AI Readiness Scorecard
A practical scorecard for deciding whether a company is ready to deploy AI into one real workflow.
Direct Answer
AI readiness is not whether a team has used ChatGPT. Readiness means one workflow has enough value, evidence, ownership, review capacity, risk control, and measurement to justify implementation.
How to use this
- Step 1: Score each dimension from 0 to 3.
- Step 2: Use 0 when the answer is unknown, not when the team is optimistic.
- Step 3: Score at least three candidate workflows before picking the first one.
- Step 4: Choose the workflow with the strongest mix of value, evidence, and reviewability.
- Step 5: Do not advance a workflow with high risk and no owner, even if the upside sounds large.
Scorecard dimensions
- Workflow clarity: Can the team name the repeated process and trigger?
- Business value: Does delay, rework, missed follow-up, or leakage matter enough to fix?
- Evidence quality: Are the required forms, notes, fields, examples, or policies available and trusted?
- Owner capacity: Is one person accountable for reviewing and improving the output?
- Risk boundary: Can the team name what AI must not decide, send, change, or promise?
- Measurement: Is there a baseline metric and a review cadence?
Worksheet prompts
- 0 points: Unknown, missing, disputed, or owned by nobody.
- 1 point: Partially clear, but the workflow would need cleanup before implementation.
- 2 points: Clear enough to design a first version with review and stop rules.
- 3 points: Ready for implementation: evidence exists, owner is named, risk is bounded, and metric is measurable.
- Recommended threshold: A first workflow should usually score 13+ out of 18 with no zero in owner, evidence, or risk boundary.
- Disqualifier: Any workflow that changes money, legal exposure, customer commitments, or records without review should be redesigned before implementation.
What the score means
A high score means the workflow is a candidate for a first implementation sprint. A low score does not mean AI is useless. It means the team should clean up the workflow before expecting AI to improve it.
What most readiness tests miss
Many readiness tests ask about culture, tools, or strategy. ADA scores whether the next workflow can be run, reviewed, measured, and corrected by the people who own it.
How to use it
Score three to five candidate workflows, then choose the one with the best mix of value, evidence, low risk, and reviewability.
Quality bar
- The scorecard compares workflows, not departments.
- The scoring conversation exposes unknowns instead of hiding them.
- The output is a decision: implement, clean up first, park, or reject.
- The chosen workflow has a named reviewer and a measurable business reason to exist.
- The team can explain why this workflow is first and why other ideas are later.
Where This Helps
Research basis
- Microsoft Responsible AI Resources: Uses impact assessment and human review as practical operating ideas.
- OECD AI Principles: Supports human agency, oversight, accountability, and risk management.
- NIST AI RMF Core: Supports documenting context, impacts, measurement, and management before deployment.
Related Resources
FAQ
- What is an AI readiness scorecard?: It is a scoring tool for deciding whether a workflow has enough clarity, evidence, owner capacity, risk control, and measurement to deploy AI.
- Who should complete it?: The person closest to the workflow should complete it with the owner who will review the AI output.
- What score is good enough?: ADA would rather see one workflow with strong evidence and a clear owner than five exciting ideas with no review path.