The Single Point of Knowing: A Sixty-Minute Rehearsal for UK Data Teams

In plenty of UK mid-market organisations, analytics continuity rests on a single person. The arrangement is rarely deliberate. It accumulates: one analyst learns the quirks of the finance extract, remembers why the customer file needs a manual adjustment, and knows that the Monday refresh fails when the upstream team runs its month-end job late. The risk is obvious but easy to defer. A resignation makes it urgent.

A better response is a structured rehearsal. It does not require a new platform, a lengthy documentation programme or a consultant. It requires one hour, one critical routine and a substitute who is willing to try. The purpose is not to produce a perfect runbook. It is to test whether the organisation can keep producing a trusted number when the usual person is unavailable.

Why a rehearsal beats a document

Most handover documents are written after the fact, often by the person who is leaving, and often under time pressure. They tend to record the steps that are easy to describe and omit the judgement that matters. A rehearsal reverses the order. It exposes the gaps while the expert is still present and while the substitute can ask awkward questions.

This is a test of operational resilience, not a performance review. The analyst's knowledge is an asset. The exercise should make that visible, not penalise them for holding it.

Set the terms before the clock starts

Choose one routine that matters to the business: the Monday refresh, the month-end pack, a regulatory return or a weekly trading dashboard. Do not attempt the whole data estate. Pick the routine whose failure would be noticed first.

Invite the analyst and one substitute with adjacent skills. The substitute might be a BI developer, a finance analyst or a data engineer. They do not need to be the long-term successor. They need to be curious and willing to drive.

Agree the success criteria in advance. By the end of the hour, the substitute should be able to explain where the number comes from, reproduce the routine with support, and name the main caveats. Full independence is not the goal. Visibility is.

Set three ground rules. First, the analyst should not say 'it is obvious' without explaining why. Second, the substitute should do the keyboard work wherever possible. Third, the group should capture decisions and dependencies, not attempt to fix everything in the hour.

Minute 0-10: Name the routine and its consumers

Write the routine on a whiteboard. Record its trigger, its output, the people who consume it and the deadline it must meet. Note the last time it failed and what happened. This frames the hour around business consequence rather than technical detail.

Ask the analyst a simple question: if you were away for two weeks, what would break first? The answer often reveals the real dependency. It may be a query stored in a personal folder, a licence tied to one account or a reconciliation step that exists only in someone's memory.

Minute 10-25: Walk the path backwards

Start from the delivered number and work backwards. Ask the substitute to trace the output to its source. The analyst should answer questions rather than deliver a lecture. The substitute should locate the query, identify the transformation steps and note any manual interventions.

Capture the locations of critical artefacts: the query, the spreadsheet, the scheduled job, the source file and the validation check. Record the caveats in plain language. For example: the customer count excludes staff accounts; the revenue figure is gross of a credit note that arrives three days later; the regional split uses a mapping that is updated only quarterly.

This is also the point to identify access dependencies. Note who owns the credentials, licences and permissions. Do not share secrets in the session. Record the owner and the route to obtain access.

Minute 25-40: Force the judgement into the open

Technical steps are only half the handover. The more valuable knowledge is diagnostic. Ask the analyst: what would you check if the number looked wrong? Which exception has no written rule? What would you do if the upstream file arrived late or the schema changed without warning?

Record the answers as decision rules. These become the basis for a short validation checklist. They also reveal where automation would reduce risk and where a second pair of eyes is genuinely required.

Minute 40-55: Let the substitute drive

The substitute should now execute or simulate the routine while the analyst stays silent unless there is a safety issue. Observe where the substitute hesitates. Mark each hesitation as amber or red. Amber means a question or a check is needed. Red means the substitute cannot proceed without the analyst's direct intervention.

Introduce one controlled failure. Make the source file late, remove a column or change a filter. Can the substitute diagnose the problem from the artefacts and the checklist? This is where tacit knowledge usually shows itself. The analyst may instinctively know that a particular warning is benign. The substitute needs a way to reach the same conclusion.

Minute 55-60: Agree the minimum viable handover

Close with a short list of actions. Limit it to five items: query locations, caveats, refresh dependencies, escalation contacts and validation checks. Assign an owner and a date to each. Decide what does not need documenting. A rehearsal that produces a forty-page manual has probably missed the point.

Schedule a second rehearsal with a different substitute in the next quarter. Repeat the exercise for the second most critical routine. Report the red flags to the executive sponsor and add them to the operational risk register.

What good looks like

A successful hour does not prove that the organisation is safe. It proves that the dependency is visible. Good signs include a substitute who can reproduce the number with support, explain the main caveat and know when to stop and escalate. Poor signs include an analyst who interrupts constantly, a substitute who cannot locate the query, or caveats that exist only in conversation.

Keep the exercise humane. The analyst is not the problem; the concentration of knowledge is. Use the findings to justify pairing, documentation, access redesign or automation. In 2026, better tooling can help, but it will not replace the context that sits with the person who has run the routine for years. The hour is a way to move that context into the organisation before circumstances force the issue.