Prioritizing AI Use Cases
Learning Objectives
- apply the impact vs effort matrix to a set of AI use cases
- evaluate use cases using five impact factors and three effort factors
- select a quick win and a strategic bet from a prioritized backlog
- distinguish high-value use cases from distractions using the matrix
Core Concepts
The impact vs. effort matrix is a two-axis prioritization tool. The horizontal axis measures effort: the cost, complexity, and risk to implement. The vertical axis measures impact: the value delivered to the business or the user. Each use case is scored and placed in one of four quadrants.
The power of the matrix is not the quadrants themselves. It is the scoring conversation that gets you there.
The Four Quadrants
| Quadrant | Impact | Effort | Decision |
|---|---|---|---|
| Quick Wins | High | Low | Do first |
| Strategic Bets | High | High | Plan and resource carefully |
| Fill-ins | Low | Low | Do if capacity allows |
| Distractions | Low | High | Avoid or defer indefinitely |
Quick Wins are where adoption starts. They deliver visible value fast, build credibility for the AI program, and give teams real feedback before they invest in harder work.
Strategic Bets are worth pursuing, but they require deliberate planning: dedicated time, clear ownership, and staged execution. They are not good candidates for a first pilot.
Distractions are the dangerous quadrant. They often arrive with technical appeal or executive enthusiasm. The matrix makes it easier to say "not now" with evidence.
Scoring Impact: Five Factors
Each use case is rated on five impact factors. Use a simple 1-3 scale: 1 (low), 2 (medium), 3 (high).
| Factor | What It Measures | Example (X-company) |
|---|---|---|
| Time saved | Hours recovered per week across the team | PRD drafting: 3 (saves 4-6 hrs/PM/week) |
| Quality improvement | Reduction in errors, rework, or variance | Contract review: 3 (reduces missed clauses) |
| Revenue impact | Direct link to conversion, retention, or expansion | Churn prediction: 3 (targets at-risk accounts) |
| User experience | Improvement to the end-user product or service | Onboarding docs: 2 (faster time-to-value for clients) |
| Strategic value | Builds capability or differentiation that compounds | Churn prediction: 3 (proprietary risk model) |
Sum the five scores. Maximum is 15. This total becomes your impact score.
Scoring Effort: Three Factors
Each use case is rated on three effort factors. Use the same 1-3 scale, where 3 means high effort.
| Factor | What It Measures | Example (X-company) |
|---|---|---|
| Data readiness | Quality and availability of data required | Churn prediction: 3 (needs 18 months of usage data, labeled) |
| Integration complexity | Systems, APIs, and workflows touched | Sales email personalization: 2 (CRM integration required) |
| Team capability | Skills gap to build and maintain the solution | PRD drafting: 1 (any PM can run it with a good prompt) |
Sum the three scores. Maximum is 9. This total becomes your effort score.
Placing Use Cases on the Matrix
Once scored, plot each use case by its effort score (horizontal) and impact score (vertical). The split points are subjective but a useful starting default is the midpoint of each range: impact above 8 is high, below 8 is low; effort above 5 is high, below 5 is low.
This produces a visual picture of where each use case sits, and makes trade-offs between options concrete rather than argumentative.
Key Points
- The matrix separates the "is this valuable?" question from the "can we do this now?" question
- Impact is scored across five factors: time saved, quality, revenue, user experience, and strategic value
- Effort is scored across three factors: data readiness, integration complexity, and team capability
- Quick Wins belong on the roadmap first; Distractions belong on a parking lot
- The scoring conversation is as valuable as the final placement: it surfaces hidden assumptions
Actionable Takeaways
- Collect all AI use case candidates from your team before scoring anything. Evaluation is meaningless if the list is incomplete.
- Score each use case independently before discussing as a group. Comparing scores before individuals have formed views produces anchoring bias.
- When a use case scores high on impact but you cannot agree on effort, it means data readiness or integration complexity is unknown. That unknowing is a risk: treat it as high effort until resolved.
- Pick one Quick Win to pilot within the next sprint cycle. Visible early progress is the most effective argument for continued AI investment.
- Designate one Strategic Bet as the six-month horizon goal. Assign an owner and begin the scoping work now, before you need the capacity.
