UK IT Tender Amendment & Deadline Watch

Keep a UK IT tender list current without counting every amendment as a separate opportunity. This Actor reads the official Find a Tender OCDS release API, filters IT services/software (CPV 72/48 by default), groups observed releases by OCID, and compares the latest observed notice with your previously delivered version.

Each candidate is checked against the official compiled record and its notice history before producing a result. It returns deadline changes, changed fields, explicit status changes, keyword evidence and source URLs. Sparse update fields are not treated as a complete snapshot. A cancellation of a previously tracked notice is retained even when it no longer passes the open-only filter. Filter matches are not a supplier eligibility or award-probability assessment.

Input example

{
  "cpvPrefixes": [
    "72",
    "48"
  ],
  "lookbackDays": 14,
  "maxPages": 3,
  "maxResults": 10,
  "openOnly": true,
  "changesOnly": true,
  "monitorKey": "example"
}

Coverage and interpretation

The default examines at most 3 API pages of 100 releases each in a 14-day update window. This is a bounded recent scan, not exhaustive UK tender coverage. Read pagesFetched, releasesScanned, recordLookups and coverageComplete. At most 10 official record packages (512 KiB each) are checked per run; this cap is reported explicitly. Known CPV matches and previously tracked notices are checked before notices with missing CPV metadata. Partial pages, the result limit, malformed releases and source errors never imply cancellation or deletion. No removal is inferred from absence. A changed deadline can be later, earlier or unknown; inspect both values.

The first successful stream creates a baseline of the matching notices it delivers. Later runs suppress unchanged OCIDs. Candidates come from the bounded scanned update window. Their official record package supplies compiled fields and notice/deadline history; PDFs, external procurement portals and contact-person details are excluded. A new amendment outside the configured window can be missed. Region filters only use the region codes actually supplied by the notice.

Source and licence: Find a Tender data/API documentation. Notice output contains the required attribution: “Contains public sector information licensed under the Open Government Licence v3.0.” OGL v3.0. Third-party attachments are excluded.

Run and retrieve results

1. Open the Input tab, keep the example or add your own filters, then select Start. No operator approval is required.
2. Open Output for the results table. Export JSON through the dataset, or download RESULTS.csv from the run key-value store for spreadsheet-safe CSV.
3. Read OUTPUT for the authoritative records, source coverage, billing checkpoints and monitor storage ID. DIAGNOSTICS.json reports failures and caps.

The Actor uses limited permissions and stores its own monitor state in a named key-value store in your Apify account. Keep the same monitor name and semantic filters to compare runs; changing page/result limits does not reset the baseline. Each account has its own state. Run a single stream at a time: overlapping runs on the same monitor can repeat observations, and a detected conflicting state update is not overwritten. State keeps up to 5,000 tender entries, with a 4 MiB overall monitor-state cap; very old evicted records may be observed again.

Pricing and spending limits

• Actor start: USD 0.01 for a run at the supported default 256 MB. Apify's start event is per GB with a one-event minimum, so larger memory settings can increase this charge.
• New or updated tender: USD 0.01. One latest matching OCID provided in OUTPUT, including an explicit update to a previously tracked notice. Unchanged, excluded and failed notices are not charged.
• No separate dataset-item event. No proxy, AI API or paid external API is used.
• Set Maximum cost per run to cover the start and the number of checks/results you want. Budget caps can stop delivery early and are shown in OUTPUT.
• The start fee applies even if no match is found or the upstream service fails. Failed or undelivered outcomes do not incur the custom result/check event.

Results are saved before the custom outcome charge. A new run performs a new paid check; retrieve the old run before restarting it. The same run's uncertain dataset or charge POST is never intentionally repeated. If a charge is uncertain, processing stops with a billing audit instead of attempting a second charge. OUTPUT remains authoritative when dataset delivery is uncertain.

Data and support

Input, output and monitor state stay in your Apify account's storage under its retention policy. This Actor sends no messages, modifies no source data, and accepts no credentials or private source URLs. Source requests use an identifying User-Agent with the developer's public support profile.

Report reproducible defects in this Actor's Issues tab using a minimal invented example and the diagnostic code. Do not post API tokens or private run links.

Actual owner-validation output

Observed on 26 October 2026 (Japan time). This is a saved owner test, not a customer result or evidence of sales. Source data changes over time.

The saved run provided 3 tender records. The 300-release scan reached its page and compiled-record lookup caps; coverage was partial. Its configured event total for an ordinary buyer at 256 MB would be USD 0.040, including startup. Owner event counts are validation, not paid customer revenue.

Abbreviated actual record:

{
  "noticeId": "100725-2026",
  "title": "Digital Analytics Tool",
  "deadlineUtc": "2026-11-20T14:00:00.000Z",
  "compiledSource": true,
  "historyNoticeCount": 1,
  "sourceCoverageComplete": false,
  "sourceUrl": "https://www.find-tender.service.gov.uk/Notice/100725-2026"
}