Credit consumption, time & efficiency of expanding into new locations
| Location | Type | Status | Msg Cr. | Int Cr. | Total Cr. | Quality | LLM Calls | Disc. | Unique | Verif. | Src | Actions |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No mapping runs yet. Create one to start tracking. | ||||||||||||
Credit tracking method: Before/after snapshots of cumulative workspace credit usage (message + integration credits separately). When a run starts, the cumulative "used this period" totals are captured; when it ends, they're captured again. Consumed = after − before (same billing period). This is a measured difference, not an LLM-call estimate.
Automatic capture: Uses the Base44 Monitoring API (credit_pool.messages.used / credit_pool.integration.used) if WORKSPACE_MONITORING_API_KEY and WORKSPACE_ID secrets are set. Requires a workspace API key with "Read monitoring data" permission (Enterprise/Business plan). If not configured, enter before/after totals manually from the billing dashboard (Settings → Credit usage).
Message vs Integration credits: shown separately. The mapping pipeline consumes only integration credits (via InvokeLLM in discovery, directory collection, and enrichment). Message credits are consumed by builder-chat AI prompts, not by pipeline execution — but both are captured in the snapshot so you can see the full workspace picture.
Completeness labels: measured = both snapshots present, same billing period, no unrelated usage flagged — the difference is actual consumption. partial = period reset occurred or unrelated usage was flagged. none = snapshots not yet captured.
Period reset handling: if the billing period resets between before and after (detected via period_end mismatch), the cumulative counter resets to zero and the automatic difference is invalid. The run is flagged and manual entry is required.
Unrelated usage: if you used Base44 for other tasks during the run's measurement window, flag it and enter the estimated unrelated credits to subtract from the raw difference. The adjusted result is labeled "partial."
Snapshot preservation: before/after snapshots are immutable once recorded — recompute never overwrites them. Only the consumed-credit calculations are recomputed (e.g. when you flag unrelated usage).
LLM call count: tracked per stage as a separate efficiency metric (not used for credit estimation). Not all LLM calls consume the same number of credits.
No dollar costs: this dashboard tracks credit consumption only. Monetary cost calculations are deferred to you.