On this page
A feature prioritization framework is a repeatable way to decide which features get built first, so the order rests on stated reasons instead of the loudest voice in the room. No single one fits every decision. RICE ranks a long backlog, MoSCoW cuts scope for a fixed release, and Kano sorts features by how customers react to them.
This guide compares seven frameworks, takes each scale and limit from the people who published it, and runs one example backlog through two of them. If your list keeps growing because nobody has decided who the product is for, that is a strategy gap. Product strategy consulting settles that choice before any scoring starts.
What is a feature prioritization framework?
A feature prioritization framework is an agreed set of inputs and a rule for combining them, applied to every candidate feature in the same way. It turns "what should we build next?" into a question the team can answer with evidence and then explain to anyone who asks.
Product feature prioritization has to settle three things, and a framework helps with only two of them:
- What fits: which features the team can build with the time and people it has.
- What order: which of those come first, and what waits.
- What matters: which customers and which goal the product serves. No framework decides this. It is the input every score depends on.
That third point is why impact needs a target. A team that has agreed on a north star metric can ask how much a feature moves it. A team without one ends up scoring opinions.
Seven feature prioritization frameworks compared
The seven frameworks fall into three groups: those that score features (RICE, ICE, weighted scoring, WSJF), those that sort them into classes (MoSCoW, Kano) and one that plots them on a grid (value vs effort). The table shows what each needs, where it works and how it goes wrong.
| Framework | Inputs | Suited to | Where it fails |
|---|---|---|---|
| RICE | Reach, impact, confidence, effort | Ranking a long backlog against one goal | Guessed inputs give an exact-looking wrong order |
| ICE | Impact, confidence, ease | Ranking many small experiments quickly | All three inputs are opinions |
| MoSCoW | Four classes, agreed by the team | Cutting scope for a fixed date | Everything is called a Must Have |
| Kano | Customer survey answers | Telling expected basics from delighters | Slow to run, and silent on cost |
| Value vs effort | Two estimates per feature | A first pass in one meeting | Nobody defines "value" |
| Weighted scoring | Criteria, weights and scores | Trade-offs across several goals | Weights get tuned to the answer wanted |
| WSJF | Cost of delay and job duration | Sequencing work when delay costs money | Relative estimates differ between teams |
The sections below take each one in turn, with its source.
RICE, ICE and WSJF: scoring a long backlog
RICE gives each feature one number: reach times impact times confidence, divided by effort. Use it when you have more candidate features than you can debate one by one, and a single goal to measure impact against.
RICE was developed at Intercom, and Sean McBride set it out in the post RICE: Simple prioritization for product managers. The four factors use these scales:
- Reach: the number of people or events in a time period, such as customers per quarter, taken from product metrics where possible.
- Impact: a multiple-choice scale of 3 for massive, 2 for high, 1 for medium, 0.5 for low and 0.25 for minimal.
- Confidence: 100% for high, 80% for medium and 50% for low.
- Effort: person-months across product, design and engineering, in whole numbers with a minimum of half a month.
Intercom describes the result as total impact per time worked. Confidence does the quiet work: it marks down exciting ideas that have no data behind them.
A worked RICE example
Here are three features for a business software product, invented for illustration, with reach counted in customers per quarter:
- Onboarding checklist: 900 × 2 × 50% ÷ 3 = 300
- Bulk CSV import: 600 × 1 × 80% ÷ 2 = 240
- Single sign-on: 40 × 3 × 100% ÷ 3 = 40
The onboarding checklist comes first with a score of 300, yet half of that rests on a guess: its confidence is only 50%. The bulk CSV import follows with a score of 240 on firmer data. Single sign-on comes last with a score of 40, because only 40 customers a quarter would use it.
The order is a starting point. Intercom's post says plainly that scores are not a hard rule: one project may be a dependency for another, or a feature may be table stakes for selling to certain customers. The MoSCoW section below picks this list up again.
ICE: the lighter version
ICE scores each idea on impact, confidence and ease, then combines the three into one number. It is quicker than RICE because there is no reach figure to look up, which suits a long list of small experiments. The cost is that all three inputs are opinions, so two people can score the same idea very differently.
WSJF: when delay has a cost
Weighted shortest job first (WSJF) ranks work by the cost of delaying it divided by how long it takes, so short jobs with a high cost of delay go first. Use it when timing matters as much as benefit: contract dates, renewals and deadlines set by someone else.
