When should you replace a spreadsheet with software?

The honest answer is later than most suppliers will tell you, and for one of five specific reasons rather than a general feeling that the spreadsheet has got out of hand.

A spreadsheet that works is not a problem. It is cheap, everybody understands it, and it changes the day the business changes its mind. Replacing one that is doing its job is how businesses end up paying forty thousand pounds for something slower than what they had.

So here are the five triggers that genuinely justify it, the three that look like triggers and are not, and what to do first either way.

The five reasons that justify it#

1. Two people need it at the same time, and they already do. Not "might one day". If two people are editing copies and reconciling by hand, or one person is waiting for another to close the file, the spreadsheet has already failed and the only question is how long you carry it.

2. A mistake in it has cost real money. An invoice sent at the wrong price, a stock figure that caused an order, a discount typed with an extra zero. Once that has happened, you are paying for the absence of rules, and that charge recurs.

3. One person is the only one who understands it. If that person being away for a fortnight is a problem, the spreadsheet is not your risk. The knowledge is. This is the trigger people recognise last and regret most.

4. Somebody outside will ask to see the history. An auditor, a regulator, an insurer, a buyer doing diligence. A spreadsheet cannot tell you who changed a figure or when. If you will need that answer, you need it built in beforehand, because it cannot be added to the past.

5. A customer sees it. The moment a spreadsheet's output goes to somebody who is paying you, its errors stop being internal.

One of those is enough. If none of them is true, read the next section before you spend anything.

The three that are not reasons#

"It looks unprofessional." Nobody outside your business can see it, and the ones who can are not judging you. This is the most expensive motive on the list because it has no end state: there is no version of the software that will finally feel professional enough.

"It has got really big." Size is not the problem. Forty thousand rows and sixty columns is fine if one person uses it and the formulas are sound. Big and working beats small and half-replaced.

"A supplier showed us a platform that does this." It probably does, and it probably does forty other things you will pay for and never use. That is not an argument against off-the-shelf, it is an argument for checking which of the five triggers it actually removes.

Off the shelf or built#

Worth answering properly rather than by instinct, because the honest answer is often the cheaper one.

Buy something when what you are doing is a standard job that thousands of businesses do identically: payroll, accounts, email, card payments. You will never build a better one and you should not try.

Build something when the thing the spreadsheet does is the thing that makes your business different from the one down the road. That logic is usually why you are competitive, and it is exactly the part a standard package will make you abandon.

The common case is both: buy the accounts package, build the small thing that feeds it. That is almost always cheaper than forcing one tool to do both.

What a package will not tell you#

The subscription is the small number. The cost that does not appear in the quote is the work of changing how your business operates to match how the software expects it to, and the ongoing cost of the parts that no longer quite fit.

Sometimes that is a fair trade and worth saying yes to. The point is that it is a cost, and it belongs in the comparison.

The cheapest honest first step#

Before you commission anything, pay for a short piece of work that produces three things in writing: which of the five triggers actually applies to you, what the spreadsheet's formulas really do, and the smallest thing that would remove the trigger.

That should cost a few hundred pounds rather than a few thousand, it should be yours to take to any supplier, and it should be allowed to conclude that you do not need software at all.

If a supplier will not do that piece of work without committing you to the build, that tells you something useful about the rest of the relationship.

What we do#

No minimum. "Look at this spreadsheet and tell me whether I actually need software" is ordinary work here, not a favour at the edges of a bigger project.

You send the file and say what goes wrong. It gets graded by size, you get the price in writing first, and nothing starts until you approve it. The written assessment is yours either way, including to take elsewhere.

We will tell you when the answer is that the spreadsheet is fine, or that one hour of fixing the formulas is the whole job. Thirty years in, that is a common enough conclusion to be worth saying out loud, and telling you is the only thing that makes the other answer believable.

The price changes with the size of the job. The standard never does.