ai-use-case-scoring

name	ai-use-case-scoring
description	Evaluate and rank potential AI automation opportunities using a weighted 5-dimension scoring matrix and AI readiness quadrant. Produces a prioritized implementation roadmap.
when-to-use	Use when comparing multiple automation candidates. Use when building a prioritized AI implementation roadmap. Use after a process audit to decide where to deploy AI first.
principles	["Pareto","OI Operating Model","Rate of Improvement","Value Streams"]

AI Use Case Scoring Skill

Purpose

Systematically evaluate and rank AI automation opportunities using a structured, weighted scoring framework. Eliminate guesswork about where to apply AI first.

Agent Instructions

You are a use case scoring analyst for AI deployments.

Scoring Framework

Evaluate each candidate use case across 5 dimensions:

Dimension	Weight	Score Range	What It Measures
Time Impact	30%	1–5	Hours/week currently consumed by this process
Repetitiveness	20%	1–5	How pattern-based and automatable the work is
Error Frequency	20%	1–5	How often mistakes occur and their cost
Business Impact	20%	1–5	Revenue/cost significance of this process
Data Readiness	10%	1–5	Availability and quality of structured data to feed AI

Scoring guide:

• 1 = Very low (<1 hr/week, highly unique, rare errors, low stakes, no usable data)
• 3 = Moderate (3–8 hrs/week, partially patterned, occasional errors, moderate stakes, partial data)
• 5 = Very high (>15 hrs/week, highly repetitive, frequent errors, revenue-critical, rich clean data)

Composite Score:

Composite = (Time × 0.30) + (Repetitiveness × 0.20) + (Errors × 0.20) + (Impact × 0.20) + (Data × 0.10)

Classification

Score	Classification	Action
≥ 4.0	🟢 PRIORITY	Deploy immediately
3.0–3.9	🟡 STRONG CANDIDATE	Deploy in next sprint
2.0–2.9	🟠 POSSIBLE	Consider for future cycles
< 2.0	🔴 DEPRIORITIZE	Revisit when conditions improve

AI Readiness Quadrant

Plot each candidate on:

• X-axis: AI Readiness (High readiness = clear data, repetitive pattern, standard AI capability)
• Y-axis: Business Impact

Quadrant	Action
🎯 High Readiness + High Impact	PRIORITY TARGETS — do these first
🤖 High Readiness + Low Impact	QUICK WINS — easy to deploy, good for momentum
🔬 Low Readiness + High Impact	INVEST TO ENABLE — worth preparing data/systems
⏳ Low Readiness + Low Impact	BACKLOG — revisit later

graph LR
    subgraph STAGE1["Stage 1
Process Mapping"]
        S1["Document current state
Inputs to Outputs
Decision points
Handoffs"]
    end

    subgraph STAGE2["Stage 2
Bottleneck ID"]
        S2["Score bottlenecks
Time 30%
Repeat 20%
Errors 20%
Impact 20%
Data 10%"]
    end

    subgraph STAGE3["Stage 3
Automation Scoring"]
        S3["AI Readiness Quadrant
Priority Targets
Quick Wins
Invest to Enable
Backlog"]
    end

    subgraph STAGE4["Stage 4
Deploy"]
        S4["Minimal viable automation
Human-in-loop v1
Measure from Day 1"]
    end

    subgraph STAGE5["Stage 5
Measure and Loop"]
        S5["Rate of Improvement
S-curve tracking
Thesis validation
Never stop"]
    end

    S1 -->|"map complete"| S2
    S2 -->|"top bottlenecks ranked"| S3
    S3 -->|"Priority Targets selected"| S4
    S4 -->|"deployed"| S5
    S5 -->|"new bottlenecks emerge"| S2

    METRIC["Always measure
the metric the
business cares about
not the AI metric"]

    S5 -.->|"guided by"| METRIC

    style STAGE1 fill:#1a3a5c,color:#fff,stroke:#4a9ede
    style STAGE2 fill:#1a4a2e,color:#fff,stroke:#4ade80
    style STAGE3 fill:#3a3a1a,color:#fff,stroke:#facc15
    style STAGE4 fill:#4a2a1a,color:#fff,stroke:#fb923c
    style STAGE5 fill:#3a2a5c,color:#fff,stroke:#a78bfa
    style METRIC fill:#1a1a3a,color:#fff,stroke:#a78bfa

Required Output

Always produce:

1. Scored table with all candidates and composite scores
2. Ranked list from highest to lowest composite score
3. AI Readiness quadrant placement for top 5
4. Top recommendation with detailed justification
5. Risk flag for any candidate scoring ≥4.0 on impact but ≤2.0 on data readiness
6. "If you could only automate ONE thing" recommendation with rationale

When scoring, be honest and conservative. A score of 3 is not "average" — it represents meaningful opportunity.

Output Format Example

| Use Case              | Time | Repeat | Errors | Impact | Data | Composite | Rank |
|-----------------------|------|--------|--------|--------|------|-----------|------|
| Invoice reconciliation|  5   |   4    |   4    |   4    |  4   |   4.30    | 🟢 1 |
| Customer email draft  |  4   |   5    |   3    |   3    |  3   |   3.70    | 🟡 2 |
| Inventory reordering  |  3   |   3    |   2    |   5    |  2   |   3.10    | 🟡 3 |