Every builder with an unlaunched repo asks the same question at 1am: does my product have potential, or am I about to embarrass myself? The bad news is that the signals most builders check — GitHub stars, likes on the demo video, a friend saying "cool idea bro" — are nearly worthless. They measure whether people enjoy watching your product, not whether anyone needs it. The good news is that real signals exist, they're checkable before launch, and most of them cost nothing but honesty. This post lays out the 4 signals that actually predict whether a product is worth launching, the vanity signals that fake them, and a rough rubric you can score yourself on in 20 minutes — before you spend another weekend polishing.
Vanity signals: applause that doesn't convert
Start by clearing the noise. These feel like validation and aren't:
Stars and upvotes. A GitHub star is a bookmark. It says "neat," not "I have this problem." Plenty of 5,000-star repos have 0 paying users, and plenty of quietly profitable tools have 12 stars.
"Cool idea" from people who know you. Friends are scoring your effort, not your product. The only version of this that counts is a stranger asking "can I pay for this?" — unprompted.
Waitlist signups with no friction. An email address costs the visitor nothing. Waitlists predict launch-day traffic, not revenue.
Your own excitement. Necessary, totally non-predictive. Every unlaunched project felt like a winner the night it compiled.
The pattern: vanity signals measure attention; real signals measure need. Attention is easy to manufacture and impossible to invoice.
Signal 1: problem severity — hair on fire, or mild itch?
The strongest predictor is how badly the problem hurts when you're not around. Severe problems leave evidence: people are already paying for a worse solution, duct-taping spreadsheets together, hiring humans to do it manually, or complaining in public in specific, angry detail.
The test: go find 5 people living with the problem (Reddit threads, niche Discords, support forums of adjacent tools) and read what they say when nobody's pitching them. If the language is visceral — "I waste 3 hours a week on this," "I'd pay anything to never do this again" — that's severity. If the language is "would be nice," you've found a vitamin, and vitamins don't survive launch friction. People will tolerate bugs, ugly UIs, and missing features for a painkiller. They won't tolerate a signup form for a vitamin.
Signal 2: willingness to pay — has money ever changed hands here?
The question isn't "would people pay for this?" It's "do people already pay for things like this?" Existing spend is the most honest market research available, and it's free to check.
Look for: competitors with public pricing pages, freelancers or agencies charging for the manual version, line items in your target user's existing budget. If your product replaces something people pay $50/month for, you have a price anchor and a proven wallet. If nobody anywhere pays anything for the category, you're not entering a market — you're proposing one, and that's a 10x harder game that mostly burns solo builders.
This flips the usual fear on its head: competition is validation, not a death sentence. A crowded category means the wallet exists. Your job is to find the angle — a niche the big players ignore, a 10x simpler version, a price point they can't match — not an empty field.
Signal 3: reachable audience — can you actually get to these people?
A product for "everyone" is a product for no one you can find. The signal here is concrete: can you name 3 specific places where your users already gather, and do you have any realistic way to show up there? A subreddit with 80k members, a newsletter that takes sponsorships, a long-tail keyword cluster with real search volume, a community you already post in.
If the audience exists but you can't reach it — they're enterprise buyers behind procurement, or they don't congregate anywhere online — potential on paper doesn't convert to users in practice. This deserves its own post, and it has one: distribution before you build. The short version: a great product you can't distribute is a hobby.
Signal 4: the moat question — why you, why now?
Last and least — but not zero. You don't need a defensible moat to launch a side project; most successful indie products have none. What you need is a credible answer to "why will this one get picked?" A sharper niche, a workflow insight from doing the job yourself, a wedge the incumbents structurally can't copy (their pricing model, their enterprise focus), or just being early to a platform shift.
Be suspicious if your only answer is "mine is AI-powered." So is everyone's, as of roughly 2 years ago.
A rough self-scoring rubric
Score each axis 1–5, honestly:
- Problem severity — 1: nobody's complained anywhere. 3: people grumble but live with it. 5: people publicly describe pain in hours or dollars.
- Willingness to pay — 1: no money changes hands in this category. 3: free alternatives dominate, some paid exceptions. 5: established competitors with public pricing and visible customers.
- Reachable audience — 1: can't name where users gather. 3: you can name the places but have no standing there. 5: you can name 3 channels and could post in 1 this week.
- Edge — 1: a clone with your branding. 3: meaningfully different on 1 axis. 5: a wedge competitors structurally won't follow.
Rough reads, hedged accordingly: 16+ suggests launch it now and stop polishing. 10–15 usually means the product's fine but 1 axis needs repair before launch — most often reachability. Under 10 is a genuinely useful answer too: better to learn it from a 20-minute rubric than from 6 months of operating something nobody wanted. These thresholds are heuristics from pattern-matching, not physics — treat them as a forcing function for honesty, not a verdict.
The hard part isn't the rubric. It's that you'll grade your own project on a curve. Everyone does.
Getting a score you can't grade yourself
That's the real argument for an outside read. LaunchBuddy — a launch studio for unlaunched side projects — scores every submission on these same 4 axes, and the assessment is free whether the answer is yes or no. AI does the legwork; a human reads it and makes the call. If we pick your project, we build it onto our harness, launch it live, and operate the growth — you keep ownership, pay flat or share revenue, and can kill or port out anytime. If you want to see what that assessment looks like from the inside, we've published the anatomy of an honest teardown.
Submitting takes 60 seconds — 1 link, no deck. And if it's a no, you get the why: which axis failed and what would change the answer. That's worth more than another month of wondering. Start at launchbuddy.app.