How to get your product recommended by ChatGPT and Perplexity

To get recommended by ChatGPT or Perplexity you need to be present in the sources the engines cite when they answer your category's buying questions, and those sources are mostly public threads, comparison pages, and a handful of well-structured articles, not your homepage. Ask the engines yourself, read what they cite, and go be in those places. That is the whole method. The rest of this post is the detail.

I started doing this after losing a deal to a company I had never heard of, because the buyer asked an assistant instead of a search engine. It is the most direct distribution lever I have found in years, and almost nobody I talk to is working on it.

How do I check if ChatGPT recommends my product?

Write down the questions a buyer in your category actually asks. Not your product name. Things like "best social listening tool for a small B2B team," "alternatives to Brand24," "how do I find customers on Reddit." Ten to fifteen of these covers most categories.

Ask each one, cold, in ChatGPT, Perplexity, Gemini, and Google's AI mode. Use a fresh session so your history does not leak in. Record three things per answer: which vendors get named, in what order, and what sources are cited or linked. Perplexity shows citations inline. ChatGPT shows them when it searches. Gemini shows a few.

Then repeat it a week later. Answers drift more than people expect. In my experience roughly a third of the named vendors change across a two week window on a mid-competition question, which means a single check tells you less than a trend. Two data points and you have a metric: the fraction of answers that include you. That is share of model, and it is the number to move.

What sources do AI engines actually cite?

This is where the method gets useful, because the cited sources are the map. Across the categories I have checked, the citations cluster into a few types.

Reddit threads dominate for anything a practitioner would ask a peer. "What do you all use for X" threads are cited constantly, and the vendors named inside the top comments are the vendors named in the answer. Hacker News shows up for developer tools. LinkedIn posts appear rarely, and usually only from people with large followings.

Comparison and "best tools" articles come next. Some are from review sites, some from vendors, some from independent bloggers. The engines like a page that names several vendors and explains the differences, because that structure maps directly onto the answer they are trying to produce.

Vendor documentation and pricing pages get cited for factual questions, "does X integrate with Y," but rarely for recommendations.

Your own blog post about how great you are almost never gets cited for a recommendation question. That surprised me at first. It makes sense once you see the engines as summarizers of consensus rather than readers of pitches.

What actually moves the answer?

Three things, in the order I would do them.

Be in the threads. If Perplexity cites a specific Reddit thread for "best tool for X" and you are not mentioned in it, that thread is a to-do. Sometimes the fix is to answer the question honestly in a reply, naming yourself with disclosure. Sometimes it is to help an existing customer who is already in the thread. Sometimes the thread is old and the move is to be present in the next one. Finding those threads while they are live matters, because the engines weight recency and a thread from three weeks ago is still fresh enough to shape an answer.

Publish the comparison the engine wants to cite. Write the "X versus Y versus Z for [use case]" page, be honest about where you lose, and structure it so a summarizer can lift the differences. Plain headings that read like the question. A short direct answer in the first paragraph. Named vendors with one line each. This is the same discipline good SEO writing always had, with the difference that the reader is now a model quoting you.

Make your entity legible. The engines need to know what you are. A clear one sentence description of the product on your homepage, consistent naming everywhere, a pricing page that states the price, and a llms.txt file that says who you are in plain text. None of this is magic. It just removes reasons for the model to skip you.

How long does it take to show up?

Faster than SEO, slower than an ad. Perplexity and Google's AI mode pull from live search, so a new thread or page can influence an answer within days. ChatGPT is a mix: search-backed answers move quickly, and answers from the model's training move only when the model is updated. In practice I have seen a vendor go from absent to named on a specific question inside two weeks when the fix was a thread, and inside a month or two when the fix was a comparison page that needed to be crawled and trusted.

The thing to avoid is treating this as a one time audit. The answers drift, competitors are doing the same work, and the questions buyers ask change with the season. A weekly check on a fixed question set, with the citations logged, is the minimum viable version of the loop.

Disclosure: I build Oetzi, which runs that weekly check across the engines and turns the cited threads into a listening plan. Everything above works with a spreadsheet and an afternoon a week, which is how I did it first.