We are paying thousands of dollars for seat licenses for several new generative AI tools, but we are not seeing any change in our payroll or our bottom line. How do we measure the actual, hard ROI of these tools instead of just relying on soft stories of saved time?

Question

Tyler Smith · Accepted Answer

If your AI tools are only saving people ten minutes here and there, you are paying a technology tax, not building an AI-powered operation. To find the actual return on investment, you must stop measuring soft time and start tracking hard operational capacity. You do this by measuring the volume of output per seat.

Use Keith Cunningham's Thinking Time to calculate your baseline metrics. Sit down for forty-five minutes with a blank notepad and ask: How might we measure the exact output of our operations team so we can prove our AI tools are increasing capacity? Convert this question into two or three specific metrics for your weekly Scorecard.

If you implement an AI tool in your customer service department, do not just look at customer satisfaction. Track the number of tickets closed per representative per week, or the average cost to resolve a customer issue. If your team is truly more efficient, you should see one of two hard results: either your revenue increases without hiring more staff, or your payroll costs decrease as a percentage of revenue.

If your Scorecard numbers are not moving, you are experiencing what Gleb Tsipursky calls action bias, which is the urge to do something new just to feel productive. Your team may be using AI to generate more noise rather than more value. Demand that every software license is tied to a specific operational Rock. If the manager owning that Rock cannot point to a measurable increase in capacity or a decrease in direct costs after one quarter, cancel the subscription.