Scan-on-Demand vs Full Archive Digitization: How to Choose the Right Conversion Strategy

Asit Dubey

Every institution holding paper, film, or bound records eventually faces the same decision: convert records only when someone asks for them, or convert whole collections in a planned program. Scan-on-demand digitization converts physical records only when someone requests them, while full archive digitization converts an entire collection, or a defined part of it, in a planned sequence.

Both options have real strengths. Scan-on-demand keeps upfront spending low, delivers value within days, and aligns cost with actual use. Full digitization makes whole collections searchable, handles fragile originals once, and can reduce physical storage, but it needs planning and budget up front and adds ongoing digital storage and preservation costs. Neither is right for every archive, and many institutions get the best result by combining them deliberately.

This guide defines both strategies, compares them side by side with an illustrative five-year cost model, covers the capture, quality, metadata, and security standards both should follow, explains how to build the business case for different decision-makers, and shows how to segment an archive into a hybrid program with clear promotion rules and measurable targets.

Key Takeaways

• Scan-on-demand minimizes upfront cost, but it only digitizes what people already know to ask for. That makes it a useful service model and a weak discovery model, because the records nobody knows about stay invisible.
• Full digitization requires more planning and budget up front, but it makes records searchable and reduces repeated handling of originals. It can also free physical storage where retention rules allow.
• The decision should follow usage, preservation risk, compliance needs, discoverability goals, and total cost over several years, including ongoing digital storage and preservation. Budget cycles matter, but they should shape the phasing of the plan, not the strategy itself.
• Many institutions benefit from a hybrid approach. High-value and at-risk material is digitized fully, while the low-use long tail is handled on demand with a concrete rule for promoting a series to full conversion, such as more than five requests a year.
• Capture to named standards in both models, such as FADGI quality levels and ISO 19264-1 image analysis, with TIFF masters and PDF/A access copies. A rushed on-demand scan that must be redone later costs more than doing it properly once.

What is Scan-on-Demand Digitization?

Scan-on-demand, sometimes called digitize-on-request, converts individual files, boxes, or volumes when a user, staff member, or legal process needs them. The request triggers retrieval from storage, scanning, basic quality checks, and delivery, usually as a PDF or image set. Over time, the digitized items accumulate into a partial digital collection shaped entirely by past demand.

The model is common in records management, legal and insurance operations, local government, and special collections where only a small share of holdings is used in any given year. It works best when requests are predictable and the value of the collection lies mainly in retrieving known items rather than discovering unknown ones.

What is Full Archive Digitization?

Full archive digitization is a planned program that converts a whole collection or defined series in sequence. It typically includes inventory, preparation, capture to a documented standard, OCR, indexing, metadata, quality validation, and delivery into a discovery platform or records system. The goal is a complete, searchable digital collection rather than a set of individually requested files.

Because the work is batched, full programs can optimize equipment, staffing, and workflows in ways that one-off requests cannot. They also make it possible to design metadata and indexing consistently across the whole collection, which is what allows people to search across records rather than open them one at a time.

How Do the Two Strategies Compare?

What are the Advantages of Scan-on-Demand?

• It requires very little initial spend, which suits institutions without a dedicated capital budget. Costs are paid gradually through operating budgets as requests arrive, which makes the model easy to approve and sustain.
• Effort goes only to records that someone actually needs. No time or money is spent converting material that may never be consulted, which keeps early spending tightly aligned with demonstrated use.
• It is easy to start. There is no large inventory or planning phase, so an institution can begin offering digital delivery within weeks rather than months.
• It works well as a bridge. While a larger program is being funded, designed, or procured, on-demand scanning keeps users served and generates request data that later helps prioritize full conversion.

What are the Limits of Scan-on-Demand?

The Discoverability Problem

People can only request what they already know exists. Records that would surface through search, such as a name in a register, a reference in an engineering report, or a clause in an old contract, stay invisible because nobody knows to ask for them. For collections whose value lies in what researchers or staff would find, scan-on-demand never unlocks that value.

