What to keep, and what to let go, in a compliance archive

There is a comforting belief in compliance that the safe thing is to keep everything. Throw nothing away, and you can never be accused of having destroyed the document that mattered. In our view this is a mistake dressed as caution. An archive that keeps everything is not safer than one that curates; it is just slower to search and harder to trust. The document a regulator asks for is in there somewhere, buried under fifteen years of superseded drafts, duplicate scans and records of works on buildings you no longer manage. Keeping everything feels responsible. What it actually does is make the things you must be able to produce harder to find at the moment you need them most.

Why hoarding is not the same as keeping records

The purpose of a compliance archive is not to hold the maximum number of documents. It is to let you produce the right document quickly and prove the building was managed well. Those two aims are in tension with hoarding. A bloated archive slows retrieval, multiplies the chance of pulling a superseded version by mistake, and erodes confidence that what you hand over is current and complete. We think the better mental model is a curated record, not a loft. Everything in it should earn its place by answering a question someone might genuinely ask: what was assessed, when, by whom, and what was done about it. A document that answers no such question is not protecting you. It is in the way.

The things that clearly earn their place

Some records are load-bearing and should be kept as a matter of course, frozen and dated:

• Assessments and audits at their point of approval, each one a fixed record of the building's condition on a date.
• Statutory certificates and inspection records for the things on legal clocks, such as EICRs, which under the gov.uk landlord rules rented homes in England need at least every five years, and LOLER thorough examinations, every six months for passenger lifts.
• The fire risk assessment and its action history, since this is the document most often asked for first.
• Plans and drawings that reflect the building as it is, plus the superseded versions that show how it got there where layout history matters for safety.
• Records that establish responsibility, naming the responsible person or, for higher-risk buildings under the Building Safety Act 2022, the Accountable Person.

These are the spine of the golden thread, and there is no sensible case for letting them go.

The things you can let go, with judgement

Other material accumulates without ever earning its keep: duplicate copies of the same scan, working drafts that were superseded before approval, transient operational notes, records relating to buildings or contracts you have exited, and documents whose retention period has genuinely expired. We would not pretend the judgement is always easy, and we would be cautious, because the cost of wrongly discarding a load-bearing record is high and the cost of keeping a useless one is low. But "low" is not "zero". Every retained item that answers no question is friction on every future search. The honest discipline is to ask of each category whether it could plausibly be needed to demonstrate the building was managed properly, and to let go of what clearly could not, against a retention policy you have written down rather than by ad-hoc impulse.

Keep the version that matters, not every version

A particular source of bloat is version sprawl. The instinct to keep every draft of an evolving document produces archives where ten near-identical files compete and nobody is sure which is current. The better practice is to freeze the approved version, keep the superseded approved versions as a clear dated chain where their history matters, and discard the interim drafts that were never approved. This gives you a clean line of succession a reader can follow, rather than a thicket. We set out the freezing principle in point-in-time records: freezing an audit when it is approved; curating versions is the archive-side companion to it.

Retention is a policy, not a mood

The reason archives drift toward hoarding is that, in the absence of a policy, every individual decision to delete feels riskier than the decision to keep, so nothing is ever let go. The fix is to make the decision once, in a written retention policy, rather than repeatedly in the moment. The policy sets how long each category is held, what triggers its disposal, and what is kept permanently. With that in place, letting go of an expired record is following a rule, not making a nervous judgement call, and the archive stays curated by default rather than swelling by inertia. Where statute or contract sets a minimum retention, the policy simply adopts it; where neither does, the policy reflects how long the record could realistically be needed to show good management.

A curated archive is a trusted one

The payoff of curation is trust, both yours and a regulator's. When the archive holds only what earns its place, the document asked for is found quickly and you can be confident it is the current one. When it holds everything, every retrieval is a hunt and every hand-over carries the risk of passing on the wrong version. In our view the discipline of letting go, done carefully and against a written policy, is not the opposite of good record-keeping. It is part of it. The companion piece on retrieval, the document a regulator asks for first, makes the same point from the other end.

This is the balance SAMRISK is built to hold: a structured, dated record where the load-bearing documents live and freeze on approval, version chains stay clean, and what you keep is what you can actually use. You can see the shape on the audits and safety case pages. Keeping everything is not caution. It is deferral, and the bill comes due the moment someone asks you to find one thing in a hurry.