AI ROI Worksheet: A Template to Decide If a Project Pays Off

An AI ROI worksheet helps you decide if a project pays off by comparing its value against its full cost. The formula is simple: (value created minus total cost) divided by total cost. The work is gathering honest inputs. This template walks through each field you need. The easiest way to fool yourself with a worksheet is to inflate the value side, counting time saved that you will never actually redeploy or ignoring the learning curve and the hours spent writing prompts and cleaning up the process. A worksheet built to justify a decision you have already made tells you nothing; an honest, even pessimistic one gives you a number you can trust. Where a benefit is quality rather than time, use a countable proxy such as error rate or rework hours instead of guessing at a dollar figure. This template covers the inputs to gather, the step-by-step calculation including a payback period, and how to read the result before you commit or scale.

Inputs to gather first

• Baseline: current time, output, and error rate for the task.
• Tool cost: subscriptions, per user per month.
• Setup cost: one-time hours at a loaded rate.
• Training cost: hours for the team to learn it.
• Expected change: realistic time or quality improvement.

The worksheet, step by step

1. 1

   Total the cost

   Add tool, setup, and training costs, then add a 20% buffer..

2. 2

   Estimate the value

   Convert time saved, capacity, or quality gains into a dollar figure where you can..

3. 3

   Apply the formula

   Subtract cost from value, then divide by cost for an ROI ratio..

4. 4

   Find the payback period

   Divide setup cost by monthly value to see how fast it returns..

5. 5

   Decide

   Proceed, adjust, or pass based on the numbers..

Use research like the NBER study of support work to sanity-check your assumptions, not to set them. Your inputs should come from your own baseline. Our full guide on calculating AI ROI explains the value side in depth.

Be conservative with value

It is tempting to inflate the value of saved time. Only count time you will actually redeploy, and be honest about the learning curve. A worksheet that flatters the project helps no one.

How to read the result

A positive ratio and a payback period inside a quarter is a healthy first project. A marginal result is a signal to shrink the scope or pick a better task. The McKinsey State of AI survey reinforces that value follows disciplined choices.

Be conservative on the value side

The easiest way to fool yourself with an ROI worksheet is to inflate the value. Count only time you will genuinely redeploy to useful work, not hours that will quietly fill with busywork. Include the learning curve and the time spent writing prompts and cleaning up the process, since those are real costs that flatter the ratio if ignored. A worksheet that is honest, even pessimistic, gives you a number you can trust; one built to justify a decision you have already made tells you nothing.

Where a benefit is quality rather than time, use a countable proxy such as error rate or rework hours rather than guessing at a dollar figure. A measurable drop in mistakes is defensible value even when it is not pure time saved. Our guide on AI ROI beyond time saved covers how to convert those proxies into the value side of the worksheet.

Use the worksheet to decide, then to scale

The worksheet earns its keep by driving a clear decision: proceed, adjust, or pass. If the numbers say proceed, run the project, then revisit the worksheet with real results rather than estimates. If they say marginal, shrink the scope or pick a better-fit task before committing. And remember that a result which holds at small scale may dip a little as more people do the work, so plan for that and keep measuring rather than assuming the early ratio carries forward unchanged.