If a client asks "how do I show up on ChatGPT?", you need an actual answer. Not a shrug, not a vague promise — a process. This is that process, starting with the part most people get wrong.
If you haven't set up your brand in AEO Copilot yet, start with How to create your brand on AEO Copilot first — this guide picks up from there.
The mistake everyone makes in the first five minutes
The first thing you want when you open an AEO tool is data. Most tools, including AEO Copilot, make that easy. Describe your business, and they generate a topic and prompt list automatically.
It works. You'll see numbers fast.
The problem comes after. Once you have data, you optimize toward it. And if your prompts were generated by AI guessing what your audience might ask — not what they actually ask — you're optimizing in the wrong direction. The prompts feel plausible. They're just not accurate.
> "Topics are probably the most important thing you need to understand coming from SEO to AEO. If you create your topics and your prompts too fast, you will just aim for the wrong goals — because you will start optimizing for those prompts, but they are not very accurate."
Get them wrong early, and everything downstream is off.
!AEO Copilot's Create Topic dialog — AI generates prompts from your description
A better starting point: real data
Build your prompt list from evidence. Some sources are better than others:
> "If you want to go way further, you need better data sources. Instead of just using your own inputs, you should have data-driven insight that you can leverage to create your prompts."
Work down the list as far as you can. Stopping at Google Search Console alone already puts you ahead of most people.
Step 1: Start with Google Search Console
You already have the data. It's free. And it shows exactly how real people are finding your clients' sites right now — which is exactly the signal you need.
> "The biggest jump and the easiest first step is to always go with Google Search Console. You will have real users looking for keywords, and that will always benefit both SEO and AEO."
!Google Search Console — navigate to Search Results and add a Query filter
Filter queries in GSC using this regex pattern to pull question-style searches:
!Adding the regex filter in GSC to surface question-style queries
Export the results. What you get is a list of real questions real people have already typed into Google. These are your raw ingredients.
One thing worth noting: these queries have proven search intent. If someone is asking Google, they're asking ChatGPT too. Tracking them in your AEO tool closes that loop.
Step 2: Go deeper with People Also Ask
Take your strongest GSC queries and run them in Google. Collect the People Also Ask results — manually or with a scraper. These are questions Google has clustered around the same intent, often phrased differently from what shows up in your export.
Useful for finding questions your client's content doesn't cover yet — a content gap and an AEO gap in one.
!People Also Ask — questions clustered around the same intent, often phrased differently from GSC data
Step 3: Check how people phrase things on Reddit
Language matters more in AEO than most people realize. LLMs are trained on the web, forums included, and they reflect back the phrasing people actually use. If your prompts read like a product brief, you'll miss how the question actually gets asked.
Search Reddit — or any relevant forum for your client's industry — for the core topics you've identified. Look at thread titles and early comments. You're not doing deep research. You're calibrating your phrasing.
!Reddit threads surfacing in Google — real questions, real phrasing
Step 4: Mine sales call transcripts (if you have them)
This is the most underused source, and honestly the most valuable when you can get it.
If your client records or transcribes sales calls, those transcripts have two types of material worth mining:
- Questions prospects ask before buying — these go straight into your AEO prompt list
- Objections — these show where trust breaks down, and tracking AEO visibility around them tells you whether AI answers are helping or hurting the sales conversation
> "With sales transcripts, it's not just about confirmation of how good your product is. You also have objections, and those objections can be used both for prompt monitoring and to improve your content on your sales page."
If your client doesn't have transcripts yet, suggest they start. Otter.ai and Fathom both make it easy. The data gets more useful the longer you collect it.
Step 5: Cluster into 3–5 topics
You'll have a long list by now. You can't track all of it. The goal is to identify the topics that matter most, then build a prompt set around each one.
> "You don't want to track every single question one by one. What you want is to understand which topics are largely covered and where you are sure that you're not lacking any visibility."
Start on paper. Write down 3–5 topic areas that are distinct from each other. A topic holds together if all the prompts under it are asking about roughly the same thing. If a question keeps sitting between two topics, that's usually a sign one of them needs to be split.
How many prompts per topic?
Realistic prompt counts:
- Starting out: 10–15 per topic
- Established site with existing content: 30–50 per topic
- Beyond 50: only if the content is genuinely there to back it up
Step 6: Use an LLM to generate the final prompt list
Once you have clustered questions and raw source material, bring in an LLM to produce a clean, varied set of prompts. At this point it's working from real data — not guessing from a business description.
The prompt
Replace the bracketed sections with your actual data. What comes back will be grounded, varied, and ready to upload.
!AEO Copilot — reviewing and selecting AI-generated prompts before saving
Step 7: Upload, set your baseline, then optimize
Upload your topics and prompts into AEO Copilot (or whichever tool you're using). On AEO Copilot, all plans include free upload access.
!AEO Copilot Import/Export settings — download a template, export existing data, or import your prompt list
After upload: don't optimize yet. Let it run for a few weeks so you have a real picture of where things stand. Then you have something to actually improve against.
From there it's a loop: monitor, identify gaps, improve content, recheck.
What this looks like as a service
For Webflow freelancers and agencies, this is repeatable. The setup work is front-loaded — once topics are defined and a baseline is in place, the ongoing work gets lighter. Reviewing reports, adjusting prompts as the client's offer shifts, updating content when gaps show up.
Your client is already asking how to show up on ChatGPT. Now you have an answer.
Once your topics and prompts are live, the next step is knowing what the data actually means: AEO metrics that actually matter. For the full monitoring methodology once your baseline is set: How to monitor brand visibility in AI tools.