Educational
How To Appear In AI Search
A practical 7-day process for finding the prompts your customers ask, tracking where ChatGPT gets its recommendations, and earning your way into the sources AI already trusts.
Published on
May 24, 2026
Written by
Maciej Czypek
Founder
Appearing in AI Search is not only about publishing more content. It is about understanding which questions your customers ask, which businesses AI recommends, and which sources AI trusts when forming those recommendations.
The good news: you can research this manually for free. It takes patience, a spreadsheet, and a willingness to look beyond your own website. The process below is the practical version of AI visibility work: prompt research, controlled tests, source capture, daily aggregation, and source-level action.
The goal is not to trick AI
The goal is to become easier for AI systems to understand, verify, and recommend. If ChatGPT is repeatedly using the same directory, Reddit discussion, local guide, review site, or niche publisher to answer questions in your market, that source is part of your AI Search environment.
Your job is to identify those sources and decide where you can credibly appear. Sometimes that means adding helpful context to a public discussion. Sometimes it means improving a directory profile. Sometimes it means earning a mention in a relevant article. The work is less about gaming rankings and more about being present where AI already looks.
The 7-day AI Search visibility process
Step 01
Start with prompt research
Write down the real questions your customers would ask ChatGPT before choosing a provider. Keep the language casual. Customers rarely type polished SEO keywords into AI Search. They ask messy, human questions with constraints, doubts, preferences, and context.
Aim for four or five prompt ideas. For a local dentist, that might include questions about nervous patients, transparent pricing, weekend availability, emergency care, or best-rated clinics near a specific neighborhood.
Step 02
Test each prompt in a clean chat
Run each prompt one by one in ChatGPT. Use a separate chat for every prompt, and avoid using an account where ChatGPT already knows that you own the business. The goal is to simulate a normal customer, not your personalized assistant.
If you can, delete or clear the previous chat before testing the next prompt. This reduces cross-prompt context and makes the daily result easier to compare.
Step 03
Capture every source ChatGPT uses
After each response, open the sources view and capture the pages ChatGPT used. Do not only record the final citations attached to the answer. Also capture pages it browsed, considered, or used during the search process.
This can include Reddit threads, local directories, review pages, blog articles, news stories, listicles, publisher pages, forum discussions, and competitor websites. Copy the URLs into a sheet, or take screenshots if that is faster.
Step 04
Aggregate the daily results
At the end of the day, turn the raw observations into a small dataset. Count how often your business appeared, what position it held when it appeared, and which sources showed up repeatedly across prompts.
You are looking for patterns, not one-off screenshots. A source that appears across several prompts is more important than a source that appeared once in a strange edge case.
Step 05
Repeat for at least seven days
AI answers can shift from day to day. Run the same basic process for a week before drawing conclusions. This helps separate durable market signals from random variation.
After seven days, you should know which prompts matter, whether your business is visible, which competitors appear most often, and which third-party sources AI seems to trust in your market.
What to track each day
Keep the dataset simple. You do not need an enterprise dashboard to learn something useful. For each prompt, record the prompt, the businesses recommended, your position if you appeared, the cited sources, and the extra pages ChatGPT appeared to browse or consider.
Daily metrics
- How many prompts included your business in the final answer
- Your average position when your business was recommended
- Which competitors appeared most often
- Which sources were cited or browsed most often
- Which sources were realistic places for you to earn a mention
The most useful metric is usually source frequency. If one page, platform, or publisher keeps appearing across multiple prompts, that source may be shaping AI recommendations in your category.
How to turn sources into action
After seven days, sort your source list by frequency. Then ask two questions about each source: how much does this source seem to influence recommendations, and how realistic is it for us to appear there?
Some actions are simple. If ChatGPT often reads a Reddit discussion or forum thread, you may be able to add a genuinely useful comment that answers the question and mentions your business only when relevant. If it uses a local directory, improve your profile, categories, photos, reviews, and description there.
Other actions take more work. If a respected industry blog, local publisher, or comparison article appears repeatedly, you may need to pitch an educational article, contribute expert commentary, sponsor a useful resource, or build something worth citing. The point is not to force your brand into every source. The point is to earn context in the few sources that matter most.
What good conclusions look like
A good seven-day conclusion is specific. Not: we need better AI SEO. Instead: ChatGPT recommends us in one out of five customer prompts, our average position is third when we appear, and three sources seem to influence most answers in our market.
From there, you can create a source action plan. Choose the realistic high-impact sources first. Add useful context where a discussion already exists. Improve weak profiles. Pitch educational content where a publisher is repeatedly shaping the answer. Then repeat the research later and check whether your visibility changes.
This is how AI Search optimization becomes operational. You stop guessing what AI might care about and start working from the sources it actually uses.
A quieter shortcut
This is also the idea behind AI Places
You can do the process above manually, and it is worth doing at least once because it teaches you how AI Search behaves in your market. AI Places in aeoh is built around the same idea: find the places ChatGPT already uses for recommendation prompts, then turn those sources into realistic actions you can take.
Manual research gives you the intuition. Source-level action is what turns that intuition into visibility.
Create your AI Visibility Audit