procurement_coverage_cells: 58
This data as json
| rowid | source_key | source_label | record_kind | record_kind_label | fy_key | state | n_records | prev_n_records | why_code | why_text | evidence_kind | evidence_sql | evidence_citation | evidence_lf_id | probed_for_publication | is_crawl_dependent | drill_url | computed_at | run_id |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 58 | gvibuy_po_detail | GVIBuy purchase-order detail pages | line-item | Purchase-order line items and their commodity classes | not-applicable | PARTIAL | 22957 | 22957 | detail-layer-held-against-a-bounded-list | Line items are published by GVIBuy on each document's own detail page, so this was a fetch cost and not a publication gap - which is why probed_for_publication is 1 and the state is about us. The fetch is not finished - 4 documents with a requestable number wait on the detail crawl, see the header-detail row - and so far we hold 22,957 line items across 4,391 of the 4,393 purchase-order documents, and 22,952 of them carry a NIGP commodity class. It stays PARTIAL because it inherits the boundary of the document list it was fetched from, which is one walk on one day - see the header-detail row of this matrix for what that costs. Nothing here is a count of what was bought: an item is a line on a document, and a document may be a contract ceiling rather than a purchase. | MEASURED | SELECT COUNT(*) FROM gvibuy_po_items | 1 | 1 | /procurement/chain?source=gvibuy_po_detail&kind=line-item&fy=not-applicable | 2026-11-21T22:40:46Z | 2026-11-21T22:36:15Z |