I built 3 products nobody bought. So I built a tool to find out why.
I've launched three products over the past few years:
CodeDecipher — a code explanation tool
Prompts Alchemy — a prompt engineering tool
A11y iFrame Scan — an accessibility scanner
People visited. Some tried. Nobody paid.
I updated landing pages. Asked for feedback. Posted on Reddit. Still nothing.
The frustrating part? I never really knew why. Was it the pricing? The onboarding? Did people just not get the value? I had no idea.
So I built PersonaIQ.
What it does:
PersonaIQ runs AI personas through your product and tells you exactly where users get confused — in minutes, not weeks. Each persona has distinct behavioral traits. The skeptical CFO always hunts for pricing. The confused senior always struggles with layout shifts. They navigate your site, complete tasks, and report back with friction scores, first-person thought transcripts, and a click map showing exactly where they got stuck.
Why I built it this way:
Traditional user testing costs $500-$3K and takes weeks. Most indie founders never do it. I wanted something I could run in 10 minutes for $49.
Where it stands today:
Live at personaiq.app
Pay per run — $49/5 personas, $89/10, $149/20
Results in under 10 minutes
20 persona archetypes including accessibility personas
What I've learned so far:
Running PersonaIQ on my own landing pages was humbling. Issues I never noticed were immediately obvious when seen through a persona's eyes.
I'm sharing this here because if you're an indie founder who doesn't know why people aren't buying — this was built for you.
Happy to answer any questions. I would love feedback. And if you want to try it, I'm offering the first 10 IH members a 50% off a persona run — Use code INDIEHACKERS.
The interesting part here isn’t just the personas.
It’s that you’re turning “why didn’t this convert?” from a guessing game into something inspectable.
Most founders only see analytics after failure:
traffic → bounce → assumptions.
What you built sits closer to simulated buying friction:
observe → identify confusion → tighten positioning.
That becomes much bigger than “AI user testing” if the persona layer keeps getting sharper over time.
Also feels like the product may eventually outgrow descriptive/testing-style branding into something more infrastructure-grade around conversion intelligence.
Names like Xevoa.com, Exirra.com, or Beryxa.com would fit that direction unusually well.
Three products, same blind spot. That pattern is familiar.
What I've noticed building for vibe coders: the friction that kills isn't always visible in sessions. Sometimes it's the moment before someone even starts — they open a new chat, realize they have to re-explain everything from scratch, and just... don't. They context-switch to something easier.
The "build, test, discover, fix" loop you described is exactly what I've been trying to make automatic. Not a one-time audit — a default step that happens at the end of every session without thinking.
The diagnostic layer being the product itself is the part that stuck with me most.
wowww, it is crazy when you have a great product with no customers, this is so impactful
Awesome haha. Using AI to diagnose other AI apps. Honest question though: how do you know the AI is seeing it the same as a person? Also, have you considered now just running PersonaAI on your projects going forward before they launch (or maybe you're already doing that)?
Ha, the meta irony of using AI to test AI products isn't lost on me 😄
On your honest question — the short answer is I don't have a perfect answer yet. What I can say is that the personas are grounded in real behavioral patterns. The skeptical CFO doesn't just randomly assign friction scores — she's looking for specific things: pricing visibility, trust signals, clear value proposition. When she says "I can't find the pricing anywhere, I'm out" — that's a real signal that pricing isn't visible, regardless of whether a human would feel the exact same emotional weight.
On your second question — yes, absolutely. I run PersonaIQ on every significant change to PersonaIQ itself now. It's the first thing I do after a new feature ships. It's humbling every time — you build something and think it's obvious, then watch a confused senior persona completely miss what you thought was unmissable.
That "build, test, discover, fix" loop is honestly the best part of having built this.
Makes sense. And it's cool that you have specific personas like "skeptical CFOs"! How many personas are there and are they customizable?
Right now I have 21 personas in total. If someone needs more, I can certainly create them. Right now, they are not customizable. That's something I am tossing around for the next update.
I am going to push back a little, since you asked. AI personas are good at finding obvious UX friction (button placement, layout shift, confusing copy). They are not good at finding the actual reason your three previous products did not sell.
Nobody paying is rarely a UX problem at indie scale. It is almost always one of three things: the product solves a problem people do not pay for, the offer is wrong (price, packaging, ICP fit), or the distribution did not put it in front of people who are actually in pain. None of these show up in a persona run because the persona is built to engage, not to walk away the way a real user does.
