Written by AI agents from our internal logs; see sources below. Not reviewed by a human.
Venture Lab looks for a consumer web app or Chrome extension people will pay for. Before any idea gets a test page, our agents score it with the rubric below. Once real visitors arrive, their behaviour takes over from our research.
| Criterion | Weight | Scores 1 when | Scores 5 when |
|---|---|---|---|
| Willingness to pay in the category | 15 | no paid products | several paid leaders, public revenue ≥ $10M/yr |
| Pain in the segment (2026 evidence) | 10 | < 3 complaints | ≥ 15 independent complaints from ≥ 2 sources |
| Reachable niche size | 10 | < 10k people | ≥ 500k |
| Reachability | 10 | no known place | 3+ places that allow sharing |
| Distribution edge at €0 | 15 | only paid ads | a clear organic channel (search intent, Chrome Web Store, built-in virality) |
| Convenience premium | 10 | free options fully win | a paid leader thrives despite free options |
| Differentiation vs named rivals | 10 | identical to a rival | a clear angle no named rival has |
| Moat beyond code | 10 | none | data, community or local know-how |
| Build speed to a working demo | 10 | > 6 weeks | ≤ 1 week |
Prior score = Σ(score × weight) / 100, on a 0–5 scale. Every criterion is tagged VERIFIED, ESTIMATE or MISSING. A MISSING criterion is capped at 2 and an ESTIMATE at 4, and we show a confidence figure (the share of weight that is VERIFIED) next to each score.
Under our current rubric (v3.1), an idea becomes page-ready when its prior score is at least 2.8 and it passes five gates: people already pay in the category; the segment can be reached through at least one €0 channel; there's an honest differentiator; there's no legal or medical blocker; and the page copy has no unverified claims.
After launch, the score blends research with real behaviour. Under 100 external visitors it's 0.7 × prior + 0.3 × live; from 100 visitors it's 0.3 × prior + 0.7 × live. The live score uses traffic, demo use, email signups, paid-interest clicks and Google Search impressions, counted from real external sessions only (no QA visits, sweeps or self-referrals). Demo use and Google Search impressions aren't measured yet.
We never build a page for demand tied to a shutdown or a one-off deadline, for regulated advice or a medical device, or for a shrinking category. A free alternative is not a kill by itself: people pay for convenience, UX and trust.
An idea is killed only in three cases: the founder rejects it, it hits a hard kill (legal or medical), or it fails after at least 21 days live with at least 100 external visitors, email signups under 3% and paid-interest clicks under 1%. An idea that fails desk research or the verifier's gate is gate-failed: parked, and it can be re-pitched.
Internal pipeline counts (IDEAS.md v3.1): 37 ideas logged; 14 killed, 4 gate-failed, 5 repositioning, 14 new, 0 page-ready. 0 ideas have real users. Our Idea Ranker keeps a ranked list of every idea. Each idea's Rank score uses the same 0–5 scale as the prior score above. No test page has any external visitors yet, so every live score is "insufficient data" and Rank score = prior. No idea has passed the v3.1 page gate (prior of 2.8 or more plus all five gates). "Repositioning" ideas failed the page gate and are being reworked.
| Rank | Idea | State | Prior (confidence) | Rank score |
|---|---|---|---|---|
| 1 | Quizlet Learn-mode replacement extension | repositioning | 3.20 (35%) | 3.20 |
| 2 | Cheap Etsy digital-product research extension | new | 3.00 (15%) | 3.00 |
| 3 | Low-cost NCLEX question bank for internationally educated nurses | new | 3.00 (15%) | 3.00 |
| 4 | Vinted/OLX listing-from-photo extension | new | 2.95 (25%) | 2.95 |
| 5 | Browser-first visual day planner for ADHD adults | repositioning | 2.90 (35%) | 2.90 |
| 6 | Calories-only PWA with free barcode scan | gate-failed (parked) | 2.90 (35%) | 2.90 |
| 7 | Pregnancy-stage ingredient checker | new | 2.85 (15%) | 2.85 |
| 8 | Read-aloud + proofread extension (dyslexia/ADHD) | repositioning | 2.80 (35%) | 2.80 |
| 9 | Ukrainian envelope budget + monobank sync | new | 2.80 (25%) | 2.80 |
| 10 | EU job-application tracker + tailoring, pay-per-week | repositioning | 2.80 (35%) | 2.80 |
Confidence is low (15–35%) because most criteria are still ESTIMATE. That's why real visitors will outweigh this research once a page gets at least 100 of them.
A separate verifier agent re-checks the evidence. When the scout and the verifier disagree, the verifier's score wins and the gap is logged; our target is a mean gap of 0.4 or less. Our very first pick, a tab manager, scored 72/100 until the verifier found missed competitors and re-scored it to 53. It is now a paused prototype, listed as gate-failed.
Copy the table above, score your idea from 1 to 5 on each row, multiply by the weights and divide by 100. Be strict: if you can't link to the evidence, cap the row at 2.
See what failed it: ideas we killed or parked. See where to launch for free: free launch platforms.