Ask how long a scanning project will take and you will often get a page count divided by a machine rating. That calculation is close to meaningless. It describes the one part of the process that is automated, fast and rarely the constraint.
Watch a production line for an hour and the picture is different. Most of the people in the room are not operating scanners. They are at desks with staple removers, at screens verifying index fields, or on the phone asking someone at the client what an unlabelled folder is.
What sets the duration
| Factor | Effect on the schedule | What you can do about it |
|---|---|---|
| Preparation effort | Usually the largest single block of elapsed time, and it scales with fastener density and damage, not with page count | Sample a box and count the staples. Consider preparing low-complexity categories in-house to the supplier's standard |
| Undecided index rules | Stops indexing entirely, and unresolved rules generate rework on everything already processed | Lock the field list and the document-splitting rules before the pilot, not during production |
| Exception volume | Every unreadable, unidentifiable or oversized document leaves the line and waits for a human decision | Agree an exceptions protocol and a named decision-maker with authority, available on a short turnaround |
| Client approval cycles | Frequently the biggest source of dead time. A pilot awaiting sign-off blocks the whole line | Name one approver, agree a response window, and give them a deputy |
| Access to the records | On-site work is bounded by your building hours, security escorts and available floor space | Confirm access hours and room availability before the schedule is drawn |
| Document mix | Reconfiguring a line between formats costs time on every switch | Sort by format during preparation so each configuration runs continuously |
| OCR and extraction scope | Adds processing time and, more significantly, verification time on low-confidence results | Set confidence thresholds deliberately. Verifying everything and verifying nothing are both wrong |
| Live retrieval during the project | Every recall of a file already in the pipeline interrupts the flow | Agree a retrieval process up front, or digitize the most-requested categories first |
It also helps to separate effort from elapsed time. A project can consume a modest amount of labour and still run for months, because the labour spends its days waiting: for a decision, for a department to release its boxes, for a pilot to be reviewed. When a schedule slips, the useful question is not who worked slowly but which stage the work queued in front of. The answer frequently sits on the client side, which is uncomfortable and much better known early.
How a credible schedule is built
- 1
Assessment
A physical survey of representative boxes across every category. This produces the condition profile the whole plan rests on.
- 2
Pilot batch
A small, deliberately representative batch taken through every stage to delivery. It calibrates preparation rates, exception rates and index accuracy against your real material, and it surfaces the rules nobody thought to write down.
- 3
Rule freeze
Index fields, document splitting, naming, output format and exception handling are agreed and signed off. Changes after this point cost rework, and the schedule should say so.
- 4
Ramp-up
Production starts below full rate while operators learn your material and the quality feedback loop tightens. Any plan that assumes full throughput from day one is optimistic.
- 5
Steady-state production
Preparation, capture, quality control and indexing run in parallel on different batches. This is the phase a page count can genuinely predict, and it is usually a minority of the elapsed time.
- 6
Reconciliation and closeout
Counts are reconciled, exceptions are cleared, the final delivery is loaded and verified in your system, and the originals are returned, stored or destroyed as instructed.
Note where the parallelism sits. Stages overlap in steady state, which is why adding people helps in the middle of a project and helps very little at either end.
What actually stalls projects
- A pilot delivered and then left unreviewed for weeks while the line sits idle.
- An index field that turns out not to be printed on most documents, discovered after ten thousand pages.
- Nobody empowered to decide what an unlabelled 2011 folder is, so exceptions pile into a queue that becomes a project of its own.
- A department that quietly refuses to release its boxes because it still uses them daily.
- A quality standard defined as high accuracy rather than as a measurable threshold on a defined sample, which makes sign-off a matter of opinion.
- Boxes that turn out to hold far more paper than the survey assumed, because the survey box was the tidy one.
Compressing a timeline without breaking it
Deadlines move in. Some responses to that are legitimate and some quietly trade quality for speed.
- Legitimate: adding shifts, adding parallel capture lines, prioritising the categories with the real deadline and scheduling the rest afterwards, and preparing in-house alongside the supplier.
- Legitimate: reducing index fields on low-value categories, provided the decision is explicit and recorded.
- Risky: cutting the pilot. It removes the calibration that the rest of the plan depends on.
- Risky: lowering the quality-control sample rate. It does not make errors disappear, it makes them arrive later.
- Usually pointless: buying faster scanners. The scanner was not the constraint.
Frequently asked questions
Why can a supplier not give a completion date up front?
Because the two variables that dominate the schedule, preparation effort and exception rate, are only measurable once someone handles your actual documents. A date given before a survey and a pilot is an assumption presented as a commitment. Expect an indicative window before the pilot and a firm schedule after it.
Does scanning more pages per day shorten the project?
Only if capture is the constraint, which it usually is not. If preparation is the bottleneck, more scanning capacity produces an idle scanner and a longer queue of unprepared boxes. Look at where work accumulates between stages, and add capacity there.
Can we keep using files while they are being digitized?
Yes, with an agreed retrieval process. Records in the pipeline can normally be located and returned, but every recall interrupts a batch. Where daily access is genuinely needed, the better approach is to schedule those categories first so they come back digital rather than physical.
How long does the preparation stage take?
It depends almost entirely on condition rather than volume. Loose, uniform sheets move quickly. Heavily stapled files with taped inserts, mixed sizes and torn pages take substantially longer per box. This is exactly why a physical assessment and a pilot batch come before any schedule is committed.
About this article
Written and reviewed by the digitization delivery team at Document Digitization Services, the specialist division of Athena Global Technologies LLC. Content is reviewed against how projects are actually run, and updated when that changes.
Read next
- How to prepare documents for scanningThe stage that usually sets the schedule.
- What drives digitization costCompressed timelines and cost are the same conversation.
- How we run a projectThe stages above, as an engagement.
- On-site vs off-site scanningWhere the work happens changes what bounds the schedule.