Practical Examples
X-company's Eight Use Cases: Full Scoring
X-company's leadership team and product lead scored all eight candidates independently, then compared results. Here is the consolidated scoring:
Impact Scores
| Use Case | Time Saved | Quality | Revenue | UX | Strategic | Total |
|---|---|---|---|---|---|---|
| PRD drafting | 3 | 2 | 1 | 1 | 2 | 9 |
| Customer support triage | 2 | 2 | 2 | 2 | 1 | 9 |
| Contract review | 2 | 3 | 2 | 1 | 2 | 10 |
| Onboarding documentation | 2 | 2 | 2 | 3 | 1 | 10 |
| Sales email personalization | 1 | 1 | 2 | 1 | 1 | 6 |
| Sprint planning | 2 | 1 | 1 | 1 | 1 | 6 |
| Churn prediction | 2 | 3 | 3 | 2 | 3 | 13 |
| Competitive analysis | 2 | 2 | 1 | 1 | 2 | 8 |
Effort Scores
| Use Case | Data Readiness | Integration | Team Capability | Total |
|---|---|---|---|---|
| PRD drafting | 1 | 1 | 1 | 3 |
| Customer support triage | 2 | 2 | 2 | 6 |
| Contract review | 2 | 3 | 2 | 7 |
| Onboarding documentation | 1 | 1 | 1 | 3 |
| Sales email personalization | 2 | 2 | 2 | 6 |
| Sprint planning | 1 | 2 | 1 | 4 |
| Churn prediction | 3 | 3 | 3 | 9 |
| Competitive analysis | 1 | 1 | 2 | 4 |
Matrix Placement
Using the default split (impact above 8 = high; effort above 5 = high):
| Quadrant | Use Cases |
|---|---|
| Quick Wins (high impact, low effort) | PRD drafting, Onboarding documentation |
| Strategic Bets (high impact, high effort) | Contract review, Customer support triage, Churn prediction |
| Fill-ins (low impact, low effort) | Sprint planning, Competitive analysis |
| Distractions (low impact, high effort) | Sales email personalization |
X-company's Decisions
Quick Wins selected: PRD drafting and onboarding documentation.
Both scored effort of 3, meaning any PM with a well-constructed prompt can start this week. PRD drafting saves an estimated four to six hours per PM per week across a nine-person product team. Onboarding documentation reduces the time professional services clients spend getting up to speed, directly improving their early experience with the product.
Strategic Bet selected: Churn prediction.
X-company's highest impact score (13) but also the hardest to build (effort 9). It requires eighteen months of labeled usage data, a model training pipeline, and integration into the account management workflow. The CEO nominated it as the six-month horizon initiative. The engineering lead will begin a data audit in the next sprint.
Deferred: Contract review and customer support triage are high-impact but involve legal data sensitivity and support tooling integrations that X-company's current team cannot own cleanly. Both go to the backlog with a six-month revisit date.
Parked: Sales email personalization scored low on impact (6) and moderate on effort (6). The revenue attribution is unclear and the CRM integration adds complexity without enough upside to justify it now. Competitive analysis and sprint planning remain as fill-ins: lightweight experiments the team can run when capacity allows.
Discussion Prompt Look at your own organization's AI candidate list. Which use cases are being discussed most loudly? Score them using the five impact factors and three effort factors. Does the loudest idea end up in Quick Wins or in Distractions?
Implementation Workflow
Complete this workflow with your team using your own organization's AI use case candidates. If you do not have a list yet, generate one before starting: ask each stakeholder to submit their top two or three ideas before the session.
Collect all candidates. Gather every AI use case idea your team has surfaced. Write each one as a single action phrase: "automate X", "generate Y", "predict Z". Aim for six to ten candidates. Fewer than five gives you nothing to compare; more than twelve makes scoring sessions impractical.
Score impact independently. Each participant scores every use case on the five impact factors (time saved, quality improvement, revenue impact, user experience, strategic value) using a 1-3 scale. Do this before any group discussion. Record scores in a shared spreadsheet or table.
Score effort independently. Each participant scores every use case on the three effort factors (data readiness, integration complexity, team capability) using the same 1-3 scale.
Compare and align. Reveal scores and identify divergence. When two participants score the same factor more than one point apart, discuss why. This conversation surfaces hidden assumptions about data availability, team skill gaps, and integration risk. Do not average disagreements away: resolve them.
Calculate totals and plot. Sum each use case's five impact scores into an impact total (max 15). Sum each use case's three effort scores into an effort total (max 9). Plot each use case on a two-axis chart using impact as the vertical axis and effort as the horizontal axis.
Draw the split lines. Use impact 8 and effort 5 as your default split points. Adjust if your distribution clusters: the goal is a useful separation, not mathematical precision.
Assign each use case to a quadrant. Work through the list together. For anything on a boundary, ask: "If we had to pick a side, which quadrant would this fall into and why?"
Select your Quick Win. From the Quick Wins quadrant, choose one use case to pilot in the next two to four weeks. Assign an owner, set a scope limit (no more than one week of engineering time for a first pilot), and define what success looks like.
Select your Strategic Bet. From the Strategic Bets quadrant, choose one use case as your six-month horizon initiative. Assign an owner and schedule a scoping session within the next two weeks.
Document the parking lot. Record every Distraction and Fill-in with a brief rationale for why it was deprioritized. This prevents the same ideas from re-entering the discussion at the next planning cycle without new evidence.