Short answer: when several people work in it, it is rebuilt every period, someone else reviews it, the data comes from several systems, it breaks under its own size, and other companies have the same file. Four or more of those six usually means an application pays for itself. With one or two, a spreadsheet is the right tool and replacing it buys you nothing but cost and disruption.
Almost every company has one: it has been running for years, everyone knows it rattles, and nobody dares touch it. The question is not whether it could be nicer — it always could — but whether it is worth it. Six questions settle that.
The six criteria
Together they say more than any one of them alone. A single yes is no reason to change anything; four or more means your spreadsheet has already become an application, just without the protections that normally come with one.
| # | Question | When this is a “yes” | Why it counts |
|---|---|---|---|
| 1 | Do several people work in it? | Two or more, especially with people outside the company | Spreadsheets have no real concurrent editing. Every second copy is a source of discrepancies. |
| 2 | Is it rebuilt every period? | Daily, weekly, monthly or quarterly | Recurring work can be automated. One-off work cannot — and that is exactly what spreadsheets are good at. |
| 3 | Does anyone else review it? | Accountant, controller, client, auditor or regulator | The moment someone looks over your shoulder, you need to show who changed what and when. A spreadsheet cannot. |
| 4 | Does the data come from several systems? | Two or more sources, or sources plus manual entry | Retyping between systems is the largest error source and most of the work. |
| 5 | Does it break down somewhere? | Slowness, errors, people getting in each other's way, or no usable reporting | Without a concrete bottleneck there is no business case. Dissatisfaction alone is not enough. |
| 6 | Do other companies have this same file? | It is common in your industry | Then off-the-shelf software may already exist, and custom work is a waste of money. |
When it pays — and when it doesn't
Worth replacing
- three or more people work in the same file
- it takes hours every month just to fill
- wrong numbers have gone out at least once
- the data comes from two or more systems
- your accountant or a client reviews it
- whoever built it has left, or is about to
Leave it alone
- one user, who knows exactly how it works
- you use it a few times a year
- it is a one-off calculation or analysis
- its shape changes completely every time
- mainstream software exists for this process
- the bottleneck is a process, not the file
That last point is the one most often missed. A spreadsheet that rattles because nobody knows who is responsible for what does not improve with software. You would just be moving the confusion to a more expensive screen.
What it takes
Scope follows the number of sources that have to be connected, not the number of screens. Screens are the easy half. You get the price after an hour-long conversation, as a fixed amount with a fixed delivery date.
| Size | Characteristics | Lead time |
|---|---|---|
| Small | one source, one process, few rules | 3 weeks |
| Average | two sources, several users and roles | 5 weeks |
| Large | three or more sources, many rules, long history | 9 weeks |
The monthly fee is not a support contract on paper. It covers hosting, backups, monitoring, updates, and keeping up when your accounting package or shop changes an API — which happens a few times a year. Without that part you have the same problem again in two years, only a more expensive one. See how we work for exactly what is and isn't included.
Why it is rarely about carelessness
The usual assumption is that spreadsheet errors come from sloppiness. The research points elsewhere. Field audits find errors in 24 to 94% of business spreadsheets; a recent study put it at 94% of the files used in decision-making. And the most telling figure: people estimate their own error rate at 10 to 18%, while 86% had actually made one.
You cannot see your own mistakes. A mistyped formula looks exactly like a correct one. That is why “be more careful” is not a fix and making the mistake impossible is.
What you can do without replacing anything
- Separate input and calculation onto different sheets, and protect the calculation sheet.
- Use data validation on input fields — that catches half the errors.
- Replace long nested formulas with helper columns in short steps.
- Remove references to other workbooks; they break without warning.
- Save each period as its own file so you stop overwriting your history.
If that solves it, you have saved yourself the money. If it doesn't, you now know why.
Frequently asked
At how many rows does a spreadsheet become too big?
There is no hard limit. In practice it starts to hurt around 20,000 to 50,000 rows with formulas, or earlier when lookups run across several sheets. Slowness is usually not even the first problem — more often something breaks quietly: a formula not dragged down, a filter left in place.
Isn't a low-code tool a cheaper alternative?
Often yes, and you should try it first. Low-code platforms fit well when the process is simple and the data lives in one place. They get expensive per user at scale, and they struggle once several sources have to be tied together and the rules genuinely have to hold.
What if not all six criteria apply?
Four or more usually means an application pays for itself. Two or three often means a database with a reporting layer is enough — the data is the bottleneck, not the screen. One or two means a spreadsheet is the right tool.
Curious where your file stands?
Upload it and answer six questions. You get a verdict in under a minute — including the verdict that you should leave it exactly as it is.
Check your spreadsheet