How Much a DMS Actually Costs

Most document management business cases are built around a per-user list price, and most of them are wrong by a wide margin for the same reason: the licence is the smallest line in the first year.

Here is where the money actually goes.

Licence#

The visible number, and the one vendors compete on. Per user per month, sometimes with storage tiers, sometimes with a platform fee.

Two things distort the comparison. First, the definition of a user varies — some vendors charge for read-only access, some do not, and if you have many readers and few authors that distinction moves the total more than the headline rate. Second, capability tiers: the feature that made you shortlist a product is often in a higher tier than the one quoted.

Ask for the quote against your actual user shape, not against a seat count.

Migration#

Usually the largest single first-year cost after licence, and the most variable.

It is governed by the number of decisions rather than by volume: what stays behind, how permissions map, which content keeps version history, how existing retention rules translate. Extraction throughput is predictable. Decisions are not, which is why migration estimates given before an inventory exists are guesses.

Integration#

A document system that does not connect to the systems where work happens gets used as an archive. The connections that matter are usually identity, email, the finance or ERP system, and whichever line-of-business application generates the documents.

Standard connectors cost configuration time. Anything non-standard costs development time, and development against a vendor API is where a modest project becomes a large one. Establish which of your required integrations are standard before signing, not after.

Configuration#

Document types, metadata, retention classes, permission roles, approval workflows and templates. This is real work whether a consultant does it or you do, and doing it badly is expensive later because a bad taxonomy is difficult to change once documents are filed against it.

The cost here is mostly internal time, which is why it disappears from business cases — it is not an invoice.

Training and adoption#

Underspent almost universally. The system that people do not know how to file into becomes a system with a shared drive still running alongside it, which means you now pay for both.

Budget training by role rather than uniformly: authors and approvers need real training, readers need very little, and administrators need considerably more than a session.

Internal time#

The line item that does not appear anywhere and is frequently the largest. Somebody has to decide the taxonomy, agree the retention schedule, own the permission model, run the reconciliation and answer questions for six months.

If that person is doing it alongside their existing role, the cost is real but hidden, and the project takes longer than planned for reasons that look inexplicable afterwards.

Running administration#

Someone maintains the document types, adds the new retention class, fixes the workflow that broke when a department reorganised, and reviews the permission exceptions. This is a permanent role, usually part-time, and it is the cost that appears in year two when the implementation team has dispersed.

A reasonable planning figure is a named administrator with a defined proportion of their time, and a documented deputy. Systems without either drift, and the drift is expensive to correct because it is discovered during an audit rather than during a review.

The exit cost nobody models#

Whether you can leave. A repository that cannot be exported completely — content, metadata, version history and audit trail, in a documented format — carries a future migration cost that is unbounded because it depends on the vendor's cooperation at the moment you least have it.

This is worth testing during evaluation by requesting a sample export, not by reading the documentation.

Building a case that survives review#

Model the seven lines separately, state the assumptions behind each, and be explicit about which are estimates. A business case showing a licence figure and a benefits figure invites the question nobody in the room can answer; one showing seven lines and their basis invites a decision.

It is also worth stating which benefits you are not claiming. Time-saved calculations built on an assumed number of minutes per search are easy to produce and easy to dismiss, and a case resting on one tends to collapse under the first sceptical question. The benefits that hold up are the ones with a countable base: the audit finding you no longer expect to receive, the contract renewal nobody missed, the storage contract you can end, the migration you will not have to run twice.