Blog · how to get startup ideas
A 7-Day Startup Idea Validation Sprint (With a Ranked Shortlist)
A practical 7-day sprint: constrain your filters, run Match Pro, save a shortlist, talk to users, and decide build / park / discard—without drowning in the catalog.
Published 2026-09-24 · startup idea validation sprint
Introduction
A 7-Day Startup Idea Validation Sprint (With a Ranked Shortlist): ignore motivational fog. Under startup idea validation sprint, ask what evidence would make you stop, continue, or narrow.
Original insight: your unfair advantage is usually a boring asset—access, scars, dataset, community—not a clever slogan.
Real-world example: Airbnb’s early unscalable behavior (professional photography, direct host help) taught what the product had to become.
Principles before the calendar
- Constraints beat inspiration. Filters first, browsing second.
- Shortlists beat folders. If you cannot recite your candidates, you do not have a shortlist.
- Fit score is not revenue. Match Pro ranks preference match, not willingness to pay.
- Conversation beats another article. Including this one.
Before Day 1 — write your constraints
Put four lines where you will see them daily:
- Hours/week you can ship (be unkindly realistic)
- Skills depth (no-code, low-code, full-stack, AI wrapper, etc.)
- Capital (none / low / medium)
- Audience you can reach this month (classmates, coworkers, niche community, local operators)
If you cannot name a reachable audience, fix distribution access before you debate idea quality. An elegant problem with no path to five conversations is a hobby brief.
Day 1 — Run Match Pro (no random browsing)
Open Match Pro and set industry preference (or leave open), skills / technical depth (honest over aspirational), stage / audience, and optional value interest and customer market.
Run Find matching ideas. Read % fit and why-fit chips. Open at most 15 idea pages today.
Anti-goal: do not "research the whole database." Your job is a candidate pool.
Day 2 — Save ruthlessly (max 10)
Click Save only when you can finish this sentence: I could talk to a real [target user] about this problem within seven days.
Open Saved. If you have more than 10, remove until 10 remain. Use Copy shortlist and paste into a note you will actually reopen.
Original insight: a shortlist you can recite beats a bookmark graveyard you fear opening.
Day 3 — Kill criteria (desk pass)
For each saved idea, open the page and read problem / target user / proposed solution, the builder brief (who it is for, first moves, risks), and capital / technical depth tags.
Score 1–5 on: Can I reach five target users this week? Is the pain weekly (not annual)? Can I demo something in 14 days or less at my real skill level? Do I have any unfair access (community, job, data, channel)?
Discard under 12/20. You want 5 or fewer ideas entering conversations. If everything dies, tighten Match filters and repeat Days 1–2 once—not five times.
Day 4–5 — Talk to five humans
For the top idea, contact five people who resemble the target user. Use a listening script: How do you handle [problem] today? What have you tried? What does a bad week cost you (time, money, stress, reputation)? If something removed the weekly version of this pain, would you try it?
Do not lead with your solution. Steal their language for an offer sentence later.
Students: classmates, lab partners, club chats. Employed builders: adjacent teams, not only founder Twitter. See student and research ideas if that is your lane. Write notes the same day. Patterns matter more than compliments.
Day 6 — Fake door / one offer sentence
Draft one sentence: For [user] who struggle with [pain], I help them [outcome] without [common objection].
Put it on a simple page or form. Share with the same five people plus five more. Measure replies, "yes I'd try," and whether anyone asks price or timeline unprompted.
If nobody cares enough to reply, that is data. Do not "improve the landing color." Change the problem or the audience.
Day 7 — Decision ritual
Choose exactly one: Build only if repeated pain language showed up (define a one-week MVP); Park in Saved with a note; or Discard and remove from Saved—no zombie backlog.
Then stop. Do not immediately open fifty new ideas tonight. If you discarded everything, schedule a new sprint next week with stricter constraints.
Optional accelerator — scorecard
If painkiller vs vitamin language is muddying decisions, run the painkiller vs vitamin scorecard on your top two Saved ideas before Day 4.
Common failure modes
Saving 40 ideas "for later"; treating fit percent as proof of demand; skipping conversations because more research feels safer; building a full stack before a fake door; restarting Match Pro instead of deciding.
How you'd know this sprint guide failed
Match Pro or Saved is unavailable where you are testing; you end the week with the same undifferentiated backlog; you cannot point to five conversations or a written offer sentence; you shipped code with zero external signal.
Start now
Build today's shortlist: Match Pro then Saved ideas. When you open a candidate, use the builder brief as your Day 3 checklist. For how we keep the catalog from going thin, read curation and noindex policy.
Field notes (read these before you build)
Unexpected challenge: friends will praise the idea and still refuse to pay. Design tests that cannot be passed with politeness.
Counter-intuitive advice: read less “thought leadership” this month and rewrite your offer until a stranger can repeat it accurately.
Distribution bottleneck: if your only path is “go viral,” you do not have a path. Pick one repeatable channel and run it for thirty days.
Hidden cost: support and onboarding labor you pretend software will erase in week one. It will not.
One caution: multiplayer roadmaps with co-founders who never talk to users will produce elegant irrelevance.
One recommendation: compare two research-style entries on Research and write a one-page “why not” memo for the weaker one.
Straight take: uniqueness in writing is an SEO tactic and a thinking tactic—if the page could be any other page, the strategy is probably generic too.