every comment carrying your name should be one you read and approved.
Step 3: Publish comparison and alternatives content
Write honest comparison pages, because AI engines answer "X vs Y" prompts by summarizing whoever already wrote that comparison.
If you skip this, the model will still answer. It will just quote a competitor's version of the story. A fair "[your product] vs [competitor]" page, plus an "alternatives to [competitor]" page, gives the engine a clean source that describes your strengths accurately.
Keep them fair. Say who the competitor is better for. That honesty is exactly what makes a page quotable, and it matches a pattern AI engines seem to reward: clear statements of who a product is for and who it is not for.
Step 4: Make your own site easy for AI crawlers to read
Your site will not carry you alone, but it needs to be readable, accurate and open to the crawlers that feed AI search.
Check robots.txt. Make sure you are not blocking OAI-SearchBot (ChatGPT search) or PerplexityBot. Blocking GPTBot, which collects training data, is a separate decision.
Render key content as HTML. Pricing, features and your one-line description should not depend on client-side JavaScript to appear.
Answer the basic questions directly. What is it, who is it for, what does it cost, how is it different. Short headings, short answers.
Use one description everywhere. The same product name and one-liner on your site, social bios, directories and launch pages help models treat every mention as the same entity.
How long does it take to get recommended by ChatGPT?
Expect weeks to months, with live-search engines like Perplexity responding first and training-based recall lagging behind.
A well-placed thread can start appearing in Perplexity citations soon after it gains traction, because Perplexity searches the web on most queries. ChatGPT search behaves similarly when it decides to browse. Recall from training data is slower, since it only updates when a new model ships. That is why consistency beats bursts. A few genuine contributions every week compound. One big launch day fades.
How do you track AI search visibility?
Track it with a fixed prompt list you re-run every month, plus one question in your signup flow.
Keep a simple sheet. Columns for the prompt, the engine, whether you were named, who else was named and which sources were cited. Re-run the same prompts on the same day each month so the trend means something.
Then close the loop on revenue. Add "ChatGPT or another AI assistant" as an option in your "How did you hear about us?" field. Self-reported attribution is imperfect, but it is the only way to see AI-referred signups that never show up as a referrer. Pair it with comment-level tracking so you know which threads actually drive signups. We cover that in stop guessing which posts bring users.
Map each thread you answered to the engagement and signups it produced.
Where Xoru fits
Xoru does the slow parts of this playbook so you can spend fifteen minutes a day on the part that matters.
It watches Reddit, X, Hacker News and LinkedIn for people asking for tools like yours. It drafts replies in your voice. Nothing posts until you approve it, which keeps your accounts safe and your footprint clean. Over time those approved, genuinely helpful answers become the discussion signals that ChatGPT and Perplexity read when they decide who to recommend.
If you want to see where your category's conversations are happening right now, start at xoru.app or compare plans on the pricing page. Either way, run the prompt test today. Knowing where you stand is the first step.
Karma is not a vanity score. It is a gate. Some subreddits will not let you post without it. Here is how much you actually need and how to earn it without farming.
