Why one process, one quarter
Spreading effort across five AI ideas at once produces five half-finished pilots. One process, done properly for a quarter, produces evidence your board can act on.
The best candidate is a task your team repeats every week that eats time and frustrates people. If you have used the AI Opportunity Finder, take your top scorer.
The one-page plan
Fill it in with your leadership team in about 45 minutes. It covers:
- the process, and the problem in one sentence
- a success measure you already count, such as hours per week, turnaround time, error rate or cost per case
- today's baseline and the target for week 13
- one named owner, the team, the tools and a budget ceiling
- what data the tool will see, where it lives and who signs it off
- the risks, how likely they are and who watches them
- the go or no-go decision date
Tip: write down today's number before you start. Without a baseline, you cannot prove the change.
The 13 weeks
| Phase | Weeks | What happens |
|---|---|---|
| Map and prepare | 1 to 4 | Map the process as it runs today, record the baseline, shortlist tools, agree the data rules, choose the pilot group. |
| Pilot | 5 to 9 | A small group uses it with a person checking every output. Fix the obvious problems, compare against the baseline, write a simple how-to. |
| Measure and decide | 10 to 13 | Pull together the results, work out the real cost and saving, present to leadership, make the call. |
Use the tracker in your weekly meeting: five minutes, tick what is done, note the blockers.
The decision: scale, change or stop
The end-of-quarter review asks whether you hit the target and stayed within budget, what worked, what did not, and what staff in the pilot said. Then you choose:
- Scale. It worked. Roll it out to more of the team.
- Change. Promising, but adjust and run another short pilot.
- Stop. It did not pay back. Keep the lessons and move to the next process.
Stopping is a good result if you learn cheaply. Use the AI ROI Worksheet to put numbers behind the decision.
Questions
Why 90 days?
Long enough to run a real pilot with a baseline and results, short enough to keep everyone focused and stop early if it is not working.
What makes a good success measure?
Something you already count, such as hours per week, turnaround time, error rate or cost per case.
Who should own the roadmap?
One named person who is accountable, with a small team doing the work and testing it.
What if the pilot fails?
Stopping is a valid result. You keep the lessons, spend less than you would have, and move to the next process.