# Why double counts are discounted

> How Eevo finds consequences that count the same effect twice, and what it does about them.

People list the same effect twice without noticing: "customers get correct totals" and "fewer support tickets about totals" are partly one thing. Counted twice, it makes an option look better than it is.

So for each stakeholder Eevo asks, for every pair that could double count, how much of one is already in the other. It looks for four kinds of overlap:

- **Restatement**: the same outcome in other words.
- **Containment**: one consequence is part of the other.
- **Causal**: one consequence causes the other.
- **Mirror**: an option avoids what another option causes ("B avoids X" next to "A causes X").

Eevo always takes overlap into account: the shared part is counted once, and only the overlapping share is left out. Mirrors are reported but count in full. They disappear when every option is described against one baseline, what would happen if none of them were taken, which is why the docs ask for that.

The overlap checks are where most of a large request's calls go: up to 120 pairs per stakeholder.