I ran 5 fake customer interviews with AI. Found out why nobody was buying."
How solopreneurs use AI to simulate five real customer types, surface the real objections, and fix their offers before wasting more traffic
Why aren’t they buying? Ask AI to simulate 5 customers and find out
You built the thing. You wrote the sales page. You posted about it. And the conversions are... fine. Or worse than fine. Or you just launched something new and you genuinely have no idea if the positioning is landing.
The problem usually isn’t the product. You’re just too close to it to see what a skeptical stranger sees when they hit your offer page for the first time.
I use a 10-prompt system to simulate five different customer types and have them react to my offer, my pricing, my messaging, whatever I’m trying to figure out. They argue with each other. They raise the objections real buyers have but never actually say out loud. And by the end, I have a clear picture of what’s stopping people from buying, before I spend another week guessing and tweaking things that aren’t the problem.
Here’s the full system.
What you’ll get:
Five realistic simulated customers built around your actual audience
A raw first-reaction debate about your offer, pricing or message
The single biggest reason they’re not buying, stated without softening
A stress test on your planned fix before you build the wrong thing
The five questions your sales page still isn’t answering
A gap analysis of what your offer is missing
A rewrite of your positioning aimed at your most skeptical buyer
A simulated Q&A based on your current copy
A “would they buy?” verdict with specific conditions attached
A one-page planning brief you can act on this week
Prompt 1: Build your customer panel
What it does: Creates five distinct simulated customers with different buying mindsets, situations and reasons to hesitate.
When to use it: Before anything else. Every prompt after this one builds on these five people, so take this one seriously.
The prompt:
I run a digital business as a solopreneur. Here’s what I sell: [DESCRIBE YOUR OFFER IN 2-3 SENTENCES].
My target customer is: [DESCRIBE WHO YOU’RE SELLING TO: their situation, their main problem, what they’ve already tried].
Create a panel of 5 simulated customers who would realistically consider buying this. Give each one:
- A name and a one-line situation (specific, not a demographic label)
- What they’re hoping this offer will do for them
- One reason they’d be interested
- One reason they’d hesitate or not buy
- How they make buying decisions in one sentence (e.g., “reads every word of the sales page,” “buys on impulse if the price feels low-risk,” “needs social proof before they’ll trust a new seller”)
Make them feel like real people with real circumstances, not buyer personas from a marketing textbook.
How to use it:
Be specific about your offer. “A course on productivity” is weak. “A 6-week async course for freelance designers who want to stop working nights” gives the model something real to work with.
Describe your actual target customer, not the ideal one. Include what they’ve already tried and why it didn’t fix the problem.
Save these five people somewhere. Every prompt in this system refers back to them by name.
Example input:
Offer: “A self-paced email course teaching solopreneurs how to write a weekly newsletter in under 90 minutes, with templates and a repeatable system.”
Target customer: “Coaches and consultants who know they should be sending a newsletter but keep putting it off because they don’t know what to write and it takes them forever.”
What you’ll get: Five named, specific customers. Something like “Dana, a business coach with 200 Instagram followers who started a Substack six months ago, published twice, and hasn’t touched it since.” Enough texture that the debate that follows feels like real people, not focus group archetypes.
Advanced note: Ask for one customer who bought something similar from a competitor and was disappointed. That person’s hesitation is usually the sharpest signal in the whole system. Their objection isn’t theoretical.
Prompt 2: Run the first-reaction debate
What it does: Puts all five customers in a room and has them react to your offer out loud, in character, with no moderator softening the edges.
When to use it: Right after building the panel. This is the raw first impression pass, before anyone’s had time to be polite about it.
The prompt:
Using the 5 customers you just created, run a debate about this offer: [PASTE YOUR OFFER DESCRIPTION AGAIN].
Here’s my current sales page headline and main pitch: [PASTE YOUR HEADLINE AND 2-3 KEY SELLING POINTS].
Format it as a transcript. Each person reacts once in the first round based on what they just read. Some will be positive. Some skeptical. Let them talk to each other, not just respond to the offer in isolation.
Then run a second round where each person responds to what someone else said. At least two of them should shift position slightly or add a new concern they didn’t mention in round one.
Keep each speaking turn to 3-5 sentences. Make it feel like five people talking, not five polished reviews.
How to use it:
Paste your actual headline. Not what you meant to write or a cleaned-up version. Whatever is live right now.
If the transcript comes back too polite, add: “These people have been burned by online courses before. Let the skepticism show.”
Pay attention to which person raises the objection that makes you feel slightly defensive. That reaction is usually pointing at your real problem.
Example input: Offer: the newsletter course from Prompt 1. Headline: “Write your newsletter in 90 minutes every week, no more blank page, no more skipped issues.” Key points: templates included, works for any niche, async so no live calls required.
What you’ll get: A 10-turn transcript of five customers reacting to your positioning. The second round is where the useful friction shows up, especially when one skeptic’s concern makes a previously interested buyer reconsider. I’ve had that moment feel genuinely uncomfortable to read, which usually means it was accurate.
Advanced note: Run this twice. Second time, add: “One of these customers has a strong opinion that the price is wrong, though they’re not sure if it’s too high or too low.” Watch what that does to the rest of the conversation.
The full system
You just ran your offer through two rounds of real customer pressure, and you can already see where the doubt is starting.
But you still don’t know which single thing is actually killing conversions. And you haven’t pressure-tested your fix yet.
The next 8 prompts handle the full diagnostic:
Isolating the one reason people aren’t buying (and why it’s probably not what you think)
Attacking your planned fix before you waste time building it
Finding the questions your sales page still isn’t answering
Rewriting your positioning for your most skeptical buyer
Getting a “would they buy?” verdict with hard conditions
Building a one-page action plan you can use this week
Prompt 3: Find the real reason they’re not buying
What it does: Pulls the single most important conversion blocker from the debate and states it directly, without softening it into something more comfortable.
When to use it: Right after the debate. This is where you stop cataloguing reactions and identify what’s actually broken.
The prompt: