Data quality and governance
Yield
Also called: Extraction yield, success rate
The share of fetched pages or requests that produce valid records.
In practice
A falling yield shows blocks, layout changes or scope problems before users notice them.
Define a valid result
Extraction yield measures how much collection work produces usable output. A request returning a page is not necessarily a valid record: required fields may be absent, values may be malformed or the result may represent the wrong item. Define acceptance criteria before calculating yield and publish the denominator so the number has an unambiguous meaning.
Example: requests versus targets
If 900 of 1,000 attempted targets produce accepted records, target-level yield is 90%. If retries made the system issue 1,200 requests, the accepted-records-per-request measure would be 75%, assuming one record per successful target. These are different metrics. Label them clearly and do not compare one provider’s target-level figure against another provider’s request-level figure.
Investigate the reason, not just the percentage
Break failures into access errors, unavailable pages, extraction errors and validation failures. Track by source and page type so an easy, high-volume source does not hide poor results elsewhere. Record exclusions and valid empty results explicitly. Improving yield should mean delivering more accepted data, not weakening validation or quietly dropping difficult targets from the denominator.
In Import.io managed engagements, every delivery is checked against a written schema contract and the source page before it ships.
Managed Services →