Unknown stays unknown
A rule the parser can't read is flagged for manual review, never reported clean.
Before you interpret a result, know whether another team's flag was live on the same users. FlagConflict compares targeting rules across both platforms and shows you the intersections.
| LaunchDarkly | Optimizely | Shared dimensions | Classification |
|---|---|---|---|
| checkout-express-v2 | one-click-cart-exp | geo, device, plan | Confirmed |
| nav-redesign | mobile-upsell-exp | plan | Possible |
| checkout-express-v2 | mobile-upsell-exp | plan | Possible |
| nav-redesign | one-click-cart-exp | — | No overlap |
Targeting rules from two vendors were normalized into one representation, then compared condition by condition.
Nothing was executed, nothing was written back, and nothing was stored.
We flag overlaps. We don't tell you your experiment is broken.
Confirmed, possible, none detected, non-comparable, insufficient information, unsupported rule.
A rule the parser can't read is flagged for manual review, never reported clean.
We surface the intersection so your team can judge interaction effects. We don't declare contamination.
A dated inventory of intersections you can attach to a release review. See the interpretation guide.
Start with the diagnostic free. Register when you need to save your workspace and track changes over time.
Run a cross-platform collision check with no account or payment required.
Create a free account to preserve one project, reopen past work, and track changes.
For teams using FlagConflict as part of an ongoing experimentation workflow.