Prioritization · pick the method, don't run them all
The arithmetic is canon the model already reproduces; the skill is choosing which one to run. Eight methods, each with what it's for, when not to reach for it, and the one calibration it quietly gets wrong.
Eight methods — the skill is picking the one the decision calls for.
Above the fold
Match the method to the decision.
Not to the backlog size. The selection table is the judgment; the arithmetic is canon the model already reproduces.
Kill anything under 50% confidence.
In RICE, confidence is an honest discount, not a fudge factor — below 50%, kill the item rather than rank it.
Classify Type-1 vs Type-2 before speed.
Naive RICE inverts on interdependent items; switch to cost-of-delay / WSJF when the call is time-sensitive.
The selection table
Match the method to the decision, not to the backlog size. ~ numbers are directional rules of thumb.
Reach × Impact × Confidence ÷ Effort
Impact × Confidence × Ease
cost-of-delay ÷ duration
Must · Should · Could · Won't
basic · performance · delighter
importance + max(importance − satisfaction)
Leverage · Neutral · Overhead
Growth / Engagement / Monetization
Watch-outs
Naive RICE inverts on interdependent items (and where reach/effort are mostly noise) — switch to cost-of-delay / WSJF when the call is time-sensitive, and classify Type-1 (irreversible) vs Type-2 (reversible) before deciding speed. Recency/capital calibration: monetization timing differs in cheap- vs tight-capital eras — a "grow now, monetize later" default is a recency artifact, not a law.
Pre-mortem · Klein; Doshi in PM
"It's 6 months out, this failed — why?"
Portfolio mode · two instruments, one move
Set strategic buckets with %-capacity before scoring, rank within each bucket, and ring-fence the small-bet bucket from the urgent-core default. CPO work shifts from prioritizing features to allocating capacity across horizons (Perri).
bucket %s directional · illustrative split The first instrument · Strategic Buckets (Cooper): core / growth / new-bets.
The second instrument · Aggregate Project Plan (Wheelwright & Clark, HBR 1992)
Buckets set the %-split; this sets what kind of work fills them. The origin of the platform/derivative/breakthrough vocabulary: classify every dev project by product change × process change into five types, set a desired type-mix % against capacity, and hold a capacity cushion.
Derivative
small change to an existing product/process
Platform
the shared base later variants build on
Breakthrough
new product and new process
R&D
create the know-how that feeds the others
Partnership
delivered with or through another party
Hold a capacity cushion
Don't commit 100% — slack absorbs crises; over-commitment is what causes cascading delays. Mature market → weight platforms; growth → weight breakthroughs; revisit every 6-12 months. Distinct from Cooper buckets (fund by horizon) and RICE (score single projects) — APP sets the mix, the capacity math, and the multi-year sequence.
8-step method, mix math, sequencing → PRO: strategyTry it
Three prioritization calculators, live — RICE with the kill rule, the WSJF/CD3 sequencer, and the Kano classifier. Edit the numbers; the verdict updates as you type.
The RICE scorer needs JavaScript to run. Run it on the Tools page.
The WSJF sequencer needs JavaScript to run. Run it on the Tools page.
The Kano classifier needs JavaScript to run. Run it on the Tools page.