If you have shipped something good but nobody can find it, this is for you. The way people discover products has quietly changed. Instead of scrolling ten blue links on Google, they ask ChatGPT, Perplexity, or Google AI Overviews "what is the best tool for X" and they trust whatever comes back.
The question is no longer how to rank on page one. It is how to get your product recommended by the AI itself.
I made a full video walking through the exact process I use for my own app, Juicy. If you prefer to watch, start here:
The rest of this post is the written playbook. Call it AI SEO, AEO, GEO, or whatever the newest acronym is by the time you read this. The mechanics are what matter.
Why your old SEO playbook stopped working
Classic SEO was about pleasing a ranking algorithm so a human would click your link. That funnel is shrinking. People are not clicking through anymore. They read the AI summary and act on it. So your real goal is to become part of that summary.
There is a catch most founders miss. If all of your competitors are spinning up the same AI generated blog posts, you are doing the exact same thing as everyone else, and none of you stand out. The move that actually gets you cited is not the obvious one. It is showing up where the AI looks for its source material, which is usually not your own freshly launched website.
The two ways AI decides what to recommend
There are only two ways your product ends up in an AI answer, and understanding the difference tells you exactly where to spend your time.
Training data. The model already knows about you from what it learned during training. This is great if you have it, but you cannot really force it. Models are only trained every so often, and a brand new tool has almost no chance of being baked in.
Retrieval. This is the one you can actually influence. When someone asks a question that needs current information, like a product comparison, pricing, or "best app for X", the model runs a live web search, scrapes the top results, and summarizes them for the user. That stage, where the model chooses which pages to pull from, is your opening. You either show up there with your own content, or you get mentioned inside content that is already being cited.
For a new product, retrieval is the entire game. And retrieval has a clear favorite source.
Reddit is the source AI trusts most
Run a quick experiment. Type a real buying question into Google, switch to AI Mode, then ask the same thing in ChatGPT and open the sources. You will notice the same pattern almost every time. Reddit. Another Reddit thread. Another one. More than 69% of recommendations AI makes it based on a Reddit thread. Let that sink in.
Here is a concrete example from my own product. Juicy is a Mac battery app. It replaces the boring default battery notifications with nicer ones, tracks battery health, and shows which apps drain your battery fastest. One question people genuinely ask is "how do I get notified when my MacBook hits 80 percent." When I search that across Google, AI Mode, and ChatGPT, the answers lean heavily on Reddit threads as their sources.
So the strategy writes itself. To get your product recommended by ChatGPT and AI search, be present in the Reddit threads that those engines are already citing. Not with ads. By genuinely showing up in relevant discussions and mentioning your product in a natural way, so the models pick up on it and surface you when someone asks for a recommendation.
The workflow: get into the threads AI already cites
You could do this manually by searching Reddit for hours. Or you can let a tool surface the exact threads worth your time. This is what I built Redreach for, so I will use it to show the process.
Onboarding takes about two minutes. You give Redreach your website, it analyzes the page and writes a company description for you (editable if you want), then you add three competitors to start. A nice detail especially for Mac/Mobile apps: when a competitor is a studio with several products, Redreach asks whether you want the full domain or just the subpath, so you only track the product that actually competes with you.
Once you finish, it scans hundreds of thousands of Reddit threads, the same kind of threads that ChatGPT, Perplexity, and Google AI Overviews build their citations on. Then you work the dashboard in three passes.
1. The Search Box: high intent threads
These are the highest intent Reddit threads Redreach found, the ones search engines and LLMs are already leaning on. Some look old, and that is the point. One thread for Juicy is from 2023, but it ranks number one for "affordable Mac battery app" and number three for "Mac battery monitoring software." Those are exactly the queries I need to win. If someone types one of them into ChatGPT, that thread is the source. Being mentioned inside it is close to a no brainer. There is a built in comment generator to get you a solid first draft, which you then tweak so it sounds like you.
2. Mentions: ride your competitors
This view shows where competitors are already being talked about, with sentiment and an AI summary of the comments. When a competitor shows up in threads where people are not fully happy, that is your moment to step in and position your product as the better alternative, honestly and helpfully. The goal is simple: wherever a competitor is mentioned, your product should be mentioned too, ideally in a better light.
3. Opportunities: be first on fresh threads
New relevant threads surface every day, so you can be one of the first useful comments rather than the hundredth. A thread like "what is the one Mac app you kept past the honeymoon phase" is a perfect, low pressure place to introduce your product to people who actually want it.
There is also an Insights section that ranks the subreddits most relevant to you, and settings to ignore irrelevant ones, tweak keywords, set custom prompts for the AI replies, and get notified by email, Telegram, Slack, or webhook.
The one rule that keeps this working
Do not spam. This is the difference between getting cited and getting banned. Do not paste your brand into every thread you can find. Focus on genuinely high quality, helpful contributions on threads that are actually being used as source material. Reverse engineer what is already cited, then earn your place in it. Thoughtful beats everywhere, every single time.
Frequently asked questions
How do I get my product recommended by ChatGPT? Get mentioned in the content ChatGPT cites when it answers buying questions. For most products that means Reddit threads, since they are the most frequently cited source in AI answers. Contribute helpfully to the relevant threads, mention your product naturally, and the models pick you up over time.
What is AEO or GEO? Answer Engine Optimization and Generative Engine Optimization are the newer names for optimizing so AI systems recommend you, rather than optimizing only to rank a link on Google. In practice it means becoming the source material that LLMs summarize.
Does this work for a brand new product with no traffic? Yes, and arguably better than traditional SEO. Your new site has no domain authority, but Reddit does. By contributing to threads that already rank, you borrow that authority instead of waiting months to build your own.
How long until AI starts recommending my product? It is gradual. As your mentions accumulate across the threads AI relies on, the models start associating your product with the relevant queries and surface it when people ask for recommendations.
Start showing up in AI answers
The takeaway is simple. People ask AI for product recommendations, AI pulls those recommendations largely from Reddit, so being present in the right Reddit threads is one of the most reliable ways to get recommended.
If you want the threads found for you, with comment drafts ready to go, try Redreach and start getting your product into the conversations that AI search is built on.
And if you want to see the whole thing in action, watch the full walkthrough here.
