Someone always asks for a number. A trustee wants a budget line. A bid writer needs a defensible figure for the NLHF form. A director wants to know if converting the card index is a two-week job or a two-year one.
The honest answer is “it depends” — but that isn’t a number, and it won’t survive a funding panel. This post breaks down what the “it depends” actually consists of, so you can put a real figure on it.
One scoping note first: this is about retroconverting your catalogue — turning an existing paper index, card system, or typed finding aid into structured digital records. It is not about digitising the source material itself (scanning documents, photographing objects). Those are different jobs with different cost shapes. If you need that distinction spelled out, see retroconversion vs digitisation. If you’re building a full NLHF bid budget covering both, our guide to budgeting for digitisation walks through the whole line item by line item. This post is narrower: what moves the price of converting the catalogue alone.
The variables that actually move the number
Volume. How many cards, entries, folders, or pages are you converting? Everything else in this list is a multiplier against this figure, so get an accurate count before you ask anyone for a quote. “About a thousand” and “1,340 across four series” produce very different budget lines.
Condition and legibility. A typed card index in a consistent hand is quick work. A box of handwritten slips in three different clerks’ handwriting over sixty years, with abbreviations nobody has used since 1974, is not. Legibility is the single biggest hidden cost driver, because it determines how much of the work can be extracted with confidence and how much needs a human to puzzle it out.
How much structure already exists. A rigorous card index held in strict series order, with consistent fields, converts faster than a shoebox of loose slips in no particular sequence. If someone has to reconstruct the original order and relationships before any conversion can start, budget for that as its own task — it’s archival detective work, not typing.
The target standard and hierarchy. Mapping flat card entries into an ISAD(G)-aligned fonds/series/file/item hierarchy takes more judgement than a straight one-to-one transcription. Decide upfront how deep you want that hierarchy to go — a shallow map is cheaper and faster than one that reconstructs every level.
Whether reference numbers must be preserved exactly. If researchers, publications, or other systems already cite your existing reference numbers, those numbers need to survive the conversion unchanged. That’s a validation step, not just a data entry rule, and it’s worth costing separately rather than assuming it’s included.
Review capacity in-house. However good the extraction — AI-assisted or manually keyed — a person still has to check it before it’s trusted. That review time is real staff cost. It’s easy to budget for the conversion and forget the checking, and then find the checking eats the time you thought you’d saved.
The sanity-check method: pricing it the traditional way
Freelance cataloguers and keying bureaux typically price retroconversion per record or per day. There’s no single standard rate we can quote you here, and be wary of anyone who gives you one without seeing your material first — legibility and structure (above) swing the rate as much as volume does.
The useful exercise is this: ask any supplier for their per-record or per-day rate, multiply it by your volume, and then add your own estimate of review time on top. That last step is the one people skip, and it’s usually where a quote that looked affordable turns out not to be. A quote that includes review is comparing like with like; one that doesn’t isn’t finished.
What to ask any quote to cover
Before you compare figures, make sure every quote answers the same set of questions:
- What happens to existing reference numbers — preserved, remapped, or cross-referenced?
- What standard does the output map to, and how deep is the hierarchy?
- What export formats are included — EAD3, Dublin Core, BagIt+PREMIS, CSV for your target system?
- Who reviews the entries the process is uncertain about, and is that reviewer’s time costed in or billed separately?
- Is there an audit trail from the original card or entry to the finished digital record, so a query can be traced back to source?
A quote that doesn’t answer all five isn’t necessarily a bad one — but you should ask before you sign, not after.
Where Archivers.ai changes the shape of this
Archivers.ai’s managed retroconversion service is built around AI-assisted extraction plus human review, rather than pure manual keying. The platform transcribes and structures entries against an ISAD(G)-aligned hierarchy, flags anything it isn’t confident about, and exports to EAD3, Dublin Core, BagIt+PREMIS, or your target system’s CSV format.
The review step isn’t a hidden extra — it’s built into the workflow. Every record carries a draft or review status, unsupported fields and low-confidence words show you where to look first, and nothing reaches a public portal without a human approving it. That means the cost of checking is visible from the start, not a surprise you discover halfway through. If you’re pulling together a full NLHF bid budget, pair this with our digitisation budgeting guide — and if you want a figure scoped to your own catalogue, get in touch and we’ll work through the variables above with you directly.