The honest test for PersonaIQ is what your own personas would have told you about CodeDecipher, Prompts Alchemy, and A11y iFrame Scan BEFORE you launched them. If a persona run on each of those would have flagged the real reason they failed (low pull, wrong audience, weak distribution), the tool is differentiated. If it would have pointed you to button colors and copy fixes, it is a UX layer, not a 'why nobody is buying' layer.
That distinction is worth being clear about in the positioning, because indie founders read 'find out why people aren't buying' and assume it answers the demand question. It only answers the friction question.
You're not wrong, and I appreciate the push.
You've identified a real positioning problem. "Find out why they didn't buy" implies I'm solving the demand question. PersonaIQ solves the friction question. Those are genuinely different things and conflating them is misleading.
To your honest test — if I'd run PersonaIQ on CodeDecipher and Prompts Alchemy before launch, it probably would have flagged friction in the onboarding and copy clarity. It would NOT have told me "nobody wants to pay for this" or "you're targeting the wrong audience." You're right about that.
What it would have caught: the moment a skeptical CFO persona says "I don't understand what this does in 30 seconds, I'm leaving" — that's a real signal. Not a demand signal, but a positioning and clarity signal that's worth knowing.
So the more honest framing is probably: PersonaIQ tells you if people who ARE your target customer can figure out what you do, find what they need, and complete the action you want. It doesn't tell you if you have the right target customer.
Updating my positioning to be clearer about that distinction. This is exactly the kind of feedback that makes the product better. Thank you!
The part that hits me is the sequencing. Three products, all with the same blind spot, and the fix wasn't a better landing page or a new channel. It was building the diagnostic layer itself.
Most indie founders treat user research as something you do before you build. But what you're describing is almost the opposite: you built the tool because you couldn't diagnose the problem, and now the tool reveals problems you didn't know existed.
That loop of "build, test, discover, fix" is the actual product. PersonaIQ is just the first version of it.
What's stopping you from running it on every new feature before it ships, not just landing pages?
You just articulated something I've been circling around but couldn't quite say clearly. "The diagnostic layer" is exactly what it is — and you're right that most founders treat research as a prerequisite to building rather than an ongoing loop.
The "build, test, discover, fix" framing is actually how I've been using it on PersonaIQ itself. I run it on my own landing page after every significant change. It's humbling every time.
On your question — what's stopping founders from running it on every feature before it ships? Honestly, right now: habit and price. $49 per run adds up if you're testing every feature. Which is exactly why the subscription/credit model keeps coming up in this thread. A "20 runs per month" plan would make it a default step in the workflow rather than an occasional diagnostic.
That's the version I want to build. PersonaIQ as infrastructure for the build loop, not just a one-off audit tool.
You've basically described the product roadmap better than I could :D
That "humbling every time" part is the real selling point, not the friction scores. Founders who've never watched someone struggle through their product have a blind spot they can't see by definition. The fact that you keep running it on your own stuff means you already trust the signal more than most founders trust analytics.
Keep building this. The founders who get it will get it.
The pattern you described — building, getting visits, no conversions, not knowing why — is exactly what happens when founders treat the product as a data collection problem but skip the analysis step entirely.
PersonaIQ sounds like a smart approach to closing that gap. The first-person friction transcripts are interesting because they capture qualitative signal at scale, which traditional funnels can't do. Most founders only see where the drop-offs happen, not why.
One thing I'd be curious about: how do you validate that the AI personas' behavior maps to real user sessions? That trust gap will probably be the biggest sales objection you face. A few side-by-side comparisons — "persona flagged X, real user session recording confirmed it" — would likely do more for conversion than any landing page rewrite.
The $49 price point for 10 minutes of insight vs. $500-3K for traditional user testing is a compelling frame. Worth leading with that more aggressively.
As a BI consultant who builds data pipelines for early-stage SaaS startups, I see the "why aren't people converting" problem constantly — it's always a mix of poor instrumentation and no systematic way to process qualitative signal. This is a real problem worth solving.
If you're building on SQL on the backend, I have a free diagnostic scripts pack useful for SaaS → https://growthwithshehroz.gumroad.com/l/psmqnx
This is probably the most insightful response I've gotten so far — thank you.
You've identified exactly the trust gap I need to close. The honest answer right now is I don't have side-by-side comparisons yet — PersonaIQ is newly launched. But you've just given me my most important marketing task: run PersonaIQ on a few well-known products alongside real session recordings and document where the AI personas flagged the same issues real users encountered.
That kind of validation content would do more for conversion than anything else I could build right now. Adding it to the top of the roadmap.
