AI Workflow ROI Estimator

Full prompt

You are a financial analyst for SaaS operators. Your job is to build a conservative ROI model for an AI automation project.

I'm automating this workflow: {{WORKFLOW_NAME}}

Current state:
- {{CURRENT_TEAM_SIZE}} people spend {{HOURS_PER_WEEK}} hours/week on this task
- Average fully-loaded cost per person: {{COST_PER_PERSON_ANNUALLY}}
- Current error rate: {{CURRENT_ERROR_RATE_PERCENT}}% (each error costs {{COST_PER_ERROR}} to fix)
- Manual rework time per error: {{REWORK_HOURS_PER_ERROR}} hours

Proposed AI solution:
- Implementation cost (tools + setup + training): {{IMPLEMENTATION_COST}}
- Monthly subscription/API cost: {{MONTHLY_COST}}
- Expected error reduction: {{ERROR_REDUCTION_PERCENT}}%
- Expected time savings per task: {{TIME_SAVINGS_PERCENT}}%
- Adoption ramp: {{ADOPTION_RAMP_PERCENT}}% of team in month 1, reaching {{FULL_ADOPTION_PERCENT}}% by month {{ADOPTION_MONTH}}

Build a 12-month ROI model that shows:
1. Month-by-month labor cost savings (accounting for adoption ramp)
2. Month-by-month error reduction savings
3. Cumulative implementation + subscription costs
4. Payback period (when cumulative savings exceed total costs)
5. Net ROI at month 12
6. Sensitivity analysis: what if error reduction is only 50% of expected? What if adoption takes 6 months instead of 3?

Use conservative assumptions. Flag any assumptions that feel risky. Format as a table I can paste into a spreadsheet.

How to use

1. Gather your baseline: How many hours does this task consume weekly? What's your fully-loaded labor cost? How often does it fail? 2. Get vendor quotes: What's the implementation cost and monthly fee for the AI tool you're considering? 3. Estimate the lift: How much faster will the AI be? How much will it reduce errors? (Be conservative—assume 60–70% of vendor claims.) 4. Fill in the placeholders and run the prompt. 5. Share the output with your CFO or finance lead. The sensitivity section is your defense against "but what if it doesn't work as promised?"

Customization tips

• Adjust adoption ramp based on your team's change tolerance. Risk-averse teams may need 4–6 months to reach full adoption; fast-moving teams might hit 80% in month 2.
• Don't forget the hidden cost: someone has to monitor the AI's output for the first 2–3 months. Add 2–3 hours/week of QA time to the implementation phase.
• Error cost is often underestimated. Include not just direct rework time but customer churn, compliance risk, and reputation damage. A single bad onboarding can cost you a customer.
• Vendor benchmarks are optimistic. Use 60–70% of claimed time savings in your model. If the vendor says 80% error reduction, model 50–60%.
• Run the sensitivity analysis with three scenarios: conservative (50% of expected gains), base case (your best estimate), and optimistic (80% of vendor claims). Present all three to leadership.