Repeated Handling

Each request means retrieving, handling, copying, and refiling originals. For fragile or deteriorating material, many small handling events over years carry more cumulative risk than one controlled conversion pass. Popular items can be handled dozens of times before they are ever captured properly.

Higher Cost per Page

One-off retrieval, preparation, equipment setup, quality checks, and delivery make each on-demand page more expensive than pages converted in batches. The five-year cost example later in this guide shows how this difference can add up for a frequently used series.

Inconsistent Quality

Rush requests are often scanned quickly on whatever equipment is available, without consistent resolution, OCR, or metadata. These files may be fine for a single user but unsuitable for long-term access, and they often need to be redone when the series is later converted in full.

Storage Costs Continue

Because most of the collection stays paper-only, physical storage, climate control, insurance, and retrieval staffing continue indefinitely. These costs rarely appear next to the scanning budget, but they are part of the true cost of the strategy.

What are the Advantages of Full Archive Digitization?

• Every record becomes searchable, which is where most of the research and operational value is created. Users can find names, dates, subjects, and references across the whole collection instead of relying on finding aids and staff memory.
• Originals are handled once under controlled conditions. This reduces preservation risk significantly, especially for brittle paper, faded ink, damaged bindings, and deteriorating film.
• Batch processing lowers cost per page and produces consistent quality. Standardized capture settings, OCR, and metadata make the collection easier to manage and integrate with other systems.
• Digital records support remote access, disaster recovery, and AI-assisted search. In practice, that means handwritten text recognition for registers and manuscripts, entity extraction that pulls out names, dates, and contract clauses, semantic search that finds records by meaning rather than exact wording, and automated classification that sorts documents by type.
• Physical storage can be reduced, consolidated, or retired where retention rules allow. For large, space-heavy series, the long-term savings on facilities can offset a meaningful share of the conversion cost.

What are the Challenges of Full Digitization?

• It requires an upfront budget and a multi-month or multi-year timeline. Institutions often need capital funding, grants, or phased budgets, which takes time to secure and plan.
• Planning takes real effort. Inventory, prioritization, standards, metadata design, and integration requirements all need decisions before capture starts, and rushing this phase creates rework later.
• Some material may have very low use. Converting everything without prioritization can spend budget on records that offer little return, which is why segmentation matters.
• Indexing and metadata require their own investment. Scanning alone does not create searchability, and programs that budget only for capture often discover the cost of recognition and structuring at delivery.

Digital collections also carry ongoing costs. Storage, backups in more than one location, integrity checks, and periodic migration to newer formats all cost money every year. The Digital Preservation Coalition’s Digital Preservation Handbook stresses that digital preservation is a continuing activity rather than a one-off project, so these costs belong in any honest comparison.

Which Factors Should Drive the Decision?

Usage Patterns

Review request logs for the past few years. Series requested frequently, or by many different users, usually justify full conversion. Series with almost no requests may be better suited to on-demand handling, at least until usage changes.

Preservation Risk

Fragile paper, acetate film, faded ink, and damaged bindings should move up the priority list regardless of demand. Every year of delay adds risk, and some damage cannot be reversed once it has occurred.

Compliance and Retention

Legal holds, audit requirements, freedom of information obligations, and retention schedules may require faster and more reliable access than scan-on-demand can provide. Missing a statutory deadline because a record was still on paper is a real operational risk.

Discoverability Goals

If the institution wants researchers, the public, or internal teams to find records by searching, full digitization with OCR and metadata is required for those collections. Scan-on-demand cannot deliver discovery at collection level.

Total Storage Cost

Compare the ongoing cost of physical storage, climate control, and retrieval against a one-time conversion cost plus ongoing digital storage, backup, and migration. As the five-year example shows, the balance often favors conversion for large, frequently used series, especially when facility leases or expansions are under review, while low-use series can remain cheaper on demand.

