METRIS™
How METRIS™ measures
Every number METRIS™ shows tells you where it came from. A rough guess labeled as a rough guess is useful. A rough guess dressed up as proof is not.
Four kinds of evidence
| Label | What it means | Example |
|---|---|---|
| Reported | What you told us | "AI saves me about an hour a week." |
| Calculated | Our math on your answers, with every assumption shown | "At $25 to $50 an hour, that time may be worth $20 to $200 a month." |
| Documented | Backed by a record | A card statement showing the tool charge, or three real AI-written posts. |
| Measured over time | Timed or counted, before and after | Five quotes timed with AI and five without, two weeks apart. |
How the numbers are worked out:
- Pays for itself at: monthly tool cost ÷ what an hour of your time is worth = hours a month the tool has to save.
- Cost per approved task: (your share of the tool's cost + time asking AI + time checking + time fixing, valued at your hourly rate) ÷ number of tasks that were accepted.
- First-pass acceptance: tasks accepted without changes ÷ tasks checked. If a check never changes anything, it may be a habit, not a check.
What your answers alone can and cannot show: they can point to likely gaps and rough value. They cannot prove savings, profit or cause.
Limits: even a measured improvement does not by itself prove AI caused it. A busy season, a new hire or a price change can move the same numbers, so we say what else changed.