The HR spreadsheet: exactly where it breaks, and at what headcount
Echipa HR 365 · reviewed 2026-09-02 · 5 min read
A spreadsheet does not become unsuitable at a particular headcount. It fails one point at a time, at six different places, each with its own threshold — and the first appears far earlier than anybody expects, at around 15–20 people, and has nothing to do with volume but with how many people write in the file.
It is not a volume problem
A spreadsheet handles thousands of rows effortlessly. What it does not handle is people: several writing at once, versions circulating by email, different access rights on the same data, and history — who changed what and when. All six points below are variations of those four gaps.
| # | Breaking point | Typical threshold | The sign you have reached it |
|---|---|---|---|
| 1 | Several people write in the same file | 2 people | There is a “final version 3” on email |
| 2 | Sensitive data sits next to ordinary data | 15–20 employees | You cannot send the file to anybody without deleting columns |
| 3 | Leave records with individual entitlements | 25–30 employees | The formulas have exceptions hand-written for particular people |
| 4 | Employees want to see their own data | 30+ employees | HR answers the same questions every month |
| 5 | You need history: who changed what and when | the moment the first dispute arises | A discussion is settled from memory, not from records |
| 6 | The same data is needed in two places | 40+ employees | You copy by hand from the personnel file into the timesheet |
The first threshold is earlier than you think
The moment a second person starts writing in the file is the moment parallel versions appear. It does not matter whether you have ten people or a hundred. Cloud syncing partly solves it — but it does not answer “who changed this figure and why”, which is the fifth break and the most expensive.
What you do at each break, without changing everything
You do not have to move to a system at the first sign. Each break has a proportionate answer, and some can be postponed for years:
- One file with one declared owner. Everyone else requests changes rather than making them. Cheap and effective up to about 30 people.
- Separate the sensitive data into its own file, with restricted access. It takes an hour and removes most of the risk.
- Take the exceptions out of the formulas and put them in their own columns. Formulas with exceptions written into them break at the first person who touches them.
- Send each person an extract of their own data, once a year. It reduces the questions and catches errors early.
- Here a spreadsheet has no answer. Change history is the one break that cannot be repaired by discipline.
- Nor here. Duplicated data guarantees divergence, and manual syncing fails in exactly the busy months.
When it is worth moving
The practical rule: once you have hit breaks 5 or 6, any further effort put into the spreadsheet is spent on a problem that cannot be solved there. Until then, a well-structured file with a clear owner is a perfectly respectable solution — and far better than a system bought and unused.
How you prepare the move, wherever you go
- HR — cleans the data first: one row per person, columns in a consistent format, no merged cells
- HR — sets the identity key — a unique code per person, not the name, which gets spelled differently every time
- HR — fixes balances and states in writing at an anchor date; what came before stays history, what comes after enters the system
- HR — keeps the old file unchanged, as a reference, and stops updating it in parallel
The last step is the most often broken and the only one that guarantees failure: if the file keeps being updated in parallel “just in case”, within three months you have two sources that no longer match and nobody trusts either.
The signs you crossed the threshold without noticing
The breaks do not announce themselves. They show up as extra work that feels normal, because it grew gradually. Check yours:
- You spend more than a day a month just compiling reports from different files.
- There is at least one formula nobody understands any more and everybody avoids.
- When somebody asks how many leave days they have left, you check in two places to be sure.
- You have a file called “to check” or “corrections” that you have been postponing for over a month.
- At the last dispute, the answer was reconstructed from emails, not from the records.
- You are reluctant to hand the file to somebody while you are on leave — so you never quite go on leave.
Three ticks out of six means you have already passed breaks 4 and 5. The last point is the best indicator of them all: a record-keeping system you cannot hand over for two weeks is not a system, it is a personal skill — and that is a risk for the company, not merely an inconvenience for you.
Does a cloud spreadsheet solve the versions problem?
For the first break, yes — syncing and cell history help a lot. For the second, the fifth and the sixth, no: fine-grained rights per column, a structured trace of changes and the link between modules stay outside what a spreadsheet can do.
How long does the move take?
The technical part, hours. The real part — cleaning the data and fixing the balances — one to two weeks of one person’s work, and that is where all the effort goes. A vendor who says “we will import everything” has not seen the file yet.
Spreadsheet import, with identity on a unique code and balances fixed at an anchor date
The data you already have comes in as it is, with leave balances taken from the date you choose — so nothing gets re-entered by hand.
You can create an account in a few minutes and use every module for 7 days, no card required.