Opportunity scoring framework

Opportunity scoring is a quantitative discovery technique from Anthony Ulwick’s Outcome-Driven Innovation (ODI) methodology that ranks desired customer outcomes by how important they are and how poorly they are currently satisfied. It tells you which outcome to target, not which feature to build. Candidates who get this distinction on the first try clear the bar. Candidates who apply the formula to features or roadmap items reveal a category error that experienced interviewers reliably catch.

The formula

Opportunity Score = Importance + max(Importance - Satisfaction, 0)

The max() guard is doing real work. If users are more satisfied than an outcome is important, that excess satisfaction adds nothing to the score. There is no negative score, only a floor at Importance alone. This prevents a high-satisfaction, low-importance outcome from appearing as a drag on others.

The double-weighting of Importance is intentional: Ulwick’s argument is that satisfaction without importance is waste. A feature users love but do not care about should not compete for resources with a critical unmet need.

Critical input detail: the formula uses top-2-box percentages (the share of respondents rating 4 or 5 on a 1-5 scale), not raw averages. Applying the formula to mean scores is the most common calculation error.

Score interpretation

Scores run on a 0-10 normalized scale after converting the percentages:

Above 10: strong underserved opportunity, high priority for investment
7-10: moderate opportunity, worth investigating further
Below 7 (falling): well-served or overserved territory

An overserved outcome (high satisfaction, low importance) is information too. It signals over-engineering. A disciplined PM names it as a candidate for reduced investment, not just deprioritized.

Worked example

You are the PM on a B2B crypto trading platform. You have surveyed customers on the outcome: “track the cost basis of each position across exchanges.”

Importance: 74% rate 4 or 5, normalized to 7.4
Satisfaction: 38% rate 4 or 5, normalized to 3.8

Opportunity Score = 7.4 + max(7.4 - 3.8, 0) = 7.4 + 3.6 = 11.0

That score sits above the 10 threshold. In the interview room: “This outcome is clearly underserved. Seventy-four percent of customers say it matters and fewer than four in ten are satisfied with the current state. That is a strong signal to bring solutions against this outcome into discovery.”

If satisfaction were 8.2 instead: Score = 7.4 + max(7.4 - 8.2, 0) = 7.4 + 0 = 7.4. Moderate territory. Investigate, but do not lead a design sprint.

What you are actually scoring

ODI starts from jobs-to-be-done: the functional job customers are trying to accomplish. You decompose that job into desired outcomes, the sub-steps of the job, and score each outcome. You are not scoring product features, user stories, or “areas of the product.” Scoring features with this formula produces noise, not signal. The question on the survey is: “When you [perform the job], how important is it that you are able to [desired outcome]?” and “How satisfied are you currently with your ability to [desired outcome]?”

Running the survey

Ulwick recommends a minimum of 180 respondents for reliable market-level signal. That sample also supports segment-level analysis (e.g., power users vs. occasional users). Lean teams often use 30-50 respondents for early-stage directional cuts, which is defensible as a rough filter but does not support segment breakdowns. In an interview, acknowledge the sample limitation if you are describing a scrappy version of this method.

Where it fits in the discovery process

Opportunity scoring surfaces which outcome to target. The opportunity solution tree then structures how you explore solution ideas and experiments against that outcome. These two tools are paired, not interchangeable. Interviewers at companies with rigorous discovery cultures expect you to name both and explain which layer each operates at.

How it differs from adjacent frameworks

vs. RICE: RICE scores solution ideas (features, experiments) on Reach × Impact × Confidence ÷ Effort. Opportunity scoring operates one level upstream, scoring customer outcomes before you have a solution in mind. Mixing them is a category error.

vs. Kano model: Kano segments satisfaction nonlinearly, separating basic needs, performance attributes, and delighters. It does not surface the magnitude of unmet need directly. Opportunity scoring surfaces that magnitude, making it better for deciding where to focus discovery.

vs. standard gap analysis: Importance minus Satisfaction as a simple difference double-penalizes overserved outcomes. The max() guard in opportunity scoring eliminates that artifact.

The 2026 update

The original formula assumes the binding constraint is knowing what to build. In 2026, engineering effort is no longer that constraint. A lean team can ship against a high-scoring outcome within weeks. This creates two implications Ulwick’s original method does not address.

First, satisfaction scores have shorter half-lives. A gap that existed 18 months ago in an AI-adjacent category is likely smaller now. Incumbents and entrants can close satisfaction gaps quickly. Re-survey more frequently than you would have in 2022.

Second, importance alone does not prove viability. You can find an outcome scoring 11.0 and build a solution in a month. But if the segment will not pay a price that covers cost and generates margin, feasibility being cheap just means you reach failure faster. The necessary addition is a third survey question: “If a solution fully addressed this need, how much would you pay for it?” This is not in Ulwick’s original method. In 2026 it is the required update, because viable is now the scarce variable, not buildable.

Third, a high opportunity score does not tell you the right delivery modality. Whether the outcome should be solved by a proactive agent, an embedded workflow integration, or a deliberate user action is a question opportunity scoring cannot answer. That is where the opportunity solution tree and qualitative research still carry full weight. The lovable standard applies to how the solution is delivered, not just whether the need is real.

The interview failure mode

Most candidates describe the formula correctly and then stop. The interviewer follows up: “You have an outcome with a score of 11. What do you do next?” The weak answer: nothing specific, or a vague “we would explore solutions.” The strong answer maps directly to a decision: bring the outcome into the OST as the target outcome, map assumptions about solutions, and add a pricing question to the next survey pass before committing a design sprint to it. Knowing the score is only useful if it changes what you do.