Spreadsheets are not the problem. They are the best tool ever built for working something out. The problem is when a spreadsheet stops being a tool for thinking and becomes the system a business runs on.
Here are the specific symptoms that mean the problem has become structural, and the arithmetic for working out what it costs you.
The twelve signs
Two of these is a nudge. Four is a business case.
1. The same number lives in three places and they disagree
Stock in a spreadsheet, in the accounting system, and in someone’s head. Somebody reconciles them weekly. Everyone quietly distrusts all three.
Why it matters: decisions get made on numbers nobody believes. Eventually people stop looking them up at all.
2. One spreadsheet runs a critical process and one person understands it
It has nested formulas, hidden columns and a macro written in 2019. When that person is on leave, the process stops.
Why it matters: this is the single largest unmanaged risk in most small businesses, and it is almost never on anyone’s risk register.
3. Month-end takes more than a couple of days
Mostly spent assembling data by hand — exporting, pasting, cross-checking, fixing.
Why it matters: that is expensive skilled time doing clerical work, every month, for ever.
4. Staff retype data between systems
Order details typed into accounting. Delivery addresses copied into the courier portal. Stock counts entered twice.
Why it matters: every retype is an error waiting to be discovered three weeks later by a customer.
5. You cannot answer a simple question quickly
“What is our actual margin on this product line this quarter?” should take seconds. If it takes a day and a half, your data is not a system — it is an archive.
6. Growth means hiring administrators, not makers
Headcount rising faster than output, with the new roles being coordination rather than production.
Why it matters: this is a systems failure presenting as a staffing need, and hiring is the expensive way to not fix it.
7. Nobody knows which version is current
Stock_v4_FINAL_updated_JK.xlsx and Stock_v4_FINAL_(2).xlsx both exist. Both have been edited this week.
8. There is no history
Something changed. Nobody knows who, when, or what it was before. Spreadsheets have no meaningful audit trail, which matters when a customer disputes a price or an auditor asks a question.
9. Permissions are all-or-nothing
Either someone can see the whole file or none of it. There is no “can update stock but cannot see margins”. So everyone sees everything, including salaries in the tab nobody remembered.
10. You are working around your accounting software
Xero and Sage are good at accounting. They are not stock systems, job costing systems or CRMs. If you are maintaining parallel records because the accounting package cannot represent your operation, that gap is the system you are missing.
11. Two people can overwrite each other
Shared files where simultaneous edits cause silent data loss. Cloud spreadsheets mitigate this, not completely, and not for anything with formulas pointing at other files.
12. The business stopped doing something because the admin was too painful
The most expensive sign, and the quietest. You turned down a customer type, dropped a product line, or stopped offering a service because tracking it was too hard. That is your systems constraining your strategy.
What it actually costs
Most people underestimate this by a wide margin because the cost is spread across many people in small amounts.
Work it out
| Measure | How |
|---|---|
| Admin hours per week | Everyone, across the business, on data entry, reconciliation and chasing |
| Fully loaded hourly cost | Salary × 1.3 ÷ 1,900 hours. £28,000 salary ≈ £19/hour |
| Annual labour cost | hours × 52 × hourly cost |
| Error cost | Errors per month × average cost to fix × 12 |
| Decision delay | Harder to quantify. Note where slow information cost you money. |
Worked example — 15-person wholesaler:
| Data entry and reconciliation across the team | 22 hrs/week |
| Fully loaded hourly cost | £19 |
| Annual labour cost | £21,736 |
| Stock/pricing errors | ~4/month at ~£180 each |
| Annual error cost | £8,640 |
| Total | ~£30,400 per year |
Against a £45,000 operational layer with £5,000 a year running: payback in under two years, and the risk in sign 2 disappears entirely.
Most people are surprised by this number. Admin time is spread across many people in small amounts, which is exactly why it goes unnoticed and unbudgeted for years. Run the calculation before you decide anything — it usually makes the case on its own.
What you probably need — and it is less than you think
Most businesses at this point do not need a full ERP.
They need:
- The accounting package they already run — untouched. Xero and Sage are good at accounting and HMRC-compliant. Keep them.
