Bulk work, such as imports, bulk skip trace submissions, scheduled exports and SiftMap auto-add snapshots, runs as activities: background jobs processed through a shared queue. Activities are built for bulk volumes, not for editing a few records.
| Guideline | Value |
|---|---|
| Minimum records per activity | 20 |
| Chunk size for large volumes | 1,000 records per activity |
| Activity creation rate | 2-3 per minute per account, any job type |
| Parallel submission | Doesn't make jobs finish faster (see below) |
- Don't create an activity to change a handful of records. A job with fewer than about 20 records doesn't get resources allocated properly, and doing it repeatedly is treated as abuse.
- Use the per-record endpoints for small changes. To change a few records (add a tag, set a status, update a custom field), call the record endpoints directly, for example
PATCH /api/internal/property/{uuid}/,POST .../add-tags/orPOST .../add-lists/. - Respect the creation rate. Create no more than 2-3 activities per minute per account, whatever the job type.
- Chunk large volumes. Split large record sets into activities of about 1,000 records instead of one huge job.
- Parallel submission doesn't help. Activities run through an internal queue, so total throughput is set by the queue, not by how many jobs you submit. Submitting faster than 2-3 per minute gives no speed benefit and may be flagged as abuse.
Watch activities in the account activity feed instead of polling records:
GET /api/internal/activity/lists recent activities and their status.
| Records to change | Use |
|---|---|
| Fewer than 20 | Per-record endpoints, paced within the rate limits |
| 20 to 1,000 | One activity |
| More than 1,000 | Activities of about 1,000 records, no more than 2-3 per minute |