Budget Model

Scan-on-demand fits operating budgets. Full digitization often needs capital or project funding. Phased programs can spread full conversion across several budget cycles, which makes a large program more achievable without losing momentum.

How Do You Build the Business Case?

A digitization strategy usually needs approval from several people, and each one weighs the decision differently. A strong business case answers each of their questions directly.

What the CFO Cares About

Finance leaders focus on the split between capital and operating expense, the payback period, and avoided costs. Scan-on-demand sits in the operating budget and grows with demand, while full digitization is usually a capital or project expense followed by smaller ongoing digital costs. Present a five-year model like the example above, show the payback year for each high-use series, and include avoided costs such as storage leases, retrieval staffing, and facility expansion.

What the CIO Cares About

Technology leaders focus on integration and security. They will want to know which systems receive the output, such as an enterprise content management (ECM) or electronic document and records management system (EDRMS), a discovery platform, or a digital preservation system, and which formats and metadata standards are used. They will also ask about access controls, encryption, storage architecture, and how files will be migrated as systems change.

What Compliance Teams Care About

Compliance and legal teams focus on audit readiness and response deadlines. Freedom of information requests, data subject access requests, litigation holds, and regulatory audits often come with fixed timelines. Show how each option affects the ability to find and produce records within those deadlines, and how chain of custody and quality documentation will be maintained.

Which Capture and Metadata Standards Should You Use?

Both strategies should capture to the same named standards, so on-demand files never need to be redone. The FADGI technical guidelines define four quality levels, from one star to four stars, and many programs target three stars for production work and four stars for the most valuable or fragile originals. ISO 19264-1 provides a standard method for analyzing the image quality of scanners and cameras using test targets, which makes quality measurable rather than subjective.

Textual records are typically captured at 300 to 400 ppi, with higher resolutions for small type or degraded originals. Preservation masters are usually stored as uncompressed or losslessly compressed TIFF files, with PDF/A access copies for everyday use. In the United States, federal agencies digitizing permanent records must follow NARA’s digitization standards in 36 CFR 1236 Subpart E, which took effect in June 2023.

Metadata should also follow recognized standards. Dublin Core suits simple descriptive records, METS and MODS describe the structure and content of complex digital objects, and EAD describes archival collections and finding aids. Choose the standards your target system supports, whether that is an ECM or EDRMS platform for operational records or a digital preservation system for long-term retention.

For delivery, the International Image Interoperability Framework (IIIF) lets images be viewed, zoomed, and shared consistently across platforms. The IIIF Image API delivers images at any size from a standard web request, and the IIIF Presentation API describes how pages fit together, so on-demand and bulk outputs appear the same way to users and can be combined with collections from other institutions.

How Do You Check OCR and Indexing Quality?

OCR quality varies widely with the condition and type of material. Practitioners generally treat character accuracy above 98 percent as good, 90 to 98 percent as average, and below 90 percent as poor. A National Library of Australia study of its digitized newspapers found that raw OCR accuracy ranged from 71 to 98.02 percent across a representative sample, which shows why targets should be set per collection.

Quality is best checked through sample-based QA with defined acceptance levels. Each batch is sampled using an acceptance sampling plan, such as those in ANSI/ASQ Z1.4, and checked against targets agreed in advance, for example 98 percent character accuracy for printed text and 99 percent accuracy for indexed fields such as names, dates, and record numbers. Batches that fall below the target are reworked and resampled, and results are recorded for audit.

How Do You Protect Security and Privacy?

Many records contain personal, financial, or health information, so security and privacy apply to both strategies. Requirements vary by jurisdiction and record type, so confirm specifics with your records officer or legal counsel.

1. Chain of custody. Log every box, file, and volume from retrieval to return, including who handled it and where it was processed, so originals are controlled throughout and nothing is lost or substituted.
2. PII redaction. Identify and redact personal information in access copies where required, while keeping unredacted masters under tighter controls.
3. Access controls. Restrict access by role during processing and apply the same controls to digital outputs that applied to the originals.
4. Encryption. Encrypt files in transit and at rest, including backups and any files shared with a digitization partner.
5. Regulatory alignment. Map the program to the rules that apply to your records, such as GDPR in the European Union, HIPAA for US health information, and India’s Digital Personal Data Protection Act, 2023.

