When a Spreadsheet Should Become a Web App

Almost every small business we talk to is running something important in a spreadsheet. Scheduling. Inventory. Job tracking. Who owes what. Which customer said yes and when to chase the ones who did not.

That is not a failure. A spreadsheet is the fastest way ever invented to turn an idea into a working process. It costs nothing, it needs no meeting, and the person who knows the job can build it themselves on a Tuesday afternoon. Half the software in the world should have stayed a spreadsheet.

But a spreadsheet is a tool for one person thinking. The moment it becomes a tool for several people doing, it starts working against you quietly, and the cost of that rarely shows up as a line on an invoice. So the useful question is not "is a spreadsheet unprofessional?" It is "has this one crossed over?"

Six signs it has crossed over

1. Two people can be right at the same time. Someone has a copy on their desktop. Someone else is in the shared version. Both are accurate as far as their owner knows, and they disagree. If your team has ever had the conversation that starts "which one is the current one", the file has outgrown the format.

2. You cannot give someone part of it. A contractor needs to see their own jobs. A customer wants to check their own balance. A new hire needs to enter data without seeing payroll. A spreadsheet is mostly all-or-nothing: you either share the file or you do not. Hiding a column is not security, it is a suggestion.

3. There is a rule that only lives in someone's head. "Don't put anything in column G until the deposit clears." "If it says pending, check the email first." Those rules are real business logic. In a spreadsheet they are enforced by memory and goodwill. Every new person breaks them at least once, and you find out later.

4. The same number gets retyped. A quote becomes a job becomes an invoice, and a human copies the figures across each time. Retyping is not slow because of the typing. It is slow because of the checking, and expensive because of the one time nobody checked.

5. Someone spends hours making the report. If the end of every month involves filtering, copying into a fresh tab, and pasting into an email, you are paying a person to be a query. That is the clearest signal on this list, because it is measurable: ask how many hours, multiply by twelve.

6. You are afraid of it. Nobody wants to touch the formula in row 4. The file takes twenty seconds to open. There is a tab called DO NOT DELETE. Fear is the stage right before an accident.

One of these on its own is survivable. Three or more together, on a process that makes you money, is the point where the spreadsheet is costing more than it saves.

What waiting actually costs

Waiting is a legitimate choice, and sometimes it is the right one. But it is worth being honest about the bill, because it is paid in places that do not look like software costs.

• Hours. The recurring admin — the retyping, the reconciling, the monthly report — is the easiest cost to size. Count the hours in a normal week and price them at what you pay for that time. That annual number is usually the real budget for the project, and it is often larger than people expect before they add it up.
• Mistakes. A wrong price sent to a customer, a double booking, a job that fell off the list. Each one costs money or goodwill, and the ones that hurt most are the ones you only hear about from the customer.
• The bus problem. If one person is the only one who understands the file, your process is a single point of failure that takes holidays.
• Work you turn down. The quietest cost of all. Plenty of businesses stop taking on more of something because the admin behind it does not scale, and never write that down as a cost of the spreadsheet. It is one.
• Data you will never get back. Spreadsheets tend to overwrite rather than record. When today's status replaces yesterday's, the history is gone, and with it any chance of answering "how long does this usually take?" later.

There is a real cost on the other side too: the build, the time you spend explaining the process, and the effort of changing how people work. Waiting is right whenever the process is still changing shape every month, because building software around a process that is still being invented just freezes a draft.

What a sensible first version looks like

The most common mistake is not building too little. It is trying to build everything the spreadsheet does, plus everything anyone has ever wished it did, all at once.

A good first version is smaller than people expect and usually looks like this:

• One process, end to end. Not "our whole operation". The single workflow that hurts most — from the thing coming in to the thing being done — working completely.
• Real logins, with roles. Each person signs in and sees what their job needs, and not the rest. This is the part a spreadsheet genuinely cannot do, so it belongs in version one.
• A database instead of a grid. A customer is a record, not a row that might also exist on another tab with a different spelling. That change is what makes everything afterwards possible.
• The rules built in. The ones that lived in someone's head become things the form will not let you get wrong.
• One screen that answers your recurring question. Usually the one you currently build by hand every month.
• A way to get your data out. Export to a spreadsheet, deliberately. You will still want a spreadsheet for the one-off analysis nobody predicted, and that is fine.

What can wait: the mobile app, the second department, the integration with the accounting package, the automated emails, the clever forecasting. All of those are easier to build well once the core exists and people are using it.

Before you call anyone

You will get a better answer — from us or from anyone else — if you arrive with four things:

1. The file itself, or a redacted copy. It is the most accurate specification of your process that exists.
2. Who touches it, and what each of them should and should not see.
3. What it costs you today, in hours a week and mistakes a month.
4. What would have to be true for the project to be worth doing. If you cannot finish that sentence, that is worth knowing before anyone writes code.

If your process is still settling, keep the spreadsheet a while longer and write down the rules as they stabilise. If it has settled, and three of the six signs sound like your week, it is probably time.

Not sure which one you are? Tell us what the spreadsheet does and we will tell you honestly, including if the answer is "leave it alone for now".