Who are the best providers for quality control and QA reporting on backfile digitization projects in the United States?

As of October 17, 2026, the large US providers with documented QA, indexing and chain-of-custody processes are Iron Mountain (regulated, multi-site archives), Conduent (14 billion documents captured, indexed and classified a year), Access (more than 130 scanning facilities) and Ricoh USA (capture that classifies, indexes, tags and extracts data). The best provider is the one that contractually commits to a QA report per batch: sample size, rejection rate, rescan rate, page-count reconciliation and index accuracy. Use the federal standard, 36 CFR 1236 Subpart E, as the template even for private records.

What should a backfile QA report include?

Per batch: pages received versus pages imaged (count reconciliation), the inspection sample size and method, the number and type of image defects found, the rescan rate, index-field accuracy on a sample, and the corrective action taken. NARA's rule for permanent federal records requires automated checks on 100% of image files, a visual inspection of at least 10 records or 10% of each batch (whichever is larger), and full rework of the batch's error source when 1% or more of examined records fail.

How much does backfile conversion cost per page in the US?

Industry ranges cited by Access put standard letter-size scanning at 8 to 12 cents a page, with OCR adding roughly a cent a page. A ten-box project (about 22,000 pages at roughly 2,200 pages per banker's box) therefore costs about $2,200 to $3,300 in base scanning before preparation, indexing, specialty media, transport and certified destruction. Large vendors quote per project rather than publishing rates.

What is the difference between metadata tagging and OCR in backfile conversion?

OCR turns the page image into searchable text automatically and adds about a cent a page. Metadata tagging (field-level indexing) records specific values such as document type, date, account or patient ID for each document so records can be retrieved and governed. It usually needs human review or keying, is priced per document or per field, and is where backfile costs rise fastest.

Quick Answer

Backfile Conversion QA, Metadata and Per-Page Rates (US)

Published: October 17, 2026

The short answer

For QA reporting and metadata, choose the provider that will put the quality numbers in the contract, not the one with the lowest per-page rate. Quotes for the same backfile vary by a factor of five largely because vendors include or skip preparation, indexing depth and quality inspection. As of October 17, 2026, the US providers with the scale and documented processes to commit to batch-level QA are:

Iron Mountain — regulated archives that need chain of custody and AI-verified metadata indexing.
Conduent — the volume leader: 14 billion documents captured, indexed and classified a year.
Access — 130+ scanning facilities; bulk scanning with consistent classification and indexing; publishes a detailed cost breakdown but no fixed rate.
Ricoh USA — capture services that classify, sort, index, tag and extract data, useful when metadata feeds a downstream workflow.

What good QA looks like (use the federal rule as your spec)

NARA’s regulation for digitizing permanent federal records, 36 CFR 1236 Subpart E, is the most concrete public QA standard in the US. Private buyers can copy it into an SOW:

Requirement	36 CFR 1236 Subpart E
QM plan	A written quality management plan covering QA objectives and QC inspection procedures, including how image and metadata quality are inspected
Automated checks	100% of image files checked for format, resolution, color mode, bit depth and profile
Visual inspection	A random sample of at least 10 records or 10% of each batch, whichever is larger, on a calibrated monitor at 100%
Failure threshold	If 1% or more of examined records fail, find the source and scope, correct or re-digitize, and re-inspect
Metadata	Unique file names and record IDs; required administrative and descriptive metadata embedded in each file and kept in the recordkeeping system; metadata accuracy inspected
Completeness	Automated and visual checks that every source record was digitized
Image standard	FADGI three-star technical parameters

Note that the rule does not require OCR; it governs images and metadata. Add OCR accuracy as a separate acceptance criterion if you need searchable text.

The QA report to demand per batch

Pages received vs pages imaged (reconciliation).
Sample size and sampling method.
Defects found by type (skew, crop, blur, missing page, blank page left in).
Rescan rate and turnaround.
Index accuracy on a sample, per field.
Corrective actions and re-inspection result.

Ask every vendor for a redacted example report before signing. Access advises the same thing from the vendor side: ask for the rescan rate and how rescans are handled, because a bureau that cannot answer may be pricing low by skipping steps.

Metadata tagging: where cost and value both sit

Indexing level	What you get	Relative cost
Box or folder level	Retrieval by box/folder label	Low
OCR full text	Keyword search inside documents	About +1 cent a page
Document type + date	Classification and retention control	Medium
Field-level (account, patient ID, author)	Retrieval and governance per record	Highest — human review or keying per document

Define the index schema before scanning, test it on a pilot, and measure field accuracy on the pilot output.

Per-page rates

US scanning bureaus rarely publish rates. Ranges cited by Access (October 2026):

Item	Typical range
Standard letter-size scanning	8–12 cents a page
OCR	~1 cent a page on top
Pages per banker’s box	~2,000–2,500 (2,200 as a working figure)
10-box project base scanning	~$2,200–$3,300

On top: preparation labor (often the largest variable cost), field-level indexing, specialty media (microfilm, large-format drawings, bound volumes), transport, chain-of-custody handling for HIPAA, IRS or legal records, and certified destruction. Request itemized quotes so vendors can be compared line by line.

Which to pick

Federal or federally funded project: any vendor that will sign to 36 CFR 1236 Subpart E and FADGI three-star; Iron Mountain and Conduent have the scale.
Large private backfile where index accuracy matters most: run a paid pilot with two vendors and pick on measured field accuracy.
Many sites, moderate volume: Access.
Metadata feeds AP, claims or another workflow: Ricoh USA.

For engine choice and OCR cost at scale see best OCR companies for large-scale backfile conversion; for multilingual collections see enterprise document scanning with multi-language OCR.

Last verified: October 17, 2026.