When to replace spreadsheets with custom software
When should you replace a spreadsheet with custom software?
Replace a spreadsheet when it has stopped being a record and become something people reconcile, chase and defend: when several copies disagree, when only one person understands the formulas, or when things go overdue and nobody notices. Until then, keep it. A spreadsheet is cheap, flexible and understood by everyone.
You rarely need to replace all of them. Replace the one at the centre of the work, the sheet everybody depends on and nobody quite trusts, and leave the rest alone until they cause trouble.
What are the signs a spreadsheet has outgrown the business?
Look for these five. If two or more are true, it is worth working out what a system would cost:
- Several copies of the truth. The sheet is emailed, downloaded and saved as 'final v3', and two people quote two different figures for the same thing. Nobody can say which copy is current without asking around.
- Manual reconciliation. Someone spends hours each week or month matching one sheet against another, or against the bank, before anyone can trust a report. The work produces nothing new; it only puts back what the copies lost.
- No permissions and no audit trail. Anyone who can open the file can change any cell, and nobody can tell who changed a price or deleted a row, or when.
- Formulas only one person understands. A chain of lookups and hidden columns works because one person built it. When they are on holiday the month-end waits, and when they leave it breaks.
- Things go overdue and nobody notices. A payment date, a service date or a renewal sits in a cell, and a cell doesn't tell anyone. It gets noticed when a customer complains or an auditor asks.
What does a system do that a spreadsheet can't?
A system stores each thing once, as its own record, and lets records point at each other. In our Real Estate Management app, properties, contracts, instalments and payments live in one system instead of five spreadsheets. A confirmed contract generates its payment schedule, and each instalment is marked paid, due soon, overdue or critical by the software rather than typed in by a person. Overdue instalments appear on a board, in a header bell and on the dashboard, which deals with the 'nobody noticed' problem directly.
A system also decides who may do what. A matrix of view, create, edit and delete rights for each role, across each kind of record, is part of our Business Platform, the shared foundation our property, hotel and file apps run on. Roles, permissions and a record of who changed what are part of how we build business systems, not an add-on.
And a system keeps history. Our File Management keeps up to 50 versions of each file with an audit log of uploads, views, downloads and changes, which ends the 'which copy is current?' argument for documents too.
What should you build first?
Build the smallest system that removes the worst sign, not a copy of every tab. In practice a first release needs four things:
- The records. Name the things the sheet tracks, such as clients, contracts, assets or orders, and the fields each one really needs. Most old sheets have columns nobody has filled in for a year.
- The states the software should work out. Paid or overdue, due for service or not, in stock or not. Every status a person types by hand today is one the system should compute.
- The people. Who enters data, who approves and who only reads. That is your first permissions matrix.
- One or two reports. The figures someone rebuilds every month. Exports to CSV or PDF cover the rest, and people can keep analysing in a spreadsheet, now from data they can trust.
Leave the nice-to-haves for the second release
A dashboard can wait; the overdue list can't. Each feature you defer makes the first release cheaper and gets it in front of the people who use the sheet sooner, and their feedback on a working system is worth more than another month of planning. The cost estimates below show how much one deferred feature saves.
How do you migrate spreadsheet data safely?
Migration is where these projects go wrong, usually because the data is messier than anyone admitted. Do it in six steps:
- Inventory the sheets. List every file and tab that holds the data, who edits each one, and which copy people treat as current. Expect to find more than you thought.
- Clean. Remove duplicates, fix dates stored as text, split cells that hold two values, and agree one spelling for each client or place. Rows nobody can explain are usually better kept and flagged than guessed at.
- Map. Write down which column becomes which field, which values become which choices in a list, and what happens to columns with no home. The map is a document both sides sign off, not something in a developer's head.
- Import in rounds. Import into a copy, check record counts and totals against the sheet, fix the map or the data, and import again until the totals match. Our Real Estate Management app, for example, imports expense lines from a workbook.
