Two audiences defined on different platforms, and the set of users that satisfies both rule sets at once.
That shaded band is what every classification below describes: how certain we are that it exists, and how large it might be.
Every pair of rules lands in exactly one. The diagram shows the shape of the finding; the note tells you what to do.
Conditions on both sides deterministically overlap — matching equality on device, geo or plan.
Do: open the detail view and decide whether the two treatments can conflict for that cohort.
A probable intersection, but missing dimension values prevent confirmation.
Do: export fuller targeting definitions, or check attribute distributions, and re-run.
The evaluated rules carry mutually exclusive constraints, such as mobile against desktop.
Limit: not a guarantee. Rules outside the export, or audience syncs, can still overlap.
The rules target on dimensions that share no common attribute, so no comparison is meaningful.
Do: confirm the attribute names genuinely differ rather than being spelled inconsistently.
Well-formed but too thin to evaluate — clauses without values, or a reference resolving to nothing.
Do: re-export with rule bodies included and run again.
A custom clause, proprietary operator or nested expression the parser does not yet implement.
Limit: flagged for manual inspection. Never counted as clean.
We verify that two targeting rules can admit the same cohort. We do not claim statistical contamination without production telemetry, and we do not re-weight your results.