I'm validating 5 ideas with a free vote before I build any of them — here's the method
The pattern I'm trying to break: pick an idea that feels clever, build for two weekends, realize nobody wants it, abandon. Repeat 6x in 12 months.
This time I'm validating first. I have 5 ideas for dev tools built on the AI-skills marketplaces: Claude Skills, MCP servers, opencode, Cursor, etc. Instead of picking my favorite and building, I put all 5 on one landing page with:
- A vote (which would you actually pay for?)
- A price ladder ($0 / $5 / $10 / $20 / $50)
The 5 ideas:
- Skills-as-Legos engine — recomposes existing AI skills into novel product concepts nobody's combined.
- Marketplace crawler + recommendation API — "compose these 7 existing skills instead of building from scratch."
- 5-minute pretotype generator — idea → deployed landing page + fake door + smoke-test ad in 5 minutes.
- Idea-to-validation pipeline (skill pack) — install, run
discover my-idea, get a go/kill/iterate verdict. - Side project radar — weekly digest of underserved niches scored by competition density × effort-to-MVP.
Methodology (in case it's useful to anyone here):
- Success threshold: ≥10% survey conversion on 500 visitors + a clear top 1-2 ideas with ≥40% of votes.
- Price ladder tests willingness-to-pay depth, not just interest.
- Per-idea vote distribution tells me which to build first — or whether to build at all.
- Kill criteria are pre-registered: <10% conversion → treat as personal-tool niche, stop pursuing "sell to others."
Page + survey: Page + survey: https://validate.danielcardosods.me
I'll post results back here in ~2 weeks — build decision, winning idea, or kill. Happy to discuss the validation approach in the comments.
Trending on Indie Hackers
I took the opposite path this year and it changed how I think about votes. I ship small data tools in a category where building one costs a weekend, so instead of validating first I launched dozens and let paying usage do the voting. The winners were consistently not the ones I would have picked in a poll, and not the ones friends were excited about. The boring niche ones sitting next to an expensive incumbent quietly outperformed the exciting ones. So my suggestion on your method: if any of the five can be built small in a weekend or two, a real launch with a kill date is the cleanest vote there is, because a vote costs the voter nothing and a signup costs them thirty seconds of belief. Your pre registered kill criteria is the part most people skip, so real respect for that. Curious which of the five you personally want to win, and whether you would still build it if it comes fourth.
love this in principle, one thing i'd watch: a vote tells you what sounds appealing, not what people will pay for. "yeah i'd use that" is cheap. the strongest validation i've seen is asking people to do something that costs them a little - join a waitlist with their real email, pre-order, reply to a cold email about it. if they'll spend 30 seconds or $5 that's a real signal, a free click is a weak one. what are you actually asking them to do?
Pre-registering your kill criteria before you see the results is the part that actually makes this work. Most founders set their thresholds after the data comes in, which means they'll always find a way to interpret borderline results as a green light. Your 10% on 500 visitors is tight enough to be honest. One thing to watch: 500 warm-network visitors and 500 cold traffic visitors will give you completely different signals on the same idea. I made that mistake with Genie 007 early on. Sent the validation to people I knew, got inflated numbers, didn't realise until 3 months later when cold conversion told a different story. What's your traffic source plan for hitting those 500?
One thing I'd be careful of is optimizing for the most popular idea instead of the strongest problem.
People can accurately tell you which concept sounds exciting. They're much less reliable at predicting which one they'll consistently pay to solve. If the votes and the willingness-to-pay signals diverge, I'd trust the money over the popularity.
honestly more rigorous than most validation posts, the pre-registered kill criteria especially. one thing i'd still watch: a price-ladder click is stated willingness to pay, and stated WTP basically always reads higher than revealed. the number that's actually predictive is someone hitting a real pay-$X button, even a fake-door that just 500s at checkout tells you more than a vote. also where your 500 visitors come from will skew it a lot, HN and a niche discord will pick completely different winners. curious to see the results in 2 weeks.