Pick your tab and score each feature 1–5 on every dimension. The Average tab blends everyone who has scored. Add your own rows with “＋ Add feature”. Edits sync live for the whole team. syncing…
Weights auto-normalise (set any to 0 to drop it — a 0-weight column is hidden from the table below). Reach only discounts the retention + session-time portion of value (population metrics); conversion, SFT, S→P, differentiation & investor pass through at full weight, so concentrated returning-user bets aren't penalised for serving a small segment. Priority = Adj.Value/5 × AbilityToDeliver/5 × 100.
Reach = what share of active users actually hit this feature, and how often — not how good the idea is.
Reach only discounts the breadth-dependent value — D1–D7 retention + session time (these are population metrics: they only move if many users are touched). The conversion, SFT, S→P, differentiation & investor value passes through at full weight, because those are concentrated, high-intent bets where the small segment is the strategy, not a penalty. So:
Adj.Value = conversion/strategic value + (retention&session value × Reach/5)
This is why a "50% discount for returning users" keeps its full conversion value even at Reach 3 — only its small retention contribution gets discounted.
| 5 | Every user, every session. |
| 4 | Most users, regularly. |
| 3 | A meaningful segment / opt-in surface. |
| 2 | Small subset (beginners, one cohort, opt-in widget). |
| 1 | Niche (B2B-only, ambassadors). |
Scoring rubric — what each 1–5 means (so we all anchor the same way)
| Column | Wt | What it measures | 1 | 3 | 5 |
|---|---|---|---|---|---|
| VALUE — does it move the goals? | |||||
| Conv | .25 | Pushes a returning/lapsed user through the hard paywall (OKR 1). | No effect on the paywall decision. | Indirectly warms intent. | Directly drives the convert decision (offers, pricing, the paywall itself). |
| Ret | .20 | D1–D7 retention — brings users back in week 1 (OKR 2). Reach-discounted. | No reason to return. | Mild habit support. | A core daily reason to come back (streaks, daily content, challenge). |
| Sess | .12 | D1–D7 session time — deeper time-in-app / day (OKR 2). Reach-discounted. | No change to time spent. | A bit longer. | Substantially more engaged minutes (chat, drills, audio). |
| Diff | .08 | Differentiation / moat vs Duolingo, Speak, Babbel. | Commodity everyone has. | Nice but copyable. | A genuine, hard-to-copy edge. |
| Inv | .15 | Investor convincing (3mo) — appeal to learning-community investors who value pedagogy. | Consumer/gamification fluff to them. | Neutral. | A strong learning-science story to show. |
| SFT | .10 | Started Free Trial — drives users into the trial (activation). | Irrelevant to starting a trial. | Some lift. | A direct trial-start trigger (onboarding, paywall, offer). |
| S→P | .10 | Trial → Paid — converts trialists into paying subscribers. | No effect on closing. | Builds value perception. | Directly closes the sale (discount, paywall, proven ongoing value). |
| REACH — multiplier (see box above), only discounts Ret + Sess | |||||
| ABILITY TO DELIVER — can we ship it well? | |||||
| Conf | .50 | Confidence it'll work well at quality (not just ship). Replaced "testability" — with AI the gate is quality, not pre-build validation. | Real quality/infra risk (real-time audio, generation, adaptive grading). | Doable with effort. | Proven / low-risk. |
| Eff | .50 | Low build effort — speed/cost to ship. | Months. | A few weeks. | Days. |
Weights reflect the current bet: returning-user conversion (.25) + the trial funnel SFT/S→P (.20 combined) carry monetisation; D1–D7 retention + session (.32) carry engagement; Investor (.15) is up because you're raising now; Differentiation (.08) is strategic ballast. All adjustable on the weights sliders. Per-feature reasoning is in the "Why (Claude)" column and on hover.
Top-right = Quick Wins · Top-left = Big Bets · bubble size = Priority · colour = category.