# The Four Key Metrics

Every opportunity in the funnel gets scored on exactly four metrics. These four determine priority. Teach each one before scoring the first opportunity; after that, score in rhythm.

For each opportunity, propose a suggested score with a one-sentence rationale, then let the executive confirm or override. Their number wins. The learning happens in the discussion of the gap between your number and theirs.

## The metrics (all 1 to 5)

### 1. Risk: "Could using AI for this negatively impact the business?"
1 = Low risk · 5 = High risk. Calibrate against the risk posture determined in Phase 1. A customer-facing output in a HIPAA-regulated business is not the same 3 as an internal drafting task. Data sensitivity, regulatory exposure, brand exposure, and irreversibility all push this up.

### 2. Business Impact: "Would using AI here drive business results, or avoid loss?"
1 = Low impact · 5 = High impact. Both offense (revenue, throughput, speed) and defense (loss avoidance, error reduction, compliance) count. Anchor to numbers from the company profile: hours per month, revenue per FTE, cost of the process today.

### 3. Feasibility: "Can we use AI for this?"
1 = Not feasible · 5 = Very feasible. Technical maturity plus data availability plus integration surface. A mature model with clean accessible data scores high; anything needing a data-engineering overhaul scores low.

### 4. Human vs. AI: "Should we use AI for this?"
1 = Human should do it · 5 = AI should do it. This is the mission-and-perception axis, distinct from feasibility. Some tasks are feasible for AI but belong with humans (final hiring decisions, sensitive customer moments). Some tasks humans do today but nothing about them requires human judgment.

## Priority Score

```
Priority = Business Impact + Feasibility + HumanVsAI + (6 - Risk)
```

Range 4 to 20. Risk is inverted so that low risk raises priority. This mirrors the workbook's Process Inventory logic: high impact, high feasibility, AI-appropriate, low risk floats to the top.

## Priority bands

| Band | Score | Meaning |
| --- | --- | --- |
| **Pilot Now** | 16-20 | High-priority. Advances to the Deep Economic Assessment. Launch an 8-week pilot for whatever passes. |
| **Build Readiness** | 12-15 | Strong opportunity with a data, infrastructure, or governance gap. High-priority if the gap is closable this quarter; otherwise icebox with a named unlock condition. |
| **Low-Stakes Experiment** | 8-11 | Moderate opportunity. Use for fast, cheap experiments that build the AI muscle. Goes to the icebox with an experiment suggestion. |
| **Defer** | 4-7 | Icebox. Revisit on the next audit cycle. |

## The high-priority cut

Opportunities that do not pop to the top of the list when rated by the four key metrics move to the **icebox**. Nothing is deleted. The icebox is an asset: technology costs fall, models improve, and iceboxed items become viable. Revisit the icebox every time you brainstorm.

Default cut: **Pilot Now** items are high-priority; **Build Readiness** items are judgment calls to make WITH the executive; everything else goes to the icebox. Keep the high-priority list honest. Five to eight items is a roadmap; twenty is a wish list.