Evaluate value and plan adoption

Outcome and preparation

Day 2 · 15:15–16:15 · 60 minutes. Produce a 90-day plan for three measurable pilots and a one-page operating policy. Bring your baseline, rehearsal log, business plan and templates/pilots.csv. Use 10 minutes to explain measurement, 40 to plan and 10 to review.

Measure value at comparable quality. Report generation time alone excludes checking, corrections and maintenance. Released staff capacity is not automatically a reduction in payroll or an increase in revenue.

Step by step

1. Step 1

   Compare manual preparation plus review time with assisted preparation plus review time on the same reporting task. Label actual measurements and estimates. Record reviewer corrections and test failures alongside time.

2. Step 2

   Calculate a monthly scenario. Use the formula below with your own inputs and verify the arithmetic independently. Keep unknown usage cost visible.

   Prompt / exampleCopy

   Capacity value/month = (manual hours/run - assisted hours/run including review)
                          × runs/month × loaded hourly cost.
   Net estimated value/month = capacity value - incremental licence/usage - maintenance.
   Simple payback months = one-time setup / positive monthly net value.
   If net value is zero or negative, report no positive payback under these assumptions.
   Do not count the same subscription twice or label capacity value as cash savings.
   

3. Step 3

   Reproduce the teaching example: 2 manual hours, 0.5 assisted hours, 4 runs/month and USD 80/hour gives USD 480 capacity value. Subtract 100 incremental usage/licence and 80 maintenance: USD 300 net estimated value. Setup of 600 implies 2 months simple payback if assumptions hold. Model conservative and upside cases; check the answer key only afterward.

4. Step 4

   Fill the three rows in templates/pilots.csv: executive reporting, system-health evidence assistant and business-plan evidence review. For each name business/technical owners, start date, day-30/60/90 milestones, baseline, quality target, measurement method, cost cap and stop criterion. Ask Claude to critique the gaps.

   Prompt / exampleCopy

   Review my three-pilot plan for the next 90 days.
   Each pilot needs accountable business and technical owners, approved sources, a baseline,
   quality and time targets, measurement frequency, cost cap, decision gates and a stop criterion.
   Day 30: controlled tests; day 60: supervised pilot; day 90: keep, revise or stop decision.
   Flag missing commitments rather than inventing owners or approvals.
   [Paste pilot table]
   

5. Step 5

   Write handover/operating-policy.md: approved sources/tools, who approves changes, source-refresh owner, required review, escalation contact, failure notification/check routine, retention location and pause procedure. Specify that reports remain drafts until reviewed and that a source outage is visible.

6. Step 6

   Review the business plan and pilot plan together. State the immediate decision, evidence needed before scale-up and criteria for stopping. Save the ROI calculation, all three pilots and policy. Use a partner as sponsor to challenge one benefit and one operational assumption.

Expected output and verification

Three distinct pilots with owners and dates to confirm, measurable baselines, quality thresholds, cost ceilings and decision gates. A useful reporting pilot target might be all required metrics correct and traceable on a held-out test set, fewer preparation minutes and no external sending without approval. State the sample size; two good reports do not establish production reliability.

The operating policy must explain who notices a failed weekly run. “The AI will handle it” is not an accountable owner. Report app/connector limitations alongside the proposed benefits.

Troubleshooting

If the baseline was estimated, plan a week of measurement before approving ROI claims. If review effort exceeds time saved, improve the method or stop the pilot. If owners cannot commit, keep the work as a proposal. If a target is “100% accurate,” specify the test set, metric and review process that would demonstrate it.

Completion checklist

• Whole-task time and quality measured or clearly labelled as estimates.
• Scenario arithmetic checked without double-counted costs.
• Three pilots and a pause/escalation policy saved.