How to Choose Which Startup Idea to Build This Weekend
Stop rotating through a notes app of ideas. Score three candidates in 30 minutes, kill two, and spend this weekend on the one strangers will actually pay for.
The Problem Is Not a Shortage of Ideas
You do not need another brainstorm.
You have a notes app with twelve half-ideas, a ChatGPT dump from last Tuesday, and a bookmark folder called "maybe." Every Friday you reopen it. Every Sunday you close it. Nothing shipped.
Finding ideas is a solved problem. How to find a startup idea covers the hunting grounds. This article is the next step: picking one so you can actually use the weekend.
If you would rather skip the scoring theater, open the startup ideas library. Those pages already carry market stats, competitors, and a scoped build. Pick one with a score you can live with and start Saturday.
Why Gut-Pick Fails
CB Insights' 2026 post-mortem of 431 VC-backed shutdowns found poor product-market fit in 43% of cases and bad timing in 29%. Running out of cash is how they died. Building the wrong thing is usually why the cash ran out.
Gut-pick optimizes for excitement. Markets optimize for pain people already pay to dull. Those filters disagree constantly.
The other failure mode is "keep all options open." That is not optionality. That is a refusal to spend eight hours on something that might be wrong. Wrong and shipped beats clever and imaginary.
The 30-Minute Scorecard
Take three to five ideas. Not twelve. If you have twelve, you have not filtered.
Score each 1–5 on the same five criteria, in one sitting. Write the numbers down. Do not "feel" them later.
| Criterion | Weight | 5 means |
|---|---|---|
| Pain | 25% | People already pay, or burn hours weekly, on a workaround |
| You | 25% | You have lived this, sold into it, or can reach buyers this month |
| Reach | 20% | You can find 20 of these people without a Super Bowl ad |
| Feasibility | 20% | A testable version fits one weekend with tools you already use |
| Money | 10% | You can name who pays, about how much, and how often |
Multiply, total, rank. The specific formula matters less than using the same rubric twice.
Two rules keep you honest:
- Define what a 5 is before you score the first idea.
- Score all ideas in one sitting. Calibration drifts overnight.
A 2026 DEV Community write-up on idea choice weights founder-market fit at 25% for a reason: founders who already know the market are cited as roughly 40% less likely to miss product-market fit. "Big TAM, no edge" is how you spend a year learning someone else's job.
Kill Criteria (Write These First)
Before you look at scores, write three kill lines:
- I cannot name a buyer with a job title in one sentence.
- I cannot point to a substitute they already pay for.
- I cannot ship a demo a stranger can click by Sunday night.
If an idea fails two of three, it is not "someday." It is out. Put it in an archive note. You are not killing creativity. You are protecting the weekend.
The Tie-Breaker: One Cheap Test
If two ideas land within a few points, do not average your feelings.
Run the cheap test on both, 48 hours each:
- Ten conversations with people who have the problem (not friends).
- One landing page with a price on it.
- One ask for money or a calendar hold.
Ten people saying "cool" is weaker than one person paying $29 or booking a 20-minute call. That is the Mom Test in practice.
A 20%+ email signup rate from cold, relevant traffic is a strong demand sniff. Under 5% after a couple hundred visits is a red flag — not a branding problem.
What "Weekend Feasible" Actually Means
Feasibility is not "could a funded team build this in 18 months."
It is: can you, with Cursor or Bolt, ship a one-feature product that delivers the magic moment once.
If the idea needs two-sided liquidity, a marketplace, or a model you cannot pay for, it is a later company. It is not this Saturday.
That is why the startup ideas briefs exist. Each one is already cut to a weekend wedge: problem, competitors with prices, a how-it-works list. You are not choosing among infinite universes. You are choosing among scoped bets.
A Worked Example
Idea A: "AI for small businesses." Score: pain 2, you 2, reach 3, feasibility 2, money 2. Sounds huge. Dies on every axis that matters.
Idea B: "Practice plans for independent piano teachers after each lesson." Pain 5 if you have taught or can DM 20 teachers this week. You 4 if you have that access. Reach 4 (Facebook teacher groups exist). Feasibility 5 (record → recap → email). Money 4 ($20–$49/month is already the studio-software price band).
Idea B wins even if Idea A would look better in a pitch deck. Pitch decks do not ship on Sunday.
CTA, Mid-Article
If scoring three of your own ideas still feels like homework, stop inventing the corpus. Browse startup ideas, pick the highest-pain page you actually understand, and put it on the calendar.
What To Do Friday Night
- Dump the notes app into a list of five, then cut to three.
- Fill the scorecard once. No revising weights after you see the totals.
- Apply kill criteria.
- Put the winner on Saturday's calendar with a one-sentence magic moment.
- Archive the rest. Revisit only if the winner fails a real test, not a mood.
Then build. The 3-screen MVP is enough surface area.
FAQ
What if my highest score is an idea I do not like?
Do not build it. Like is not a scorecard row, but it is a retention mechanism for you. Pick the highest score you will still open on Saturday when you are tired. If nothing qualifies, you do not have three real ideas yet. Go mining.
Should I pick the biggest market?
For a first weekend product, no. A $500M niche you can message beats a $50B category where you have no distribution.
How many ideas should I score?
Three to five. Fewer and you have not looked. More and you are stalling.
TL;DR
Write kill criteria. Score three ideas on pain, you, reach, feasibility, and money. Test the top two only if they tie. Spend the weekend on one.
When you want the scoring already done for you, start at startup ideas and ship the page that already has competitors and a how-it-works list.