Should You Digitize In-House or Outsource?

In-house digitization gives direct control and suits small, steady volumes, but it requires equipment, trained staff, quality systems, and secure facilities. Outsourcing to a digitization partner suits large programs, specialized formats, and fluctuating demand, and moves equipment and staffing costs to the provider. Common pricing models include per page or per image for bulk work, per request with a minimum charge for on-demand work, and fixed project or managed service pricing for long programs. For on-demand work, agree turnaround service levels in advance, such as two to three business days for standard requests and 24 hours for urgent or statutory requests.

Why Does a Hybrid Strategy Often Work Best?

A hybrid approach uses each strategy where it fits best. Across the programs DDD supports, it is how most mature records and archive programs operate, although this reflects practitioner experience rather than a published survey.

1. Digitize high-value, high-use, compliance-critical, and at-risk collections fully. These are the series where searchability, preservation, and fast access create the most value, so they justify the planning and budget of a full program.
2. Handle the low-use long tail on demand, captured to the same standard as bulk work. This keeps costs aligned with actual use while making sure every on-demand file is reusable later.
3. Set promotion rules that move a series from on-demand to full conversion once requests reach a defined threshold. For example, a series requested more than five times a year, or by three or more different users, moves into phased full digitization. This turns request data into an objective trigger instead of relying on ad hoc decisions.
4. Review request logs and condition surveys periodically to update priorities. Usage and preservation risk change over time, and the plan should change with them.

How Do You Segment an Archive for a Hybrid Program?

Segmentation usually starts with a high-level inventory by series, then adds usage data, condition information, and compliance requirements. The result is a tiered plan that stakeholders can understand and approve, with clear reasons for why each collection sits where it does.

How Do You Estimate Cost for Each Strategy?

For scan-on-demand, estimate annual request volume, average pages per request, retrieval and refiling time, scanning time, quality checks, and delivery effort, then add continuing storage costs. For full digitization, estimate total volume, material types, condition, preparation effort, capture, OCR, indexing depth, and quality validation, then add ongoing digital storage, backup, and format migration.

The most reliable way to compare is to convert a representative sample from each tier, measure actual throughput and cost per page, and model both strategies over a five-year horizon. Sampling replaces assumptions with numbers from your own collection, which makes the business case much easier to defend.

How Do You Measure Whether the Strategy Is Working?

Whichever model you choose, track a small set of metrics each quarter so the strategy can be adjusted with evidence rather than habit. The example targets below are starting points to adapt to your own service levels and obligations.

What are the Most Common Mistakes?

• Defaulting to scan-on-demand for collections whose value depends on searchability. The result is a digital service that answers known questions but never supports discovery, which is often the main reason the collection matters.
• Digitizing everything in the order it sits on the shelf. Converting by physical location rather than value, use, and risk spends early budget on the wrong material and delays the collections people need most.
• Scanning on demand without OCR, metadata, or a capture standard. These files may satisfy a single request, but they create a fragmented digital collection that is expensive to fix later.
• Ignoring ongoing costs when comparing options. Physical storage, retrieval staff, and climate control make on-demand look cheaper than it is, while digital storage, backup, and migration are often left out of full digitization budgets.
• Converting records that are already past their retention period. A retention review before digitization avoids spending money on material that could be lawfully disposed of.
• Treating the strategy as a one-time decision. Usage, condition, and priorities change, so tier assignments should be reviewed at least once a year.

How Digital Divide Data Can Help

DDD has delivered content digitization for 25 years. We help institutions segment their holdings, build the business case, and run bulk and on-demand content digitization to the same standard. Our teams handle inventory, condition review, and standards-based capture, then apply OCR and document conversion with sample-based quality checks so records are accurate and findable. Through data engineering, we deliver outputs structured for your ECM, EDRMS, or discovery platform, with chain of custody and access controls documented throughout.

