Skip to main content
Import a backlink export you already hold, then choose the rows worth checking. Imported candidates retain supplier provenance and dates. An import does not establish current presence, complete coverage or recurring monitoring.

​
Preserve and inspect the import

Use import_backlinks for a hosted import. Its reference defines the supported supplier values and row format. Inspect every rejected row returned by the operation and keep the run identifier. Read get_discovery_run, then page through list_discovery_candidates. For a local CSV instead:
The ledger guide describes local candidate files. A local import does not create a hosted discovery run.

​
Verify fifty selected candidates

Batch verification and explicit candidate enrollment require their matching hosted release and database migrations. Availability records the boundary. Where only single verification is available, use verify_discovery_candidate for each selected row.
  1. Select one to fifty stored candidates from the same run. Keep their actual identifiers and optional local references.
  2. Submit verify_discovery_candidates with runId, an items array of candidate identifiers, and an idempotencyKey.
  3. Save the returned batch identifier and each item’s outcome. Budget rejection can affect individual items; inspect the reservation and counts.
  4. Read get_candidate_verification_batch until the selected jobs finish. Read each observation separately: succeeded is job completion, while present, absent or unknown describes evidence.
After a lost response, replay the same payload and idempotency value. Changing the selection creates a different intended request and needs a new value.

​
Start monitoring deliberately

For a completed present candidate verification, call monitor_discovery_candidate. This preserves the candidate’s lineage and local reference. Check the returned watch; replay does not reactivate a watch someone later paused. Keep rejected or unknown rows with their reason for review. Follow monitoring for enrolled watches and troubleshooting for unavailable batches or expired evidence.