- Two or three integrations so systems stop disagreeing
- One operational system for the thing no product covers — stock, jobs, scheduling, whatever the critical spreadsheet does
Typically £25,000–£60,000, not £150,000. This hybrid is usually the right answer and it is rarely proposed, because it is the smallest invoice available.
Where spreadsheets fail at scale, technically
The specific mechanisms, so the risk is concrete rather than vague.
Silent calculation errors. A formula dragged one row short. A range that did not expand when rows were inserted. No warning, no error, just a number that is quietly wrong. Research into spreadsheet error rates has consistently found errors in a large majority of operational spreadsheets examined — and the ones that matter are the ones nobody spots.
No referential integrity. Delete a product row and the orders referencing it still exist, pointing at nothing. A database refuses this; a spreadsheet does it without comment.
No transactions. Updating stock and recording a movement are two edits. If someone closes the file between them, the two disagree and nothing flags it.
No concurrency control. Two people open the file. The second to save overwrites the first. Cloud spreadsheets improve this and do not solve it for formulas referencing other files.
No audit trail worth the name. Version history tells you a cell changed. It does not tell you why, or what the business reason was, or who authorised it.
Permissions are all or nothing. You cannot grant “update stock but do not show margin”. So everyone sees everything, including the tab somebody forgot about.
Scale limits that arrive unannounced. Performance degrades long before any row limit. A file that takes forty seconds to open is a file people avoid updating promptly.
What to do first, before buying anything
1. Count the hours. One week, everyone, honestly. This single exercise changes most people’s view of the problem.
2. Find the critical spreadsheet. The one where if it were lost tomorrow, the business would struggle. Back it up properly today, regardless of what else you do.
3. Write down what each one does. Not how it does it. What question it answers, who uses it, how often. This is the start of a specification and it costs nothing.
4. Identify the single biggest pain. Not everything. The one thing that, if fixed, would make the most difference.
5. Check whether a product already does it. Genuinely check — trial two with your real data. Many operational problems are solved by a £40/month tool and an integration.
6. Only then consider a build, and start with the biggest pain rather than everything at once.
Why it happens to good businesses
Nothing on the list above is a sign of poor management. The pattern is almost universal and it follows from how businesses actually grow.
A spreadsheet is the correct first answer. When a process is new and you are still working out how it should run, a spreadsheet is faster, cheaper and more flexible than any system. Building software around an unsettled process means building it twice.
It works, so it stays. The process settles. The spreadsheet keeps doing the job, so nobody revisits it.
It accretes. A column here, a tab there, a lookup to another file, a macro to save someone fifteen minutes. Each addition is sensible in isolation.
The business grows around it. More people, more volume, more edge cases. The spreadsheet still works, just slightly less well each quarter — which is exactly why nobody notices the threshold being crossed.
Then one of two things happens. Either someone leaves and the risk becomes visible, or a customer or auditor asks a question nobody can answer quickly.
The useful conclusion: the question is not whether you made a mistake. It is whether you have crossed the threshold where the tool costs more than it saves — and that is an arithmetic question, not a judgement about anyone.
When spreadsheets are still the right answer
- The process is still changing. Building software around an unsettled process means building it twice.
- It runs a handful of times a month. The arithmetic does not work.
- It is genuinely a modelling or analysis task. Spreadsheets are excellent at this and always will be.
- One person uses it and understands it, and nothing depends on it.
Not every spreadsheet is a problem. The ones that are tend to be shared, critical, and running something the business depends on.
What replacing them actually looks like
Not a big-bang replacement. The approach that works is narrower and faster.
Start with the one that scares you. The critical spreadsheet with the nested formulas that one person understands. Not because it is the biggest, but because it is the largest unmanaged risk in the business.
Keep the flexibility people liked. The commonest failure in replacing a spreadsheet is building something more rigid. People used the spreadsheet because they could adjust it. A replacement that forces a rigid process gets abandoned, and everyone quietly goes back to the file.
What that means practically: keep the ability to override with a reason recorded, allow bulk editing where it genuinely helps, support exporting to a spreadsheet for ad-hoc analysis. You are replacing it as a system of record, not as a thinking tool.
