Prioritization Frameworks
Prioritization is the daily act of strategy. Every roadmap, every sprint, every "can we just add one thing" is a prioritization decision, and how you make those decisions is your strategy whether you've written one down or not. This note is my map of the major methods, when each fits, and the uncomfortable truth underneath all of them.
#The real point of a framework: forcing the conversation, not finding the answer
I want to be clear about this up front, because it's the thing I coach people on most. No framework produces a correct prioritization. They produce a structured argument. Their value is that they force you to make your criteria explicit, apply them consistently, and show your work, so the decision is auditable instead of political or based on who shouted loudest. The Product Vision and Roadmapping behavior names exactly this: "structured frameworks... transparent rationale... say 'no' to focus on the highest leverage." The framework is scaffolding for the no.
The most common prioritization mistake isn't choosing the "wrong" framework. It's framework-hopping, where a new method appears every quarter and nothing is comparable over time. Consistency beats sophistication. Choose one that fits your context, use it for a year, and let the comparability of decisions become the real benefit.
#The major methods and when each fits
| Framework | How it works | Best when | Watch out for |
|---|---|---|---|
| RICE | Reach × Impact × Confidence ÷ Effort | Comparing many similar-sized features with some data | False precision; gameable inputs |
| Impact × Likelihood | 2×2: high/low impact vs likelihood of success | Quick portfolio triage; separating big bets from busywork | Too coarse for fine-grained calls |
| Kano | Classifies features as basic / performance / delight | Deciding what kind of value to invest in | Needs user data; categories drift over time |
| MoSCoW | Must / Should / Could / Won't have | Scoping a release or negotiating with stakeholders | Everything becomes a "Must" without discipline |
| Eisenhower | Urgent × Important | Personal/team time and triage, not roadmaps | Conflates urgency with importance if rushed |
| Cost of Delay / WSJF | Value of shipping sooner ÷ effort | When timing and dependencies dominate | Hard to estimate the cost-of-delay honestly |
The trick is matching the framework to the decision altitude. For triaging a big portfolio of bets, I reach for the Impact-Likelihood Matrix, which is fast and cleanly separates "Home Runs" from "Stuff You Shouldn't Be Doing." For comparing a backlog of comparable features, RICE earns its keep. For what kind of feature to build (table stakes vs differentiator), Kano is the right lens. For release scoping with stakeholders, MoSCoW gives a shared vocabulary. Don't use a scalpel where a machete works, or vice versa.
A note on effort, separate from value: Shreyas Doshi's LNO framework (Leverage / Neutral / Overhead) prioritizes how much effort a task deserves rather than which task to do. It pairs well with any of the above. Once you've decided what to do, LNO tells you how much polish each piece warrants.
#The discipline of saying no
The one thing no framework can do for you is actually say the no. The frameworks give you the rationale; the courage is on you. Saying no is the core skill of prioritization, and it's hard for three reasons: it disappoints people, it's irreversible-feeling, and a "yes" is always easier in the moment. I've learned to make the no specific and reasoned: not "we can't do that," but "we're not doing that because it serves a segment we've deliberately chosen not to prioritize, and here's the bet we're making instead." The reason is the product. A no without a why is a brush-off; a no with a why is strategy.
A scoring framework's output is an input to a decision, not the decision. I've seen teams build the top-RICE item reflexively while it actively contradicted the strategy. The score doesn't know your vision or your defensibility. Sort by the framework, then apply judgment, never the reverse.
#How AI is changing it
Auto-prioritization is explicitly on Reforge's list of commoditized PM activities, and rightly so. A model can ingest your backlog, estimate RICE inputs, and produce a ranked list faster than you can open the spreadsheet. That's useful for the grunt layer: first-pass scoring, surfacing what you might've missed. But it sharpens rather than removes the human job. The inputs (reach, impact, confidence) are estimates that encode judgment and strategy, and AI will estimate them plausibly and generically. The decisions that matter (overriding a high score because it fights the vision, making the unpopular no, deciding which segment to starve) are exactly the taste-and-conviction work the model can't own. Let AI do the arithmetic; keep the choosing.
#Continue Reading
- RICE and Scoring Models is the most popular scoring model, in detail, with its failure modes.
- The Impact-Likelihood Matrix is my go-to for fast portfolio triage.
- Building and Managing a Roadmap is where prioritization decisions get sequenced and communicated.
- The LNO Framework for prioritizing effort per task once you've decided what to do.
- Understanding Trade-offs is the foundational skill prioritization is built on.