How to Prioritize Artificial Intelligence Use Cases: Scoring 4 Criteria and Biases That Distort the Choice
By Pedro Sonda, dooop
To prioritize AI use cases, evaluate each one across four criteria: business value, feasibility (data, technology, and process), risk, and execution capacity. Use a 1 to 5 scale with clear descriptions for each score, agree on weights with leadership before scoring, have each evaluator score independently, then calibrate differences as a group. High-risk cases only proceed with defined controls. The goal is to select a few cases with clear justification and explicitly discard the others.
Updated October 2026.
Why Prioritize with a Method
Gartner predicted that at least 30% of generative AI projects would be abandoned after proof of concept due to poor data, poorly controlled risks, rising costs, or unclear value (Gartner, 2024). These four causes can be assessed before the project starts if the right criteria are used. Among organizations with high AI maturity, 63% perform return analysis and measure impact on customers (Gartner, 2025).
Interpretation: Most causes of abandonment can be anticipated with defined criteria before the project, and more mature organizations already perform this analysis.
dooop Framework: The 4 Criteria with Described Scale
At dooop, discovery prioritizes by value, feasibility, acceptable risk, and execution capacity (dooop). The following scale is a starting point.
| Criterion | Score 1 | Score 3 | Score 5 |
|---|---|---|---|
| Value | Small gain, no clear link to indicators | Relevant gain in one area, with defined indicator | High gain linked to strategic objectives, with clear value chain |
| Feasibility | Data nonexistent or inaccessible; unstable process | Partial data; known process; moderate integration | Available and reliable data; stable process; mature technology |
| Risk (high score = low risk) | Decisions about people or customers without possible control; sensitive data | Moderate risk, controllable with human review | Detectable and reversible error; no sensitive data |
| Execution Capacity | No owner, no team, no time | Defined owner, partial team | Engaged owner, available team, leadership support |
Weights
Agree on weights before scoring. A common starting point is to give more weight to value and feasibility, but companies in regulated sectors may give more weight to risk. The important point is that weights are decided beforehand to avoid adjusting them to favor a preferred case.
The Scoring Process
- Describe each case in the same format: problem, proposed solution, area, data, owner.
- Each evaluator scores independently, without prior discussion.
- Compare scores and discuss only large discrepancies.
- Calculate weighted scores and rank.
- Apply the risk rule.
- Decide the shortlist and record discarded cases with reasons.
The Risk Rule
Cases with risk scores of 1 or 2 do not enter the shortlist solely due to high scores in other criteria. They only proceed with defined controls such as human review, autonomy limits, and legal evaluation. See How to Assess Risk Before Developing a Use Case.
Biases That Distort Prioritization
| Bias | How It Appears | How to Avoid |
|---|---|---|
| Strongest Opinion in the Room | The case of the most influential director rises | Individual scoring before discussion |
| Charm of Novelty | Cases with newer technology seem more valuable | Require value chain; see How to Find AI Opportunities That Really Generate Value |
| Sunk Cost | Already started projects receive generous scores | Evaluate with the same criteria as new ones |
| Optimism About Data | "The data exists" without verification | Require evidence of availability |
| Wanting to Do Everything | Shortlist with 20 cases | Limit by execution capacity |
What to Do Now
- Adjust the scale of the 4 criteria to the company’s reality.
- Agree on weights with leadership before opening cases.
- Score individually and calibrate in a two-hour meeting.
To prioritize opportunities with shared criteria, learn about dooop’s discovery at dooop.com.br/discovery-de-oportunidades-de-ia.
Frequently Asked Questions
Should quick wins have priority? Some yes, to generate learning and confidence. But a portfolio of only quick wins rarely changes outcomes; see How to Create a Prioritized Portfolio of AI Opportunities.
How often to reprioritize? Generally quarterly, or when relevant changes in strategy, technology, or regulation occur.
Also Read
About dooop
dooop is a consultancy based in Caxias do Sul (RS) that prepares leaders and organizations to transform artificial intelligence into business capability through strategy, governance, and training. See dooop.