Why Document Processes Cost More Than Data Processes, and When They Still Pay Off

A question that comes up regularly in sales meetings runs roughly like this: the delivery notes arrive as PDF by mail, surely they can just be filed automatically. The honest answer is uncomfortable. A process that moves files costs more than a process that moves values. Not twice as much, but noticeably more, and for reasons that have nothing to do with how good the supplier is.

This is not a warning against document processes. They are often the ones with the highest return, because they replace the most manual work. It is a warning against proposals that ignore the difference, because that difference reappears later as a claim for extra payment.

Why a file is not a data field

An invoice number weighs nothing. It gets read, passed along and written, and whether there are ten or ten thousand of them barely changes the process.

A file is genuinely transported between the systems involved. It gets re encoded for transport, which makes it about a third larger, and it sits entirely in the memory of the executing system while that happens. On the way it meets at least three separate limits, and they are rarely documented in the same place: what the execution environment can hold per run, what the target system accepts on upload, and what a mailbox lets through per message. With many providers the mailbox limit sits around 25 megabytes including encoding.

None of that is a problem while files stay small. It becomes a problem when something changes outside your company. A supplier replaces a scanner and suddenly sends colour scans instead of lean PDF files. A customer attaches three photos to a complaint. A sender switches to a download link because their own mailbox stopped letting the file through.

The failure that produces no error

The real risk is not that a large file fails. It is how it fails.

When an upload runs into a timeout, what comes back is often not an error message but an empty response. To the process that looks like a result. The mail is marked as handled, the run counts as successful, and the log shows nothing unusual. The gap surfaces only when somebody looks for a document and cannot find it, and that regularly happens weeks later while reconciling an invoice.

This is why the key question in a document process is not whether the file was sent, but whether it actually arrived at its destination and can be opened. That verification step is exactly what cheap proposals leave out.

Where the extra effort really sits

Four things come on top when files are involved, and each of them is work that should be visible in the proposal.

Verification. After filing, the process checks that the file exists at the destination, that its size matches the source and that it opens. Only then does the case count as done.

Handling of oversized files. There has to be a rule for the case where a file exceeds a limit. File it and send a link instead of an attachment, split it, or deliberately route it to a person. Which of those is right is a business decision, not a technical one.

Telling documents from noise. Incoming mail carries plenty of attachments that are not records. Signature images, logos and marketing attachments all count as attachments technically. Without a rule, a company logo ends up in the document archive while the actual delivery note sits in a mail where it happened to be the second attachment.

Naming at the destination. Two files with the same name are the classic way one record overwrites another with nobody noticing. Unique naming built from date, sender and reference number sounds trivial and is the difference between an archive and a folder.

What this means for price and payback

As a rough estimate rather than a measurement: a process that moves, verifies and files documents sits noticeably above a comparable process that only transfers values. The surcharge is created almost entirely in error handling, not in the business logic.

Whether it pays off is something you can calculate with your own numbers before requesting a single proposal. Take the number of records filed, renamed or searched for by hand each month, and multiply it by the minutes a case actually takes, searching included. In companies dealing with goods receipt, freight papers or incoming invoices, that calculation usually produces a clear figure. On top of it sits the part that cannot be expressed in minutes: a record that cannot be found during an audit costs more than a full year of filing time.

Three points to settle before the proposal

How large are the files in the worst case, and what is the biggest one that has actually turned up? That figure is sitting in your mailbox and takes half an hour to establish.

What should happen when a file is too large or missing? Stop and report, or file it and route it for manual work? That is your decision, and if it is missing from the proposal, somebody else will make it later.

Who notices when a record fails to arrive? If the answer is nobody, verification is not an option. It is the core of the assignment.

Sophera Consulting settles exactly these points in the analysis before anything is built, and writes size limits, verification and the route for exceptions into the statement of work before naming a fixed price, with no subscription. The entry point is the free automation check.

The recommendation

Treat documents as their own line in the proposal, not as a side effect of a data process. Insist in writing that every filing step is verified at the destination, and that a failed case lands with a named person.

A proposal that prices document processes at the same rate as pure data processes is not cheaper. It is incomplete, and the missing part is precisely the one you will miss after the first record disappears.