A good next step is a sampled assessment of your three highest-use series, which gives you real cost, accuracy, and turnaround figures for your business case. Talk to an expert to schedule one.

Conclusion

Scan-on-demand and full archive digitization solve different problems. Scan-on-demand serves known requests at low upfront cost and delivers value quickly. Full digitization makes collections discoverable, protects originals, and can reduce storage, while adding ongoing digital costs. The strongest programs segment the archive by value, use, and risk, digitize the most important material fully, handle the long tail on demand with clear promotion rules, and capture everything to one standard. The practical way to start is to take your three highest-use series, sample them, and model both strategies over five years using your own numbers.

References

Appleby, M., Crane, T., Sanderson, R., Stroop, J., & Warner, S. (2020). IIIF Image API 3.0. International Image Interoperability Framework Consortium. https://iiif.io/api/image/3.0/

Appleby, M., Crane, T., Sanderson, R., Stroop, J., & Warner, S. (2020). IIIF Presentation API 3.0. International Image Interoperability Framework Consortium. https://iiif.io/api/presentation/3.0/

Digital Preservation Coalition. (2015). Digital preservation handbook (2nd ed.). https://www.dpconline.org/handbook

Federal Agencies Digital Guidelines Initiative. (2023). Technical guidelines for digitizing cultural heritage materials (3rd ed.). https://www.digitizationguidelines.gov/guidelines/digitize-technical.html

Holley, R. (2009). How good can it get? Analysing and improving OCR accuracy in large scale historic newspaper digitisation programs. D-Lib Magazine, 15(3/4). https://doi.org/10.1045/march2009-holley

International Organization for Standardization. (2021). Photography – Archiving systems – Imaging systems quality analysis – Part 1: Reflective originals (ISO Standard No. 19264-1:2021). https://www.iso.org/standard/79172.html

National Archives and Records Administration. (2023, May 4). Federal records management: Digitizing permanent records and reviewing records schedules. Federal Register. https://www.federalregister.gov/documents/2023/05/04/2023-09050/federal-records-management-digitizing-permanent-records-and-reviewing-records-schedules

Frequently Asked Questions

What is scan-on-demand?

Scan-on-demand digitizes records only when someone requests them. It keeps upfront costs low, and requests should still be captured to full standard so each file can be reused when the series is later converted.

Is full archive digitization worth the cost?

For high-use, high-value, or at-risk collections, usually yes. Check the payback year using your own request logs, because low-use series may never recover the conversion cost.

Which is cheaper per page?

Full digitization is usually cheaper per page because batch processing avoids repeated retrieval and setup. Remember to add digital storage and migration costs when comparing totals.

Can we combine both approaches?

Yes. Many institutions digitize priority collections fully and handle low-use material on demand, with a rule such as more than five requests a year triggering full conversion.

How do we decide which records to digitize first?

Rank collections by usage, preservation risk, compliance needs, and research value. Series that face statutory response deadlines deserve early attention even if overall demand is modest.

Do on-demand scans need OCR and metadata?

Yes, if you want them to be reusable and searchable later. Recording at least Dublin Core descriptive fields at request time makes later integration far easier.

asit dubey

Asit Dubey is a global operations leader with almost 30 years of experience across digitization, publishing, AI/ML, and LegalTech, currently serving as Executive Vice President at Digital Divide Data. He has led large-scale operations (3,500+ workforce) across APAC, EMEA, and North America, driving AI-led transformation and process excellence. A Six Sigma Black Belt, he specializes in automation, solutioning, and cost optimization, delivering productivity gains of over 300% and significant margin improvements. He has successfully scaled revenues from $750K to $3M+ monthly while turning around underperforming units. His expertise spans global delivery setup, GTM strategy, and client engagement. He is known for building resilient, multi-geo delivery models and enabling organizations to transition to AI-powered services.