Statsig

Statsig brings together feature flagging, A/B experimentation, and product analytics into a single platform, built on the premise that these three capabilities are more powerful when they share a data model. Feature flags that gate experiments automatically generate the exposure events needed for statistical analysis; analytics events that measure product metrics can be used as experiment outcome metrics without a separate instrumentation step. The founders came from Facebook's experimentation infrastructure team, and that lineage shows in the platform's emphasis on statistical rigor — sequential testing, CUPED variance reduction, and automatic detection of novelty effects are built in rather than bolt-on features.

The SDK evaluation path — where client or server code fetches feature flag values and logs exposure events — is latency-sensitive for any application that uses Statsig to gate features on the critical rendering path. Statsig maintains a caching layer that allows SDKs to evaluate flags locally using a downloaded rule set rather than making a network call for every evaluation, which dramatically reduces per-evaluation latency and provides some resilience against backend degradation. However, that cache must be refreshed periodically, and a prolonged backend outage eventually causes the cached rules to become stale relative to any flag updates made in the console during the outage window.

When Statsig has infrastructure problems, the SDK evaluation cache provides a meaningful buffer — applications continue to evaluate flags using the last-known rules for minutes to hours, depending on the cache TTL configured. The first visible impact is typically in the console: the real-time analytics feed stops updating, and experiment results stop being refreshed because event ingestion is delayed. New flag changes made in the console may fail to propagate to SDKs when the rules-publishing pipeline is impaired. Server-side SDKs without persistent local cache are more immediately affected than client-side SDKs running in browsers with cached rule sets.

Outage.gg monitors Statsig service status using community reports from product and engineering teams. If flag evaluation is broken, experiment data is not updating, or the console is inaccessible, the live status page shows current incident data.