Observable condition
Almost every business begins with spreadsheets, and for good reason. They are quick to set up, familiar to everyone and endlessly adaptable. A new requirement appears on Tuesday, and by Wednesday there is a new column, a new tab or a new formula. That flexibility is what lets a small operation move faster than software could.
Years later, the same spreadsheet runs quotes, tracks jobs and feeds the monthly figures. It has fourteen tabs, formulas that nobody dares touch and a naming scheme only its author understands. When the author is on leave, work slows. When someone sorts a column the wrong way, numbers shift without anyone noticing until a customer queries an invoice.
An illustrative example: a Port Elizabeth fleet-maintenance company schedules every service and repair in one workbook. It started as a simple list. Today it holds vehicle details, technician allocations, parts costs and invoicing notes, with colour codes whose meaning exists only in the memory of the workshop controller. A copy saved to a laptop during a power cut becomes, for a day, a second version of the truth.
Realisation
The realisation is that the quality that made the spreadsheet valuable is the same quality that makes it brittle. Flexibility means anyone can change anything at any time, with no enforced structure and no record of why. A tool that cannot be misused by mistake is also a tool that cannot protect you from mistakes.
Microsoft's own documentation on the limits and specifications of Excel is a useful reminder that a workbook is bounded in size, in the number of simultaneous users it can support well and in what it can enforce. Its co-authoring guidance explains how several people can edit at once, but it does not turn a workbook into a system with rules, roles and an audit trail. Those are properties of purpose-built systems.
Realisation prompt: choose the spreadsheet you would least like to lose. If its author left tomorrow, who could explain what every tab and formula does?
Identification
This is a pattern of useful flexibility turning into operational fragility. The visible symptom is a critical workbook that few people can safely change. The common assumption is that staff should be more careful or that the file needs a tidy-up. The structural cause is that a general-purpose tool has been asked to act as a controlled system, and nothing in it enforces the rules the business now depends on.
The Insight when a spreadsheet is a useful tool and when it becomes infrastructure sets out the distinction, and the Pattern when a spreadsheet becomes a business-critical system describes the moment of transition. This Pattern concentrates on why the same flexibility that helped early on creates risk later.
Diagnosis
- Nothing enforces structure. Dates can be typed three ways, a customer name can be spelled five, and a number can be entered as text. Reports that depend on consistency quietly produce wrong totals.
- Logic is hidden in formulas. Business rules such as discount tiers, rounding and exceptions live inside cell formulas, some copied down and some overwritten by hand, with no documentation of intent.
- Versions multiply. Copies are emailed, saved locally or renamed for a month-end. The Pattern when nobody knows which spreadsheet is the latest shows the confusion this causes.
- Access is all or nothing. Anyone who can open the file can usually change any part of it, including formulas and historical entries, and the change leaves little trace.
- One person is the system. The workbook's author knows its quirks. Their knowledge is the real control, and it is not written down.
- It does not connect. Data leaves and enters by copy and paste, so every hand-over is a chance for a slip. The Pattern why staff keep a spreadsheet between two business systems describes the bridging role these files play.
Commercial impact
The commercial value at stake is accuracy of billing and reporting, continuity of operations, staff capacity and the ability to grow without adding risk.
The first cost is error. Spreadsheet mistakes are seldom dramatic; they are small and compounding: a wrong rate carried into a quote, a missed row in a total, a formula range that stops one line short. Each can cost money, and each is hard to detect because the file looks fine.
The second cost is continuity. If the person who understands the workbook is absent, the business slows or stops. If the file is corrupted or overwritten, the loss can be severe, particularly where backups are informal. In a country where power interruptions can end a session without a save, the risk is not theoretical.
The third cost is scale. As volume grows, the work of maintaining and checking the workbook grows with it, and the people who maintain it are often the most capable operators. The Pattern why your most important process is running in Excel looks at what that dependence costs over time. The fourth cost is governance: when auditors, banks or partners ask how a figure was produced, the answer is a file, not a controlled process.
Common misidentification
The common response is to add more protection to the spreadsheet itself: lock cells, add macros, colour-code more, write a manual. Some of this helps, but each addition increases complexity and makes the file harder for anyone but its author to maintain. Another response is to ban spreadsheets, which usually pushes them underground, where they are less visible and no less used.
A third response is to rush into a large software replacement. Without first understanding what the spreadsheet actually does, including its unwritten rules, the replacement either omits needed behaviour or reproduces the old file's untidiness in a more expensive form.
Possibility
In a better state, spreadsheets remain a valued tool for analysis, drafts and one-off work, but not the home of any process the business cannot afford to get wrong. Core records live in a system that validates input, controls who can change what and keeps a history. The logic is written down, and more than one person understands it.
That does not require abandoning flexibility. It requires putting it where it is safe: exploration happens in spreadsheets, and repeatable operations happen in controlled structures.
Intervention
- List your critical workbooks. Ask each team which spreadsheets they could not work without. Record the owner, purpose and who depends on the output.
- Rank by consequence. Give priority to files that feed invoices, payments, compliance returns or customer commitments.
- Document the logic. For the top files, write down what each tab is for, which formulas encode business rules and which cells are inputs.
- Introduce basic controls now. Use a single shared location with version history, restrict editing of formulas and add simple checks such as totals that must reconcile.
- Separate input from calculation. Make the entry area structured and validated, and keep calculations in protected sheets.
- Plan the migration of the riskiest process. Move one process at a time to a system that enforces rules, using the documented logic as the specification.
Diagnostic questions
- Which workbook, if it disappeared tonight, would hurt most on Monday?
- Can anyone other than the author explain how its key formulas work?
- How would you know if someone changed a historical figure?
- How many copies of the file exist, and which one is correct?
- How much time each month is spent checking the workbook rather than using it?
Bottom line
Spreadsheets earn their place because they bend to fit the problem. The same bending is what allows errors, hidden rules and dependence on one person to build up unnoticed. The sensible position is not to condemn them, but to know which ones have quietly become infrastructure and to give those the structure, ownership and controls that infrastructure needs.
Sources and further reading
- Excel specifications and limits - Microsoft Support (last verified 10 October 2026)
- Collaborate on Excel workbooks at the same time with co-authoring - Microsoft Support (last verified 10 October 2026)
- What is Data Integrity? - IBM (last verified 10 October 2026)
Related Mellorca reading
- The Hidden Cost of Automating a Bad Process With AI
- The Hidden Cost of Re-entering the Same Information
- The Hidden Cost of Signing In to Every System Separately