Add what a spreadsheet cannot do. Permissions, so not everyone sees everything. History, so changes are attributable. Validation, so impossible values cannot be entered. Simultaneous access without overwriting. These are the actual reasons for the project.
Integrate with accounting rather than replacing it. Xero and Sage are MTD-compliant and maintained for you. The operational layer sits above them.
Run in parallel until people trust it. Both for a few weeks, numbers compared. The old file is not deleted on launch day.
The risk nobody puts on the register
Sign 2 on the list above — one spreadsheet, one person who understands it — deserves separate treatment, because it is the one that can actually stop a business.
What it looks like when it goes wrong: that person leaves, or is ill for three weeks. The formulas are undocumented. Nobody knows which cells are inputs and which are calculated. The macro references a file path that no longer exists. Orders carry on arriving.
What it costs: businesses have lost weeks of trading to this. Reconstructing an undocumented spreadsheet that encodes a decade of business rules is genuinely difficult, and it is being done under pressure by people who do not know what the right answers look like.
Three things to do this week, none of which require a project:
- Back it up somewhere the owner cannot delete. Versioned, automatic, off their machine.
- Have them write down what it does. Not the formulas — what question each section answers and where the inputs come from. Two pages.
- Have someone else run it once, with them watching. The gaps surface immediately.
That is a morning’s work and it removes most of the immediate risk, whether or not you ever build anything.
A worked example
A 28-person distributor. Four spreadsheets running the operation alongside Xero.
What was happening: stock in a shared workbook updated twice daily. Orders typed into Xero by hand. A pricing file with customer-specific discounts that three people maintained. A job tracker that one person had built and understood.
The symptoms, mapped to the list above: signs 1, 2, 3, 4, 5 and 9. Six of twelve.
The arithmetic:
| Order entry into Xero | 6 hrs/week |
| Stock reconciliation | 5 hrs/week |
| Pricing queries and corrections | 4 hrs/week |
| Chasing job status | 3 hrs/week |
| Month-end assembly | 14 hrs/month ≈ 3.2 hrs/week |
| Total | 21.2 hrs/week |
| At £19/hour fully loaded | £20,900/year |
| Pricing errors: ~5/month at ~£140 | £8,400/year |
| Annual cost | ~£29,300 |
What they did: kept Xero. Built an operational system for stock, orders and job tracking, integrated with Xero so orders flow automatically and invoices come back. Pricing rules encoded with history and permissions.
Cost: £42,000, 11 weeks. Running cost about £3,600/year.
Payback on labour and errors alone: roughly 19 months. The key-person risk on the job tracker disappeared on day one, and that was not in the arithmetic at all.
What they did not do: replace Xero, build a CRM, or attempt everything at once.
Frequently asked questions
When should a business replace spreadsheets with software?
When the same data lives in several places and disagrees, when one person is the only one who understands a critical file, when month-end involves days of manual assembly, or when staff retype data between systems. Two of those is a nudge; four is a business case.
Have I outgrown Xero?
Probably not — Xero is good at accounting and you likely still need it. What you may have outgrown is using it for things it was never designed for, like stock, job costing or CRM. The usual answer is to keep Xero and build the operational layer around it.
How much does it cost to replace a spreadsheet system?
An operational layer over existing accounting typically runs £25,000–£60,000. A full custom ERP runs £60,000–£150,000. Calculate your current annual cost first — if it is under about £10,000 a year, a build will not repay.
What is the risk of running a business on spreadsheets?
The largest is key-person dependency: one person understanding a critical file. Beyond that, no audit trail, no granular permissions, silent data loss from simultaneous edits, and decisions made on numbers nobody trusts.
Do I need a full ERP?
Usually not. Most businesses at this stage need their existing accounting package, two or three integrations, and one operational system for the thing no product covers.
How do I calculate what spreadsheets cost me?
Count admin hours across the business for one week, multiply by 52 and by a fully loaded hourly cost (salary × 1.3 ÷ 1,900). Add error frequency × cost to fix × 12.
Written by Jas Singh, Founder. Last updated 8 October 2026.
