You shipped.
Nobody came.
Find out why.

I look at your funnel and at real public conversations, and tell you what's most likely blocking growth (the problem, the audience, the message, distribution, or willingness to pay) and the one thing to test next.

Diagnose my launch See an example →

Early access: I'm doing the first diagnoses by hand. You need a URL and a few numbers you probably know. "I don't know" is a fine answer.

Built from real public evidence. The first investigation, on my own idea, checked 149 quotes word for word against their sources.

You shipped.

+built features for weeks

+posted on Product Hunt

+shared it on Reddit

+rewrote the landing page

+added another feature

+tried ads

+still no paying users

= you still don't know what the real bottleneck is.

There's a better way to find out.

How it works

Your product + your numbers + real market evidence
= your next best experiment.

You share your product and a few numbers

A 5-minute form: visitors, signups, trials, paying customers, price. Skip anything you don't track.

I test five competing explanations

Problem, audience, message, distribution, willingness to pay, with public evidence for and against each. Every quote is checked word for word against its source.

You get one answer and one next test

The most likely bottleneck, how sure I am, what I still don't know, one experiment, and what not to build yet. A 10-minute read, plus a 30-minute call.

See exactly what you get

One answer first. The evidence after.

Most likely reason your launch isn't workingIllustrative example, not a customer's data

Willingness to payModerate evidence

Why

Signup conversion is healthy
Users activate
No one converts to paid
Public market evidence shows the pain is real
Purchase intent in that evidence is weak

Other explanations considered

MessageUnlikely: people sign up, so the page is understood

AudienceLess likely: activated users match your ICP

DistributionLess likely: enough of the right people arrive

ProblemUnlikely: the pain shows up in public evidence

Next experiment

Offer the same solution manually to 10 qualified buyers at $49.

Success: 3 or more pay. Stop: none pay after 10 honest asks.

Do not build yet

Another dashboard feature, until someone pays for what already exists.

Fair question

Why not just ask ChatGPT?

You can, and you should compare.

A chatbot summarises what's out there, and it tends to tell you your idea is great. It can't see your funnel numbers, and it doesn't show you the evidence against its own answer.

What you get here instead

Real posts, quoted word for word and linked, counted per independent source (one loud thread doesn't count ten times), with evidence against the diagnosis shown next to evidence for it.

"Beware of AI reviewing AI."

A founder on Hacker News, after three chatbots approved an idea real users didn't need

Who's behind this

Hi, I'm Lior.

I'm a SaaS founder who builds products with AI tools like Claude Code. Building got fast. Figuring out why a product gets no traction, who actually wants it, and what to test next did not.

I started testing this after running into the same problem with my own products. Before offering it to anyone, I ran the diagnosis on this idea itself. It surfaced evidence I didn't particularly want to see, which is exactly the point. For now, every diagnosis is done manually, with the evidence checked before I give you a recommendation.

The first few are free in exchange for honest feedback. After that it's $49, refunded if it doesn't change anything in your plans. Your numbers stay private and nothing is published.