Full Reloads

Full reloads completely wipe the Staging and Results datasets Statsig has used for previous pulse calculations, and Statsig recalculates results from scratch.

Forwarded exposures and full reloads

A full reload recalculates experiment results from the staging and results datasets. It doesn't rebuild or backfill the forwarded exposures table.

A reload may also forward recent exposures before the daily export. These updates are best-effort and can be incomplete. Use the daily export as the authoritative record of eligible forwarded exposures. For details, go to Forwarded Data.

Choose a full reload when Statsig needs to recompute staging and analysis results, rather than an incremental reload that only appends new data or a metric reload that restates a single metric.

Generally, use full reloads for

• Initial or historical pulse loads.
• Cases where the customer side has lost or dropped data, so incremental reloads have lost state.
• Cases where data changes frequently. For example, a dbt full reload changes historical data due to chargebacks, model changes, or other reasons.
• Cases with complex data dependencies, where a team wants to ensure that the gap between Statsig and internal systems doesn't cause inconsistencies.