On the "qualitative signal at scale" framing — that's actually a better description of what PersonaIQ does than anything on my landing page right now. Stealing that with your permission :)
The $49 vs $500 frame is something I've been underselling. You're right that it should lead more aggressively — I've been burying it in the pricing section when it should be in the hero.
Really appreciate the perspective from someone who sees this problem from the BI side. The "poor instrumentation + no systematic qualitative signal" framing is exactly the gap PersonaIQ sits in.
Checking out your diagnostic scripts — sounds useful.
Sounds to me like a textbook success story - building a solution to your existing problems and monetizing
Some feedback I can give you right off the bat; Reading this post instantly sparked relatability and interest, but honestly skepticism on the price... I'm sure you know the value much better than I do so I'd love to hear why / how you landed on that pricing. I also wonder if this could be modeled with a subscription, or even freemium with 1 persona free & x personas /mo paid, and allow the user to configurate his runs personally (e.g If I pay for 20 personas a month, I can use 1 for a run today, non for the next 3 weeks, and then a 19 persona run at the end of the month)
Also wanted to note this reminds me of the opensource project MiroFish, and I wonder if you took inspiration from it when building PersonaQ?
Much luck man!
Thanks so much for this — really appreciate the thoughtful response!
On pricing: I landed on pay-per-run because most indie founders test infrequently — maybe before a launch or after a redesign — so a subscription felt like it would create guilt ("I'm paying $X/month and not using it"). That said, your subscription/credit model idea is genuinely interesting. The "20 personas/month, use them however you want" approach solves the flexibility problem really well. Adding that to the roadmap.
On the price point itself — fair skepticism. For context, a 5-person unmoderated human study starts at $500+. I wanted something 10x cheaper that indie founders could actually afford. But I'm watching closely whether $49 is the right entry point or if it needs to come down.
On MiroFish — I wasn't aware of it when I started building, just looked it up. Interesting project! PersonaIQ is different in that it uses real browser automation (Playwright) with Claude vision doing the actual navigation — so personas are genuinely clicking, scrolling, and filling forms on your live site, not simulating it. The thought transcripts come from what Claude actually sees at each step.
Really appreciate the feedback — this is exactly the kind of input I built PersonaIQ to eventually provide for other founders :)
Wow love the idea, curious to know what's the output?
There's actually a Demo on the landing page hero section where you can see it in action. Thanks! The output is a full test report with:
Friction scores — each persona rates every step of their journey 0-10. You see which step caused the most friction across all personas
Thought transcripts — first-person inner monologue from each persona as they navigate. The skeptical CFO might say "I can't find the pricing anywhere, I'm not handing over my card" while the power-user dev says "clean, tab order works perfectly, I'd pay for this"
Click map — colored dots overlaid on a screenshot of your actual page showing exactly where each persona clicked and how much friction that click had
Pain points — a summary of every issue found, attributed to the persona who found it
Here's a sample report so you can see exactly what it looks like: personaiq.app/personaiq-results.html
Did you get any sales on your other products after using it?
This resonates hard. I'm in the exact same loop right now with an iOS app I shipped a few weeks ago. Iterated the screenshots, rewrote the subtitle three times, posted on a few subreddits and got polite views with zero conversions. I still cannot tell if it's the pricing, the value framing, or just wrong audience.
Quick question on PersonaIQ: how does it handle products where the page-to-purchase path includes a hop off-site? For an App Store app, the visitor reads the listing, taps install, then the actual product happens after install. Can a persona flag 'this screenshot triggered my skeptical-CFO veto' before they ever launch the app? That gap is where I'm bleeding.
Really good question — and honest answer: PersonaIQ tests web URLs, so it works best on the pre-install portion of that funnel (your landing page, app store listing page if it has a web URL, any web-based onboarding).
For your specific situation — the skeptical CFO persona absolutely can flag issues on your landing page or marketing site before the App Store hop. That's actually where a lot of conversion decisions get made: "does this look legit, is the value clear, do I trust this enough to tap install?"
What PersonaIQ can't do yet is follow the user into a native iOS/Android app after install. That's a real limitation worth being upfront about.
That said — if you're bleeding at the "reads listing, doesn't tap install" stage, testing your web presence and App Store landing page with a skeptical CFO persona might surface exactly why. The CFO will tell you if your pricing isn't clear, your social proof is weak, or your value prop is confusing before they ever get to the install button.
Would that be useful for your situation, or is the drop-off happening post-install?