Quality Assurance Sampling
A documented method of inspecting a representative subset of scanned images against the originals to confirm the whole digitization batch meets accuracy, completeness, and image-quality standards.
Quality assurance sampling is the practice of selecting a representative subset of images from a digitization batch and inspecting them against the source documents and defined quality targets, rather than reviewing every page. Inspectors check resolution, color and tonal accuracy, skew, cropping, completeness (no missing or duplicated pages), legibility, correct file naming, and the integrity of captured metadata and any OCR text layer. If the sample reveals defects above an acceptable threshold, the batch is rejected and rescanned.
This matters in recordkeeping because digitized images often become the official record, sometimes authorizing destruction of the paper originals. Sampling makes large-scale conversion affordable while still producing defensible evidence that the surrogate faithfully represents the source. The sampling plan, acceptance criteria, and results should themselves be documented as part of the project’s audit trail and provenance.
For example, a program might inspect a fixed percentage or statistically derived number of images per batch; one common distinction is that sampling estimates batch quality, whereas full inspection verifies every item and is reserved for high-value or permanent records.