Digitisation Is Not the Finish Line: Clearing the Hidden Backlog

There is a comforting story the heritage sector tells itself about digitisation: scan the material, and the job is done. The reality on the ground is different. Walk into most county record offices, university special collections, or community archives and you will find a growing store of digital images — TIFFs of glass-plate negatives, PDFs of parish registers, recordings of oral histories — that almost nobody can find, because almost none of it has been described. The scanner did its work. The catalogue never caught up.

This is the quiet crisis of UK archives. Digitisation is not the finish line. It is barely the starting gun. An image without a record attached is invisible: it cannot be searched, cited, requested, or surfaced to a researcher who would value it. The backlog has simply changed form, from boxes you cannot open to files you cannot find.

The real bottleneck is description, not scanning

Funding bodies, trustees, and the public tend to measure progress in items scanned. It is a tangible number and it photographs well in an annual report. But scanning capacity is no longer the constraint. Flatbed and overhead scanners are cheap, volunteers can be trained on them in an afternoon, and bureau services will digitise at scale for a known price per image.

Cataloguing is where projects stall. Producing a usable archival description — a title, a date range, a scope and content note, named people and places, a reference following ISAD(G) — is slow, skilled work. A professional archivist working carefully through unsorted material catalogues, in many cases, only two to three boxes a day. That is not a criticism of archivists; it is the nature of intellectual control. Every record requires reading, judgement, and context.

Multiply that rate against a donated collection of two hundred boxes and the arithmetic is sobering. Years, not months. And so collections sit. Donated papers accepted with genuine enthusiasm in 2014 remain unprocessed in 2026, not through neglect but through sheer want of hours. We have written before about the hidden cost of uncatalogued collections — the storage you pay for, the access you cannot offer, the funding case you cannot make for material the public never sees.

Why the backlog keeps growing

The backlog is structural, not accidental. Three forces feed it.

First, acquisition outpaces capacity. Archives accept material faster than they can process it, because refusing a significant deposit is often not an option — the alternative may be its loss. Each accession adds to the queue.

Second, digitisation projects rarely fund the description that gives them meaning. A grant pays for scanning equipment and a fixed-term digitisation officer. The post ends; the images remain; the descriptive work was scoped as something that would “follow on”. It seldom does.

Third, cataloguing does not scale with headcount the way scanning does. You can double your scanning throughput by buying a second scanner. You cannot double your descriptive output by buying a second archivist, because trained archivists are scarce and the work resists shortcuts.

The result is the paradox at the heart of a digitised archive that tells only half the story: the more you digitise, the larger your invisible backlog grows, unless description keeps pace.

What AI changes — and what it must not

This is precisely the gap that AI-assisted description is built to close, and it is worth being exact about how. The point is not to replace the archivist. It is to remove the blank page.

The slowest part of cataloguing is starting from nothing: confronting a folder of unfamiliar correspondence and constructing a first-draft description from scratch. AI can do that first draft. Optical character recognition and handwritten text recognition turn the image into searchable text. Automated transcription does the same for oral histories and audio-visual material. From that text, a model can propose a title, a date range, a summary, candidate subjects, and the people, organisations, and places mentioned — a structured starting record for every item, generated in seconds rather than hours.

The archivist’s role does not shrink; it shifts. Instead of composing every field, they review, correct, and approve. Judgement moves to where it adds the most value. The two-to-three-boxes-a-day archivist is now reviewing drafts rather than building records from raw material — and the throughput changes accordingly.

But this only works if the AI sits in the right place. Archivers.ai is the AI processing and access layer that sits in front of your existing systems — AtoM, ArchivesSpace, CALM, Axiell — not a replacement for them. It proposes; the archivist disposes. Every AI-generated field carries a per-field confidence flag, so review is targeted: the archivist’s attention goes to the uncertain suggestions, not the obvious ones. Nothing is finalised without human approval, and the way we use AI is built around that review gate rather than around it.

Description that exports cleanly

Clearing a backlog is pointless if the output is trapped. The records you create have to land in your repository and your public catalogue in formats those systems accept. Archivers.ai exports standards-aligned finding aids — EAD3, Dublin Core, ISAD(G)-structured records, BagIt packages with PREMIS preservation metadata, and SPECTRUM-aligned fields for museum material. The description you produce is portable, not locked inside another silo.

Authority matters too. As you work through a backlog, the same person appears across many collections. Reconciling those mentions against VIAF, FAST, and Wikidata means one individual becomes one authority — so that, once described, the material is genuinely discoverable across your holdings rather than scattered under inconsistent name forms.

Backlog clearance for under-resourced teams

This approach is aimed squarely at the organisations that need it most: community archives run by volunteers, grant-funded project archivists with a fixed-term post and an immovable deadline, and professional archivists in record offices carrying decades of accumulated deposits. None of them has the luxury of catalogue-everything-from-scratch at two boxes a day.

For a head of service or director, the case is straightforward. The expensive asset is the archivist’s expertise. Spending that expertise composing first drafts of routine descriptions is poor value; spending it on review, arrangement, and the difficult judgements that genuinely need a human is exactly where it belongs.

Digitisation made the material visible to a camera. Description makes it visible to the public. Until the second job is done, the first is unfinished — and the backlog you cannot see is the one that costs you most.

If your digitised material is sitting undescribed, you can join the Archivers.ai waitlist and run a sample of your backlog through the workflow in early access — with transparent pricing published when we launch. Digitisation is not the finish line. Let us help you cross it.