Safeguarding as a daily rhythm, not a monthly scramble

Payment and e-money institutions across the EU are expected to keep customer funds separated from their own and to be able to demonstrate, at any point, that the safeguarded balance covers the corresponding obligation. The precise wording varies by member state and by licence; the operational reality does not.

What varies enormously is how teams organise around it. The pattern we see most often is a monthly exercise, assembled from exports, evidenced by a spreadsheet and a signature. It satisfies the reporting requirement and it does very little to control the risk.

Why monthly is structurally too slow

A shortfall is not usually a single dramatic event. It is a small mechanical divergence that persists — a fee debited from the safeguarding account that should have hit the operating account, a customer refund funded from the wrong side, a sweep that ran on a stale figure.

Each of those is trivial to correct on the day it happens, when the transaction is recent and someone remembers the context. Discovered five weeks later, the same item requires reconstructing what happened, and by then there may be four more like it. The cost of correction rises steeply with age, and so does the difficulty of explaining it.

What a daily rhythm actually requires

Daily attestation sounds heavier than monthly. Done properly it is lighter, because the volume per run is small and almost all of it is mechanical.

1. 01Ingest yesterday's statements for every safeguarding and operating account, plus the ledger position, before the working day starts.
2. 02Reconcile the safeguarded balance against the customer obligation as computed from the ledger, not as reported by a separate system.
3. 03List every reconciling item with its cause, its age and its owner. Timing items are expected; anything else is not.
4. 04Compute the resulting surplus or shortfall and compare it to the previous day. A stable surplus is fine. A drifting one is the signal.
5. 05Have a named person attest, and record who and when as part of the run rather than in a separate file.

Steps one to four should be automatic. Step five is the only one that requires a human, and it should take a few minutes on a normal day.

The reconciling items are the whole point

A safeguarding reconciliation that comes out flat every day is not reassuring — it is suspicious. Real operations have timing differences: a collection received but not yet swept, a payout instructed but not yet cleared. Those are legitimate and expected.

What matters is that every item on the list has a cause and an age, and that the list does not accumulate. Three specific patterns are worth alerting on independently.

• An item ageing past its channel's clearing window — timing that has stopped being timing.
• A reconciling item recurring with the same cause on consecutive days, which indicates a process defect rather than a transaction.
• A surplus trending downward, even while remaining positive. The direction is the signal, and waiting for the sign to change discards the warning.

Evidence as a by-product

The reason monthly processes persist is that daily ones seem to multiply the evidence burden by thirty. That is only true if evidence is produced by hand.

If the attestation is generated from the reconciliation run — the balances as reconciled, the reconciling items as classified, the approver as recorded — then thirty daily attestations cost no more to produce than one monthly one, and they are considerably better evidence. Each carries the state of the world on the day it was signed, rather than a reconstruction assembled afterwards.

An auditor asking what you knew on 14 March is asking a question a monthly process cannot answer, because on 14 March nobody had looked.

A reasonable target

For a team moving from monthly to daily, the milestone worth aiming at is: the safeguarding position is reconciled and attested before ten in the morning, every working day, with a reconciling-items list short enough to read in full.

That is achievable for most operations, and it changes the character of the control entirely. The question stops being whether you can produce the report, and becomes whether anything on today's list is unusual — which is a question a person can actually answer well.