If AI doesn't pay for itself in 3 months, don't buy it

Managing partners do not need a consultant's spreadsheet to evaluate an AI project. Two numbers and one division tell you everything.

The formula

• •Monthly saving = hours saved per month × loaded hourly cost of the person doing the work.
• •Payback period = one-time project cost ÷ monthly saving.

Loaded hourly cost means salary plus benefits and overhead, not just the wage. For most admin and clerk roles that lands between $30 and $50 an hour. For paralegals and specialists it is higher, which is exactly why automating their repetitive work pays back faster.

A worked example

Suppose an automation removes 60 hours of clerk and paralegal work a month at $40 an hour. Monthly saving: $2,400. If the pilot costs $5,000, payback is just over two months. Everything after that is margin, month after month, for as long as your firm handles files.

Our rule of thumb: a good automation project should pay for itself in under three months. If the math says twelve, the process is probably too small or too irregular to automate today, and we will tell you so in the audit rather than sell you the project.

Two things the formula deliberately ignores

First, error costs. Fixing a mistake usually costs far more than making the entry, so the real saving is higher than the formula shows. Second, growth. The formula assumes your volume stays flat. If you are growing, automation is the difference between scaling revenue and scaling headcount at the same rate. Both effects mean the simple formula understates the case, which is fine. If a project clears the bar on conservative math, it clears it in reality too.

If the conservative math works, the real result will be better. That is the right direction to be wrong in.

Bring us one process and the three inputs: hours, hourly cost, and volume. The free audit turns them into a payback number you can take to your partners. If it does not clear three months, we will be the ones telling you not to buy.