The ROI Problem
Every vendor has a slide showing 10x ROI. Every consultant has a case study showing 300% returns. And every board member is asking: "What's our return on this AI investment?"
The honest answer is usually: "We don't know yet, and here's why that's okay."
Why Traditional ROI Doesn't Work for AI
Traditional ROI is simple: investment goes in, returns come out, divide one by the other. This works for capital equipment, marketing campaigns, and new product launches.
AI is different for three reasons:
1. Benefits compound over time. A customer service AI gets better as it handles more conversations. The ROI in month 1 is different from month 12.
2. Value is often in time saved, not revenue generated. Saving a senior analyst 10 hours a week doesn't show up in revenue figures, but it's enormously valuable. How do you value time that gets redirected to higher-value work?
3. The baseline is shifting. Your competitors are also adopting AI. The real question isn't "what does AI add?" but "what do we lose by not having it?"
A Practical Framework
Instead of forcing AI into traditional ROI calculations, use this four-dimension framework:
Dimension 1: Direct Cost Savings
The easiest to measure. What processes cost X before AI and Y after?
Examples:
- Document review: 40 hours/week at $150/hour → 15 hours/week = $3,750/week saved
- Customer service: Average handling time reduced from 12 minutes to 8 minutes = 33% capacity increase
- Report generation: Manual process costing $2,000/report → AI-assisted at $400/report
Be conservative. Measure actual, not projected. Track for at least 3 months before claiming savings.
Dimension 2: Quality Improvements
Harder to measure but often more valuable than cost savings.
Examples:
- Error rate in data entry: reduced from 4% to 0.8%
- Compliance violations flagged before they become problems: 12 per quarter
- Customer satisfaction scores: improved from 72% to 84%
Define quality metrics before implementing AI. Measure the baseline. Then measure again after 90 days.
Dimension 3: Speed and Capacity
What can you do now that you couldn't before?
Examples:
- Proposal turnaround: 2 weeks → 3 days
- Research capacity: Can now analyse 10x more sources per project
- Response time: Customer enquiry to first response reduced from 4 hours to 15 minutes
Speed improvements often unlock revenue that doesn't show in simple ROI calculations.
Dimension 4: Strategic Optionality
The hardest to quantify but potentially the most important. AI capabilities create options you didn't have before.
Examples:
- Can now offer personalised recommendations at scale → new service offering
- Can now serve smaller clients profitably → expanded market
- Can now operate 24/7 without additional staff → new geographies
Don't try to put a dollar figure on optionality. Instead, document the options created and let leadership assess their strategic value.
The Honest Board Report
Here's a template that works:
AI Investment Summary
- Total investment this quarter: $X
- Direct cost savings verified: $Y (measured, not projected)
- Quality improvement: [specific metrics with before/after]
- Speed gains: [specific metrics with before/after]
- New capabilities unlocked: [list with strategic context]
- Confidence level: [high/medium/low for each metric]
- Recommendation: [continue/expand/pause/pivot]
The key: separate what you know from what you're projecting. Boards respect honesty more than inflated numbers.
Getting Your Financial Readiness Score
Financial readiness for AI includes more than budget. It covers ROI frameworks, cost modelling, and the ability to sustain investment through the learning curve.
Take the free Quick Scan to see your financial readiness score alongside seven other dimensions of AI readiness.