Skip to content

Define Prioritization Framework: Brainshelf Q3 Features

Scenario

Brainshelf is a B2C AI book-curation subscription. The PM has 8 candidate features for Q3 and engagement plus engineering estimates, so RICE and ICE both run, with MoSCoW as a scope-bounding view. The lesson is two divergent items (Mobile app, Family plan) that expose RICE’s reach bias. All reach/impact/effort numbers are illustrative PM inputs for this scenario [fictional].

Source Notes:

  • Sean McBride, “RICE: Simple prioritization for product managers” (Intercom, intercom.com) - the RICE formula and the 0.25-3 impact scale used here originate from Intercom’s framework.
  • Sean Ellis (GrowthHackers) - the ICE framework (Impact, Confidence, Ease) originated for prioritizing growth experiments; used here as the lower-input complement to RICE.
  • Dai Clegg, DSDM (MoSCoW) - the Must/Should/Could/Won’t commitment buckets used for the Q3 scope bound.

Prompt

/prioritization-framework
prioritize our Q3 candidate features for brainshelf (AI book-curation
subscription). decision context: Q3 roadmap, ~14 eng-weeks of capacity.
features (i have engagement data + eng estimates):
1. rec algorithm v2 2. social follow-friends 3. mobile app
4. public reading profiles 5. reading streaks 6. audiobook recs
7. export to goodreads 8. family plan
run whatever frameworks apply and show me where they disagree.

Output

Prioritization: Brainshelf Q3 Roadmap Candidates

Applicability Filter Summary

Reach, impact, effort, and confidence are available, so RICE and ICE both run. The decision bounds Q3 scope (~14 eng-weeks), so MoSCoW runs as a commitment view. Weighted Scoring is excluded (no competing stakeholder criteria provided). Kano is excluded: no customer-research data. A Kano survey on these eight features would unlock it.

Inputs Summary

Eight Q3 candidates with PM-supplied estimates (illustrative [fictional]). Reach is users affected per quarter; effort in eng-weeks.

Per-Framework Scoring

RICE

ItemReach/qtrImpact (0.25-3)Conf (%)Effort (wk)RICENotes
Reading streaks35,000160%210,500Cheap retention lever
Rec algorithm v240,000270%69,333Core value prop
Mobile app30,000280%124,000Big reach, big build
Export to Goodreads8,0000.590%13,600Trivial, high confidence
Social follow-friends25,000150%52,500Speculative
Audiobook recs20,000150%42,500Speculative
Family plan6,000270%42,100Narrow reach, high per-user value
Public reading profiles15,0000.560%31,500Low impact

ICE

ItemImpact (1-10)Conf (1-10)Ease (1-10)ICENotes
Export to Goodreads499324Trivial, well-understood
Family plan776294High per-user value
Reading streaks668288Cheap and solid
Rec algorithm v2974252High value, harder
Audiobook recs656180Middling
Mobile app882128Valuable but a 12-week slog
Public reading profiles367126Low value
Social follow-friends555125Speculative

MoSCoW (Q3 scope, ~14 eng-weeks)

ItemBucketRationaleRisk if dropped
Rec algorithm v2MustThe product’s core promiseErodes the differentiator
Reading streaksShouldCheap retention winSlower habit formation
Export to GoodreadsShouldCheap, reduces lock-in anxietyMinor friction
Family planShouldNew revenue lineForgone revenue
Mobile appCould12 eng-weeks blows most of Q3Mobile users wait
Audiobook recsCouldNice expansionNone critical
Social follow-friendsWon’t (this time)Low confidenceDefer until validated
Public reading profilesWon’t (this time)Low impactNone

Per-Framework Ranking Output

Each scoring table above is sorted high to low, so the per-framework ranking is the row order shown (top item first, lowest last). The side-by-side rank positions, and the items where the frameworks disagree, are consolidated in the Cross-Framework Comparison below.

Cross-Framework Comparison

ItemRICE rankICE rankMoSCoWAgreement
Reading streaks13ShouldStrong
Rec algorithm v224MustStrong
Mobile app36CouldDivergent
Export to Goodreads41ShouldClose
Social follow-friends58Won’tMostly agree (RICE 5th, borderline)
Audiobook recs65CouldClose
Family plan72ShouldDivergent
Public profiles87Won’tAgree (defer)

Divergent - Mobile app (RICE 3rd, ICE 6th): RICE rewards the 30,000 reach; ICE punishes Ease (2) because it is a 12-week build. Driver: reach vs. effort. With only 14 eng-weeks, the effort signal should win for Q3 - hence MoSCoW Could.

Divergent - Family plan (RICE 7th, ICE 2nd): RICE’s reach term punishes the narrow 6,000 reach; ICE rewards the high per-user impact and decent ease. Driver: RICE under-weights revenue-concentrated features. Because Family plan opens a new revenue line, weight the ICE/MoSCoW view and keep it as a Should.

Executive Summary with Recommendation

Fund Rec algorithm v2 and Reading streaks: both score high across frameworks and fit the budget. Add the two cheap, high-confidence items - Export to Goodreads (1 week) and Family plan (4 weeks) - the latter despite a low RICE rank, because RICE undervalues its concentrated revenue impact. That fills ~13 of 14 eng-weeks. Defer Mobile app to a quarter with capacity for a 12-week build, and drop Public profiles (all three frameworks agree) and Social follow-friends (ICE and MoSCoW both defer it; RICE ranks it a borderline 5th) this cycle. The one judgment call is Family plan: if new revenue is not a Q3 goal, swap it for Audiobook recs.

Sensitivity / What Changes the Ranking

  • If Mobile app effort were 6 weeks (not 12), its ICE Ease rises and it becomes a credible Should.
  • If Rec algorithm v2 confidence dropped below 50%, Reading streaks becomes the clear single priority.
  • Family plan’s standing depends entirely on whether Q3 has a revenue goal; with one, it is a Must.

Recommendations (Sequencing)

  • Fund now: Rec algorithm v2, Reading streaks, Export to Goodreads, Family plan
  • Defer: Mobile app (needs a low-effort quarter), Audiobook recs
  • Drop this cycle: Social follow-friends, Public reading profiles
  • Data that would change this: A Kano survey to confirm which features are delighters vs. table stakes; a firmer mobile-app effort estimate

Limitations and Biases

  • RICE under-ranked Family plan purely on reach; do not let it auto-cut revenue-concentrated features.
  • No sequencing dependencies are modeled (e.g., Family plan may need billing changes first).
  • Every score is a PM estimate; a Kano survey would replace the guessed impact values with customer signal.