Data Collection Proxy Workflows for Deleted Source Pages
A data collection proxy workflow needs a clear response when a source page disappears. The event can affect provenance, monitoring conclusions and approved retention, but a failed request alone does not prove deliberate deletion. When using Go2Proxy data collection options, preserve the observed response and route the appropriate follow-up through the project's source policy.
Keep historical observation and current availability separate. A page can have been validly observed earlier while being unavailable now.

Classify the current source result
Distinguish a confirmed HTTP removal response, a redirect, an access condition and a collection failure. A timeout does not establish the same event as an observed removed page.
Validate the intended URL and final destination. A migration can move content without deleting the underlying source item.
Keep the observation time and non-secret diagnostic context. The record should explain what was actually seen without retaining sensitive access tokens in the URL.
Preserve provenance history
Retain the source reference and version relationship needed by the approved dataset process. Do not overwrite earlier evidence with a current unavailable label that erases when valid content was collected.
Keep current status as a new observation. This allows a report to distinguish historical content from a claim that the source is still publicly available.
Use stable source identifiers where available. A changed URL should not automatically create a second independent source or destroy the relationship to earlier records.
Apply the approved retention decision
Source disappearance does not let the collection worker decide every retention question automatically. Follow the project's reviewed source-use, privacy and removal processes.
Record the owner and decision for affected raw files and derived records. A dataset can have different systems holding copies, so remediation needs an actual dependency map.
Keep pending review explicit. Do not continue distributing affected material merely because the status field has not yet been updated.
Update downstream interpretations
Monitoring reports should not present an old value as freshly observed after the source becomes unavailable. If a value is carried forward, label that method and its age.
Preserve unavailable status separately from a verified business event. A removed product page may suggest lifecycle change, but the source response alone may not establish the reason.
Where a permitted replacement source exists, review its identity and allowed use before linking it to the earlier record. Similar text does not automatically establish the same authoritative source.
A moved report example
Imagine a public report URL stops returning the document and redirects to a new archive location. A collector marks all associated records deleted and removes their history.
Instead, verify the approved destination and document identity, preserve the migration relationship and record current availability separately. If the source is genuinely removed, apply the appropriate retention review without rewriting past observations as though they never existed.
This keeps the dataset's history coherent while allowing current source status and approved use to change. The guide to website change-detection workflows explains how stable identity, preserved evidence and confirmation rules make a reported change reviewable.
Deleted source questions
Does a timeout prove deletion?
No. Preserve the actual failure category and use suitable approved evidence before classifying a source removal.
Should the last known value appear as current data?
Only under an explicit carried-forward method with its age visible. It is not a new observation.
Can one worker delete every downstream copy automatically?
Only through a specifically approved workflow with known targets. Retention and remediation require the responsible policy and dependency mapping.
Keep history and current status distinct
Record removal evidence, provenance and the approved follow-up decision. That lets a proxy-supported dataset remain understandable when its sources change or disappear.



