# Automation opportunity scorecard

Compare a few real opportunities using the same criteria. This worksheet makes assumptions explicit; it is not a validated scoring model or a promise of returns.

Revision: 2026-09-24

## When to use it

When several departments propose AI projects and you need to decide what to investigate with limited data, people and budget.

## Expected output

A reasoned comparison, including missing evidence, blocking conditions and a decision about the next experiment.

## 1. Document the value

Complete one worksheet per opportunity. Use a consistent observation period and avoid counting the same operational improvement under multiple benefits.

### Outcome and beneficiary

Identify who benefits and which operational behaviour should change.

________________

### Volume, current effort and source

Record cases per period, observed handling time, known error costs and where each figure came from.

________________

### Expected benefit and uncertainty

Separate released capacity, actual spending reductions and commercial outcomes that remain unproven.

________________

> Illustrative example, replace with your own: Illustrative example: preparing 200 monthly summaries uses 20 hours. A trial must measure remaining review time; it cannot assume that all 20 hours disappear.

## 2. Check feasibility

Check dependencies before comparing scores. An attractive project without access or an available owner is not ready to begin.

### Data, access and integrations

List what is already available and what needs authorisation or third-party work.

________________

### Consequences of errors

Describe a plausible harmful action, how it would be detected and whether it can be reversed.

________________

### Owner availability

Identify who can supply examples, clarify rules and review the trial.

________________

> Illustrative example, replace with your own: Illustrative example: summaries can remain internal drafts. Automatic price updates need approval rules and write controls before a meaningful trial can begin.

## 3. Record the decision

If using a scale, define its values in advance. Keep non-negotiable conditions separate from any combined score.

### Agreed criteria and weights

Define how every proposal will be compared on value, effort, data reliability and reversibility.

________________

### Conditions that must be met

List requirements that must hold before the trial starts, even when its overall score is high.

________________

### Decision and next review

Choose trial, gather evidence or defer; assign an owner and a review date.

________________

> Illustrative example, replace with your own: Illustrative example: trial summaries because usable records and a reviewer exist. Defer automated pricing until its approval rules and recovery process are defined.

## Pitfalls to avoid

- Treating a numerical average as evidence of feasibility.

- Rewarding visibility without checking data and ownership.

- Equating released capacity with cash savings without an operational plan.

Stolen Orbit — https://stolenorbit.com/en/templates/automation-opportunity-scorecard/