The Scaled Agile Framework (SAFe) uses WSJF to order its backlogs. Its page on weighted shortest job first defines the measure as the relative cost of delay divided by the relative job duration. The inputs it names are user and business value, time criticality, risk reduction or opportunity enablement, and job size. The same page opens with a line from Don Reinertsen's book The Principles of Product Development Flow, on putting a number on the cost of delay.
WSJF asks for more estimates than any other framework here, and all of them are relative, so numbers from two teams cannot be compared directly. For a small product team it is often more process than the decision needs.
MoSCoW: cutting scope for a fixed release
MoSCoW sorts requirements into four classes: Must Have, Should Have, Could Have and Won't Have this time. Use it when the date and the team are fixed and the scope is the thing that has to give.
The method belongs to DSDM, an agile project framework first created in 1994 and published by the Agile Business Consortium. Its guide, What is MoSCoW Prioritization?, defines the classes this way:
- Must Have: there is no point delivering on the target date without it, or the product would be unsafe or not legal. If a workaround exists, even a manual and painful one, it is not a Must Have.
- Should Have: important but not vital. Leaving it out hurts, but the solution is still viable.
- Could Have: wanted, with less impact if left out. These are the first items dropped when the deadline is at risk.
- Won't Have this time: agreed as out of scope for this timeframe, and written down so it is not quietly added back.
The part most teams skip is the limit. DSDM recommends that Must Haves take typically no more than 60% of the effort, with around 20% set aside as Could Haves. For a release with 10 person-weeks of capacity, that means at most 6 on Must Haves and about 2 on Could Haves, which leaves about 2 for Should Haves.
Now return to the RICE example. Suppose two signed contracts depend on single sign-on. It scored lowest, yet it is the only Must Have on the list, and the other two features are simply ranked by score. A team that used RICE alone would have shipped it last.
Kano model feature prioritization: sort by reaction
Kano model feature prioritization sorts features by how customers feel when a feature is present and when it is missing, instead of by one score. Use it before you rank anything, to learn which features are expected basics and which would set the product apart.
The model comes from Noriaki Kano and three co-authors, in the paper Attractive Quality and Must-Be Quality, published in the Journal of the Japanese Society for Quality Control in 1984. The authors argued that quality has two dimensions, not one, and tested the idea with a questionnaire survey of consumers about a TV set and a table clock. The paper names three classes of quality:
- Must-be: customers expect it. Its absence causes complaints, and its presence earns little credit. Password reset is an example.
- One-dimensional: the more of it, the happier the customer. Speed and storage often behave this way.
- Attractive: customers did not expect it and are pleased to find it, but do not miss it when it is absent.
For prioritization, the order usually follows from the classes: cover the must-be features first, compete on the one-dimensional ones, and choose a small number of attractive ones on purpose.
What a feature prioritization survey should ask
A Kano-style feature prioritization survey asks about each feature twice: how would you feel if the product had it, and how would you feel if it did not. Use the same answer options for both questions, and keep the list of features short enough that people finish it.
Send the survey to one customer segment at a time, because a basic for one group can be a delighter for another. A survey also tells you what people say, not why. Interviews fill that gap, and they are part of our product discovery services.
The feature prioritization matrix: value vs effort
A feature prioritization matrix plots each feature on two axes, value and effort, and reads the order from the four quadrants. It is the fastest framework to run and the easiest to get wrong, because the word "value" does a lot of unexamined work.
- High value, low effort: do these first.
- High value, high effort: plan them as deliberate bets, one at a time.
- Low value, low effort: use them to fill gaps, and expect them to add up to little.
- Low value, high effort: drop them and say so.
The matrix works well for a first release, when there is little usage data to score with. Our MVP scope planner applies the same idea to a first version.
Weighted scoring: when value means several things
Weighted scoring splits "value" into named criteria, gives each a weight, and scores every feature against all of them. Say the weights are customer value 50%, revenue 30% and strategic fit 20%, and single sign-on scores 2, 5 and 4 out of 5. Its weighted score is 2 × 0.5 + 5 × 0.3 + 4 × 0.2 = 3.3.
The risk is that weights get adjusted until a favourite feature wins. Agree the criteria and weights before anyone scores a feature, and change them only when the strategy changes.
