Picture the end of this quarter. Two pieces of work your team does every week now arrive prepared, to the standard of your best person, and the people who run them are asking what comes next. A third is ready to start, because the one question holding it back has been answered. That is the team you have, doing what looked impossible when the quarter began: 10X what it can do, starting with the work that was ready to go first.
The way there is a good first choice. Most leadership teams already hold the list: fifteen or twenty ideas from every department, each with a sponsor. The usual next step is to ask what each one is worth, and the list stalls there, because the value of a workflow shows once it runs. Rank first, on what you can already see. Measure value on the winners.
Score what you can see today.
Five things about any piece of work are visible before anything is built: how often it happens, how much friction it carries, whether its inputs are ready, what a miss would cost and how much there is to set up. Every one is scored so that a higher number means a better first win.
The people who do the work know the answers, so score with them in the room. Build your AI plan with the people who do the work gives five questions that fill the list in the first place.
One rule makes the ranking trustworthy: unknown is a value. When nobody in the room knows whether the records can be exported, the score is “Unknown”. It is never a zero, which buries a strong opportunity, and never a hopeful 2, which hides the question. An unknown becomes a task with a name and a date beside it, and it is often the most valuable line on the sheet.
The opportunity scorecard.
Copy it and give every opportunity one row.
Opportunity scorecard: [team or outcome]
- Frequency. 3: daily or weekly. 2: monthly, or several times a month. 1: quarterly or less.
- Friction. 3: the work waits at a handoff, is re-keyed or is regularly redone. 2: slow but steady. 1: runs smoothly today.
- Input readiness. 3: the records exist in one place and the team may use them in its approved AI workspace. 2: they exist but are scattered or need tidying. 1: they live in people’s heads, or permission is refused.
- Consequence of error. 3: a named person reviews the output before anything leaves the team. 2: a reviewer exists, and the output goes to a customer or carries commercial terms. 1: a miss would reach a customer, a contract or money before anyone saw it.
- Implementation effort. 3: one role, working from files and exports. 2: two or three roles, or one live connection. 1: several teams and several systems.
- Total. Add the known scores, out of 15. For each Unknown, write a range: the known total plus 1 at the low end, plus 3 at the high end.
- Find out. For each Unknown, write the question, who answers it and by when.
Rank by total. Where an opportunity’s range reaches the top of the list, answer its question before you commit: one answer could change the order.
One company’s ranked list.
An invented company scored five opportunities in one ninety-minute session.
Illustrative example with invented data. An invented 90-person industrial equipment service company. Scored by the operations lead, the controller, the sales manager and the service manager.
| Opportunity | Frequency | Friction | Inputs | Error | Effort | Total |
|---|---|---|---|---|---|---|
| Weekly receivables summary | 3 | 2 | 3 | 3 | 3 | 14 |
| New-customer handoff, sales to service | 3 | 3 | Unknown | 3 | 2 | 12 to 14 |
| Onboarding plan for new technicians | 1 | 2 | 3 | 3 | 3 | 12 |
| Sales proposal preparation | 2 | 3 | 2 | 2 | 2 | 11 |
| Supplier contract renewal review | 1 | 2 | 2 | Unknown | Unknown | 7 to 11 |
What the team decided
- Start now
- The weekly receivables summary. Every score is known, the controller reads every summary before it reaches the owners, and it runs from two exports the finance team already produces.
- Find out first
- The new-customer handoff. Nobody knew whether signed service agreements can be exported from the sales system into the approved AI workspace. The sales manager asks IT and reports back by Friday. A 3 ties it for first place and it starts alongside receivables; a 2 means the agreements are gathered in one place first.
- Next in line
- The technician onboarding plan, then proposal preparation once the approved company material is gathered in one place.
- Hold
- Supplier renewals. Two unknowns: who reviews a renewal before it is signed, and how many systems hold the contracts. The operations lead answers both before the next quarterly scoring.
The scorecard ranks the work. The leadership team chooses, and a named person reviews every output of whatever is chosen.
Four of the five have a starting point ready: weekly receivables summary, account handoff, employee onboarding plan and sales proposal preparation are capability templates you adapt to your own rules.
What the score leaves to you.
The scorecard orders a list. It leaves three things with you. Strategy can overrule arithmetic: if the handoff is the outcome your customers feel, it may deserve the first slot at 12. The score says an opportunity is ready to pilot, and the pilot proves the workflow on real cases. And value comes next: once you have chosen, What is an AI workflow worth? shows how to measure it from your own baseline.
Rescore every quarter. Unknowns get answered, records get gathered and a 9 becomes a 13. The first 90 days of an AI adoption plan takes your top-ranked opportunity from selection through baseline, pilot and rollout. All of it runs on the platform your team already has: ChatGPT, Claude, Copilot or Gemini. You set the pace, and your organization keeps control of how the work gets done.
Bring your list of opportunities, scored or not, and the one unknown you most want answered.