Records Are Evidence: Document Control as Institutional Memory

A capital project generates an enormous quantity of documents. Drawings, specifications, method statements, material submittals, shop drawings, test certificates, inspection records, correspondence, minutes, instructions, reports, contracts, invoices, permits, warranties, as-builts and operating manuals. On a large project the count runs into hundreds of thousands.

Almost all of it will never be read again. A small fraction of it will determine the outcome of a dispute involving very large sums, and nobody knows in advance which fraction.

This is the fundamental economics of document control, and it explains why organisations systematically under-invest in it. The cost is certain and immediate. The benefit is probabilistic and deferred. Under budget pressure, the certain immediate cost loses.

The three things a record must survive

For a document to be useful when it eventually matters, three properties must have been preserved from the moment of its creation.

It must be findable. Not "stored somewhere" but retrievable by someone who does not know it exists, searching by subject, by date, by project element, by party, or by relationship to another record.

Its version status must be unambiguous. The single most expensive document control failure in construction is work executed to a superseded drawing. It occurs constantly, and it occurs because the recipient had no reliable way to know that the version in their possession was no longer current.

These three properties are what an Electronic Document Management System within a PMIS provides as a structural matter. Documents are registered with controlled metadata. Revisions supersede predecessors automatically and visibly. Transmittals are records with recipients, dates and acknowledgement status.

The submittal and RFI cycle

Two processes account for a disproportionate share of both administrative volume and project delay: material and shop drawing submittals, and requests for information.

The mechanics are simple and the failure modes are consistent. A submittal is issued, reviewed, returned with comments, revised and resubmitted, sometimes several times. An RFI is raised, routed to the party who can answer it, answered, and closed. In principle both are straightforward. In practice both routinely take three to five times longer than the contract allows.

The reasons are almost never technical. They are queueing failures. The item sits with a reviewer who has forty others. It is routed to the wrong discipline and returns after a week. It is answered by email to one person who does not circulate it. Nobody has a consolidated view of what is outstanding and how long it has been outstanding, so escalation happens when someone complains loudly rather than when a threshold is breached.

Digitalizing these cycles addresses the queueing directly. Every item has a status, an owner, a target date and an elapsed time. The register is complete and shared by all parties. Overdue items escalate automatically. The distribution of a response is defined by the workflow rather than by whoever happens to be on the email.

The effect on programme is frequently larger than anything achieved by construction acceleration, and it costs nothing to implement beyond configuring the process properly.

Correspondence is a contract instrument

Letters, notices and formal correspondence on a construction project are not administration. They are the mechanism by which contractual rights are preserved and lost.

Notices of delay must be served within contractual periods. Notices of claim likewise. Instructions must be confirmed. Objections must be raised in time. Approvals must be given or deemed given. A right not exercised within its window is frequently a right lost permanently, regardless of the underlying merits.

Managing this through email is managing it without a system. There is no register, no time bar tracking, no assurance that the correct recipient received it, no reliable way to identify which correspondence relates to which contractual provision, and no institutional record when the individual leaves.

Correspondence managed within a PMIS is a registered record: numbered, categorised, linked to the contract clause, the project element and the related records, with transmission and acknowledgement captured, and with time bars tracked as a status rather than remembered as an obligation.

The handover problem

There is a moment on every capital project when the accumulated cost of poor document control becomes payable in a single instalment. It is called handover.

The entity requires as-built drawings, operation and maintenance manuals, test and commissioning records, warranties, spare parts schedules, asset data and training records. These are contractual deliverables, usually tied to final payment.

Where document control has been weak, this material is assembled at the end from whatever can be found. Test certificates for equipment installed three years earlier are missing. Warranty start dates cannot be established. As- builts do not reflect late variations. The asset register is compiled by someone walking the building.

The consequence is a delayed handover, a facilities management function that begins operating an asset it does not understand, warranties that lapse unclaimed, and a maintenance regime built on incomplete information. The cost falls on the operating budget for the life of the asset, which is decades.

Where handover documentation is collected progressively — the test certificate registered when the test passes, the warranty registered when the equipment is accepted, the asset record created when the asset is installed — handover becomes a verification exercise rather than an archaeology exercise. This requires nothing more than deciding at the outset that the O&M deliverable is a running obligation rather than a final one, and configuring the system to enforce it.

Records and the audit

Entities in the government, semi-government and PIF categories are audited, and the audit population includes procurement, commitment, certification and payment.

An audit conducted against a well-maintained digital record is a different experience from one conducted against shared drives and mailboxes. Sampled transactions are produced with their full approval history and supporting evidence in minutes. Population-level testing becomes possible rather than sample-based, because the population is queryable. Findings relate to substance rather than to the inability to locate documentation.

The reverse case is familiar to anyone who has lived through it: weeks of staff time diverted to document retrieval, findings raised because evidence could not be produced rather than because a control failed, and a management response that commits to improvements in record-keeping, again.

The stakeholder's view

Documents are the least visible part of project governance and among the most consequential. For a stakeholder assessing an entity's control environment, one request is diagnostic.

Ask for the current approved revision of a specific drawing, the transmittal that issued it, the list of parties who received it, and the previous revision it superseded.

If that arrives within the hour, the entity has document control. If it takes days and arrives with caveats, then the entity has document storage — and the difference will become apparent at handover, in the first significant claim, or in the next audit.

Next in the series: the processes that protect people and assets — risk, quality, safety and compliance.