How to choose a feature prioritization framework
Choose a feature prioritization framework by the decision in front of you, not by which one is most popular. The table matches common situations to a starting point.
| Your situation | Start with | Why |
|---|---|---|
| A long backlog and one clear goal | RICE | Reach and confidence expose pet ideas |
| A fixed date and too much scope | MoSCoW | The effort limits build in contingency |
| Unsure what customers expect | Kano | It separates basics from delighters |
| A first release with little data | Value vs effort | It needs only two estimates per feature |
| Several goals pulling apart | Weighted scoring | The weights make the trade-off visible |
| Dozens of small experiments | ICE | It is quick enough to run every week |
| Deadlines and renewals set the order | WSJF | It puts a number on the cost of waiting |
The frameworks also combine well, because they answer different questions. A common sequence is Kano to learn what customers expect, RICE to rank what is left, and MoSCoW to cut the result down to one release.
Whatever comes out feeds the next two documents. The order goes into the roadmap, which our guide on how to build a product roadmap covers, and each chosen feature becomes stories for the team, as in these user stories examples.
Where every framework fails
Every framework fails in the same place: it is only as good as its inputs and the strategy behind them. A score can rank features for a chosen customer and goal. It cannot choose the customer or the goal.
- Guessed inputs. A reach figure nobody measured still produces a score with a decimal point. Mark guesses with low confidence, and replace them with product metrics as soon as you can.
- No goal. Impact on what? Without one agreed target, each person scores impact against their own.
- Gaming. Once people learn which input moves the result, favourite features drift upward. Have someone outside the feature's sponsor set the effort and confidence.
- Dependencies and table stakes. A low-scoring feature can block a high-scoring one or a sale, as single sign-on did in the example above.
- Scoring once. Evidence changes. Score again when a test finishes or a large customer's needs shift, not only at planning time.
When the scores keep producing an order nobody believes, the framework is rarely the problem. The usual cause is that the team has not agreed who the product is for.
How we prioritize features with your team
We work with founders and product leads whose backlog grows with every customer call. As your product strategy consultant, we first settle who the product serves and which goal comes first. Then we pick the framework that fits the decision and score the backlog with your team, using your usage data and customer interviews.
You leave with an ordered list and the reasons behind it. Every engagement is under an NDA, and the IP is always yours. Each project is different, so we quote it after a short call.
Frequently asked questions
Which feature prioritization framework should a startup use?
Start with the decision. For a first release with little data, a value vs effort matrix is enough. Once there is usage data and a backlog too long to debate, move to RICE. When a date is fixed and the scope is not, use MoSCoW. Most teams end up combining two of them.
What is the difference between RICE and ICE?
RICE multiplies reach, impact and confidence, then divides by effort. ICE uses impact, confidence and ease, with no reach figure. RICE takes longer because reach should come from product metrics, but it stops a team from favouring features only a few customers would use. ICE is faster and suits lists of small experiments.
How do RICE, MoSCoW and Kano compare?
They answer different questions. RICE produces a ranked list from four estimates. MoSCoW sorts requirements into four classes so a fixed release keeps some contingency. Kano uses a customer survey to separate expected basics from features that please. Many teams use Kano to learn, RICE to rank and MoSCoW to cut scope.
What is a feature prioritization matrix?
A feature prioritization matrix is a grid with value on one axis and effort on the other. Each feature is placed in one of four quadrants: do first, plan as a bet, fill in, or drop. It takes one meeting and two estimates per feature, which makes it a sensible first pass before any detailed scoring.
What should a feature prioritization survey ask?
Ask about each feature twice: how the customer would feel with it, and how they would feel without it. Use the same answer options for both questions. Keep the list short, survey one customer segment at a time, and follow up with interviews, because a survey shows what people say and not why.
Can you use more than one prioritization framework?
Yes, as long as each has a separate job. A team might use Kano once a year to check what customers expect, RICE each quarter to rank the backlog, and MoSCoW for each release. Running two scoring frameworks on the same list at the same time mostly produces two arguments.
Sources and update history
- Intercom blog, RICE: Simple prioritization for product managers (Sean McBride)
- Agile Business Consortium, What is MoSCoW Prioritization?
- Agile Business Consortium, The DSDM Agile Project Framework
- Kano, Seraku, Takahashi and Tsuji, Attractive Quality and Must-Be Quality, Journal of the Japanese Society for Quality Control 14(2), 1984 (J-STAGE)
- Scaled Agile Framework, Weighted Shortest Job First
Update history
- First published.
