Score Value, Readiness, and Risk
Score an AI automation opportunity by business value, operational readiness, and risk without hiding weak evidence in one total.
Keep three scores visible
One total score creates false certainty. A workflow can have high business value and weak data. Another can be easy to automate but barely matter. Keep business value, readiness, and risk visible as separate dimensions, then write the reason behind each score.
| Dimension | Question | Evidence |
|---|---|---|
| Business value | Would a better result change capacity, speed, quality, customer experience, revenue, or cost? | Volume, time, delay, errors, rework, missed actions |
| Readiness | Can the team define, access, test, and own the workflow? | Clear rules, usable data, integration access, owner, examples |
| Risk | What happens when the system is wrong, unavailable, or misused? | Sensitive data, commitments, reversibility, exception load, oversight |
Build the business-value case from visible inputs
For repeated labour, record occurrences, minutes per occurrence, loaded hourly cost, the share that is repetitive, and the share a proposed system may support. Include recurring software, data, review, and maintenance costs. Leave uncertain inputs visible and test a range.
The Automation Opportunity Calculator can organize these assumptions. Its result does not automatically value fewer errors, faster response, or better customer experience. Give each additional outcome its own baseline.
If the estimate depends on future revenue, state the chain of assumptions. Faster response may improve the chance of a conversation, but it does not prove booked revenue. Choose an operational measure the team can observe first.
Score readiness from the current-state map
Use a five-point scale for each factor: boundary clarity, rule clarity, input quality, system access, example availability, exception knowledge, business ownership, and operating ownership. A low score is not failure. It identifies discovery work.
Watch for readiness theatre. A clean spreadsheet does not prove that the inputs are accurate. An API does not prove that the team has permission or budget to use it. A written rule does not prove that employees agree on its meaning. Verify the constraint before funding the build.
Describe risk as a consequence and control
Risk should not be a vague red flag. Write the failure, consequence, and proposed control. “AI risk: high” does not guide a design. “The system could route an existing customer to a new-business queue; check account status before routing and require review when records conflict” does.
NIST describes its AI Risk Management Framework as a voluntary way to include trustworthiness considerations in the design, development, use, and evaluation of AI systems. For this course, use that practical posture: assign an owner, test realistic failures, monitor actual use, and revise controls as evidence changes.
Worked example: comparing two candidates
The enquiry workflow has high value because response delay is visible and repeated. It has medium readiness because the ownership rules and company data are inconsistent. Its risk is medium: a poor route wastes time and may produce an awkward customer handoff, but a review queue can limit the consequence.
A weekly internal summary has medium value, high readiness, and low risk. The data is accessible and a person already reviews it. The team may choose the summary for a quick learning project, or choose enquiry handling after a short rule and data cleanup. The scorecard makes that tradeoff explicit.
Your turn: score the shortlist
Complete Module 3 of the workbook for the top two candidates. Use a one-to-five score for each factor and add one sentence of evidence. Then select one of four decisions: proceed to design, investigate a named gap, repair the process first, or stop.
Your selected workflow and scored rationale become Section 3 of the Automation Decision Pack.
Quick Check
1. How should value, readiness, and risk appear on the scorecard?
2. A workflow has high value but unreliable input data. What is the best next decision?
3. What belongs in a practical risk statement?
What you have produced
Your selected workflow and evidence-backed scorecard are Section 3 of the Decision Pack. Module 4 decides which steps need AI, ordinary automation, human judgment, or process repair.