Custodians Without Succession: Cataloguing the Spreadsheets Your UK Firm Cannot Retire
Photo: Joe Haupt from USA, CC BY-SA 2.0, via Wikimedia Commons
Every organisation of a certain size has one. A workbook that produces the monthly commission run, or the price list the sales team quotes from, or the thirteen-week cash forecast the finance director trusts more than the ERP. It works. It has always worked. And the person who built it left in 2019.
The instinct on discovering such a file is to classify it as shadow IT and schedule its demolition. That instinct is usually wrong, or at least premature. A workbook that has survived three finance directors, two auditors and a systems migration is not a failure of governance. It is a working artefact that encodes decisions nobody wrote down. Before it is replaced, it needs to be understood — and understanding starts with a register.
Heritage, not hazard
The conservation framing is useful because it forces a different question. Instead of asking why does this still exist?, ask what does the business depend on it for, and who could rebuild it? Those are answerable questions with practical consequences. They also tend to surface a smaller, more alarming set of files than a general spreadsheet audit would.
Most organisations have thousands of workbooks. Perhaps forty matter. Of those, a handful are single points of failure in the literal sense: one person holds the tribal knowledge, there is no version history worth the name, macros perform the actual logic, and the file pulls from a shared drive folder that was decommissioned in 2023. That set of forty is what the register is for.
The shape of the register
A heritage register is deliberately modest. It is a single table — a spreadsheet will do, which is either appropriate or ironic depending on your disposition — with one row per file and columns covering purpose, owner, dependencies and a risk score. It is not an inventory of everything. It is a schedule of assets with a documented condition report.
The working fields that earn their place:
- File and location, including the folder path someone would actually need to find it
- Business purpose, in one sentence a non-technical director would recognise
- Custodian, the person who can currently explain it, and whether anyone else can
- Succession status, from "documented and handed over" to "one person, no notes"
- Consumed by, the reports, board packs or external filings that depend on the output
- Upstream dependencies, the source systems, extracts and manual inputs it assumes exist
- Risk score, derived from a short, repeatable rubric
Finding the forty, in a week
A one-week exercise is enough to get a first pass on the register, provided the scope is honest. Five working days, with roughly half a day of effort per day from a small team — a data steward, someone from finance, and a business analyst who knows where the bodies are buried.
Day one: collect candidates from humans, not tools. Automated discovery will find files. It will not find the ones that matter. Ask twenty people across finance, sales operations, commercial and supply chain one question: which spreadsheet would cause a real problem if it disappeared tomorrow? Expect duplicates and expect consensus to form quickly around a small number.
Day two: confirm by dependency. For each candidate, trace what it feeds. If the answer is "nothing outside the team", the risk is lower than it looks. If a commission payment, a statutory filing, a customer quote or a board figure depends on it, the file moves up. This is the day that separates genuine heritage assets from well-loved clutter.
Day three: score succession risk. A simple four-factor rubric works better than a sophisticated one nobody applies consistently.
| Factor | Low (1) | High (3) |
|---|---|---|
| Custodianship | Two or more people fluent | One person, no backup |
| Version control | Named, dated, consistent | "Final_v7_FINAL_use this one" |
| Embedded logic | Plain formulas, readable | Macros, VBA, external links |
| Upstream feeds | Live, documented connections | Manual extracts from retired systems |
Scores of ten or above go on the immediate handover list. Scores of six to nine get a note and a review date. Below six, record the file and move on.
Day four: write the handover note. One page per high-risk file, attached to the register row or stored alongside the workbook. It should cover what the file does, where its inputs come from, what the outputs feed, what would break it, and which quirks are deliberate. That last point matters more than any other. Half the value of a long-lived workbook is in the accidents that were later baked in — the manual adjustment for a customer with unusual terms, the rounding convention that reconciles to the ledger.
Day five: socialise and set a cadence. Circulate the register to the relevant directors. Name the custodians formally, which is often the first time anyone has done so. Diarise a quarterly review, and a mandatory update whenever a custodian leaves or changes role. A register that is not refreshed becomes a heritage problem of its own.
What conservation does not mean
Documenting a workbook is not the same as endorsing it. Some files in the register will be genuinely unfit: fragile, unauditable, or carrying control risk that an external auditor will eventually flag. Conservation is what allows that judgement to be made on evidence rather than instinct.
A file with a full handover note, a named secondary custodian and a traced dependency chain is a candidate for replacement with an agreed specification. The same file without those things is a candidate for replacement with a guess — and the guess is usually wrong in ways that only become visible when the new system produces a number that differs from the old one.
There is a further benefit, less discussed. Heritage registers tend to reveal that some spreadsheets are not artefacts at all but prototypes: a pricing model that has been refined over eight years is, in substance, a specification for the system the business should have bought. Reading it as a requirements document rather than a liability is often the fastest route to a business case.
Starting this week
The barrier to entry is low and the returns are asymmetric. Two people, five days, one shared table. The output is not a beautiful thing — it is a list of files with names and notes attached — but it converts an invisible dependency into a managed one.
Begin with the question rather than the tooling. Ask twenty people which spreadsheet they would miss. Then write down the answer, because the person who could have told you without looking may not be there next quarter.
