When to Replace a Spreadsheet with Software (and What It Costs)

Every business we have ever worked with has one. The spreadsheet that started as a list and became the system: the job tracker, the quote book, the vendor register, the commission calculator, the month-end workbook with eleven tabs that only one person understands. It is not a bad thing. A spreadsheet is the fastest way to build a system that did not exist, and the person who built it was solving a real problem. The question is when it stops being the right tool, and what to do then.

The tells that a spreadsheet has become a system

• More than one person edits it, and "who changed this?" is a question that comes up.
• There is a copy. Then there are three copies, named FINAL, FINAL2 and FINAL-use-this-one.
• Data gets typed into it from somewhere else: an email, a form, another system. Someone is the integration.
• It has a tab nobody opens because the formulas on it broke and nobody dared fix them.
• A decision waits on it. The truck cannot leave, the invoice cannot go out, the report cannot be sent until the spreadsheet is updated.
• The person who built it is the only one who can change it, and they have started to mention this in salary conversations.
• You have asked it to send an email, chase a date or notify someone, and it cannot.

Any two of those and the spreadsheet is doing the job of software. Four or more and it is doing that job badly, and the cost is already showing up somewhere you are not measuring: a late invoice, a missed renewal, an insurance certificate that lapsed, a quote sent with last year's price.

What it is actually costing you

The direct cost is easy to see once you look. Count the people who touch the spreadsheet in a week, estimate the hours, and multiply by fifty. A job tracker that four people spend forty-five minutes a day updating is about eight hundred hours a year, which is a fifth of a full-time salary spent moving information between places it already exists.

The indirect cost is larger and harder to see. Spreadsheets do not tell you when something is late; a person has to notice. They do not stop two people from booking the same crew, or from quoting a price that no longer exists. They cannot chase anyone. Every one of those gaps is a small loss that happens often, and the total is usually more than the direct hours.

There is also a cost that does not appear until it does: the spreadsheet as a single point of failure. When the person who understands it leaves, or the file is corrupted, or someone sorts one column without the others, the business finds out how much it had been leaning on a file.

When the spreadsheet should stay

Not every spreadsheet should become software, and a firm that tells you otherwise is selling. A spreadsheet is the right tool when the work is genuinely one person's, changes shape every time, happens a handful of times a year, or is a piece of analysis rather than a process. A model someone builds to answer a question, uses, and closes is exactly what spreadsheets are for. The ones worth replacing are the ones that run a repeating process with more than one person in it.

What the replacement should look like

Here is where most replacements go wrong. The instinct is to buy a platform: a CRM, a project tool, an ERP module. Sometimes that is right. Often it means bending your process to fit someone else's product, paying per seat forever, and discovering the one thing your spreadsheet did that the platform cannot. The other instinct is to rebuild the spreadsheet as a bigger spreadsheet, which postpones the problem.

The replacement that works is small, specific and connected. It has the columns your spreadsheet had, because those columns are the process, and it adds the four things a spreadsheet cannot do:

• One record, many people. Everyone sees the same current state, with a history of who changed what.
• Intake without re-keying. Forms, emails and other systems create records directly, so nobody is the integration.
• Rules that act. A date that is about to pass sends a reminder. A status that changes notifies the next person. A price that is stale cannot be quoted.
• Reporting from the records, not from a copy. The month-end view is the same data as the daily work, so it is never out of date and never reconciled by hand.

That is what we mean by an internal app or a tracker rebuilt: not a platform, a small system shaped exactly like the work, that plugs into the accounting, email and CRM you already have.

What it costs

This used to be the hard part. Custom software was a five-figure quote and a three-month wait, which is precisely why the spreadsheet got built instead. That has changed. Building software with agentic AI tools, under engineers who own the design and review every change, means a working replacement for a single spreadsheet-driven process is a matter of days, not months.

Our version: a scoping conversation, then a fixed price for a first release, then five days of build and it is live for one team, with the source code. Support and enhancement after that start at $999 a month. Set against eight hundred hours a year, or one lapsed insurance certificate, the arithmetic is not close. The longer version of this, with market ranges, is in how much AI consulting costs.

How to start

• Pick the spreadsheet that most people complain about, not the biggest one. Complaints are a measure of cost.
• Write down what it does today in ten lines: what goes in, who touches it, what comes out, what breaks. That is most of the scope.
• Decide the one thing the replacement must do that the spreadsheet cannot. Usually it is a reminder, a notification or a form. That is the first release.
• Keep the spreadsheet running alongside for two weeks. Kill it when nobody has opened it.

If you would rather recognise your spreadsheet than describe it, the Rebuild Library has an entry for most of them: the job tracker, the vendor register, the quote book, the month-end workbook. Each one sets out how it runs today, the tells, and what it looks like rebuilt.