- Run in parallel. For one full cycle of the work, such as a month-end or a payment run, keep both going and compare the results. Every difference is either a data problem or a missing rule, and both are cheaper to find now.
- Cut over. Pick a date, do a final import, then make the old sheets read-only and keep them as an archive. A sheet that can still be edited will be, and you will be back to two copies of the truth.
What catches people out in a spreadsheet migration?
Three things. First, meaning stored in formatting: a red row means disputed, a struck-through one means cancelled, and an import sees neither, so ask what every colour and comment means before you map. Second, rules hidden in formulas, such as a discount applied only to some clients or a date pushed to the next working day; they must become rules in the software or they quietly disappear. Third, the person who knows the sheet best is usually the busiest, so book their time for the cleaning and the parallel run before the project starts.
What does it cost to replace spreadsheets with custom software?
It depends on how many kinds of user, features and connections the first release needs. We ran two typical cases through our MVP cost calculator, which is anchored on our own published prices:
- A focused first release: a web app for two kinds of user, such as office staff and managers, with an admin panel, email reminders and alerts, and dashboards and reports, no integrations, a proven design system and a clear scope. The calculator puts it at 13 developer-weeks: $17k to $24k as a fixed-scope build, with two developers and about 8 weeks from kick-off to first real users, or $12k a month for about two months as a dedicated team. Leaving the dashboards and reports for a second release saves about $4k.
- A broader system: three or more kinds of user, an admin panel, notifications, file uploads, search and filters, reports, more than one language and one or two integrations, such as an accounting package, with a scope that is still moving. That comes to 24 developer-weeks: $31k to $45k as a fixed-scope build, with two developers and about 13 weeks. Because the scope is still moving, the calculator suggests a dedicated team instead, at $12k a month for about three months.
Estimates are a starting point, not a quote
The weights behind the calculator are judgement, and the written scope is where the real price is set. Our fixed-scope builds start from $12k, and a dedicated developer costs from $6k a month.
Running costs come after launch: hosting, backups, security updates and the occasional change. Our care plans start from $900 a month, and the software maintenance cost calculator estimates your own case. Compare all of it with what the spreadsheet already costs you: the hours spent reconciling, and the payments nobody chased.
Should you buy a ready-made product instead?
Sometimes. If the sheet tracks something common, such as accounting, payroll or a simple contact list, a mature product will usually fit and cost less. Build when the sheet holds your own process: the way you sell, schedule, approve or collect that no product matches without workarounds. We go through that decision in custom software or off-the-shelf.
How we replace a spreadsheet
We start by asking what breaks today and who it hurts, then write a scope with a price and a date. Work moves in two-week slices with a demo every Friday, so the people who use the sheet can check the new system against it while changes are still cheap. You own the code from the first commit. That is our custom software development service.
How long does it take to replace a spreadsheet with software?
For a focused first release with two kinds of user, an admin panel, reminders and reports, our calculator estimates about 8 weeks from kick-off with two developers. Allow one cycle of the work for the parallel run before you switch the sheet off. A broader system with more roles and integrations comes to about 13 weeks.
Will we lose our historical data when we move off spreadsheets?
No, if the migration is done in rounds. Old records are imported and checked against the sheet's counts and totals, and after cut-over the sheets are kept read-only as an archive.
Can we still use Excel for analysis after the switch?
Yes. Reports export to CSV, and some systems connect to BI tools directly; our Real Estate Management app can connect to Power BI. The difference is that the export now comes from one trusted record instead of a copy of a copy.
Can't we just lock cells and tidy up the spreadsheet?
That helps for a while: locked cells, validation and one shared online copy fix some of the problems. They don't give you records that point at each other, rights per role on each kind of record, or states the software works out and puts in front of the right person. When those matter, a system is the fix.
Who owns the software once it is built?
You do, from the first commit. It lives in your repository and your cloud account. If you want a second opinion on your own sheets, tell us what they track.