Methodology
How we measure the return on your AI
AIReturn turns the work data you already have + your AI costs into one number — the return on your AI. No new logging, no surveys. Here's exactly how.
The three inputs
Output
what a team produces, defined per function: engineering ships changes; support resolves tickets; sales advances and wins deals; product ships specs and features.
Rework
the friction and redos it took to reach the outcome. The hidden waste, read from your tools:
| Team | What we read as rework / friction |
|---|---|
| Engineering | code rewritten right after merge, PR review rounds, reverts, reopened issues |
| Support | reopened tickets, back-and-forth replies, reassignments, escalations |
| Sales | deals moving backward a stage, pushed close dates, time stuck in a stage |
| Product | specs/issues reopened, review loops, mid-sprint scope churn |
Cost
your AI spend (tokens today), attributed to each team through your connected tools.
The return
Return on AI = the value your teams deliver vs. what the AI costs — with the rework (the waste) taken out. Shown as a multiple (e.g., 3.1×) and in dollars.
Two principles that keep it honest
Compared only to your own history.
Each team is measured against its own baseline over time, never ranked against another team — output means different things in support vs. engineering.
Measured by team, not used to score individuals.
Team and workflow level only. No employee scoreboards.
From diagnosis to fix
We don't just score you. We pinpoint why the rework is happening — usually the AI is missing context — and give the next move, ranked by impact, then track whether it improved.
See it on your own numbers.
Get a clear return-on-AI picture in weeks.