Signs Your Business Has Outgrown Spreadsheets (and What Replacing Them Costs)

Quick answer

Spreadsheets fail predictably: when more than two or three people edit the same file, when one workbook feeds decisions it was never designed for, and when “the spreadsheet person” leaving would hurt. If your business runs on a file with a name like MASTER-jobs-2026-v7-FINAL-USE-THIS-ONE.xlsx, you’re past the threshold. The escape path doesn’t have to be a big-bang custom build — the sensible ladder runs from better tools (Airtable, purpose-built SaaS) at £10–£50/user/month up to custom web applications from £5k–£15k for an MVP.

The seven failure signs

1. Version chaos. Multiple copies circulating by email; nobody is sure which is current. Every duplicate is a future bad decision.
2. The silent-error problem. Studies of real-world workbooks consistently find errors in the majority of complex spreadsheets — a mistyped formula doesn’t crash, it just quietly misprices your quotes for six months.
3. One person owns it. If a single employee’s resignation would take your operations knowledge with them, that’s not a spreadsheet, it’s a liability.
4. Manual re-keying. Copying data between the spreadsheet and your accounting package, email, or job sheets. Re-keying is where errors breed and hours vanish — often 5–10 staff-hours a week by the time businesses notice.
5. Access anxiety. You can’t safely let a subcontractor or new starter see part of it. Spreadsheets have no concept of roles or permissions.
6. It’s become the database. Customer records, job history and pricing in one workbook — with no audit trail of who changed what, and GDPR obligations you can’t actually meet in a shared file.
7. Growth stalls it. Ten thousand rows, cross-workbook lookups, five-minute recalculations. Excel is superb software being used as infrastructure.

The escape ladder (cheapest first)

Rung 1: Purpose-built SaaS — £10–£50/user/month

If your spreadsheet is doing a job that thousands of businesses share — CRM, job scheduling, quoting, stock — there’s a product for it (and switching to it is configuration, not development). This is the right answer more often than software developers like to admit.

Rung 2: Structured no-code — £20–£100/month

Airtable, NocoDB and friends give you the database behaviours spreadsheets lack — roles, forms, views, audit history — while keeping the grid familiarity. Good for workflows too specific for off-the-shelf SaaS but too simple to justify custom development.

Rung 3: Custom software — from £5k

Justified when the workflow is your competitive advantage, when rungs 1–2 would force your process into someone else’s shape, or when integrations (accounting, suppliers, your website) need to flow without re-keying. UK pricing: MVPs £5,000–£15,000, production systems £15,000–£75,000 — full comparison in our guide to custom vs off-the-shelf software. Increasingly, the first custom step is smaller: automating the copy-paste between existing tools rather than replacing them.

What a migration actually involves

• Data archaeology (the underestimated bit): cleaning years of inconsistent entries — “J Smith”, “John Smith” and “Smith, J” are three customers until a human says otherwise. Budget real time here.
• Workflow mapping: documenting what the spreadsheet actually does, including the workarounds nobody wrote down. Often the first time the process has ever been examined — and where the improvement comes from.
• Parallel running: two to four weeks of both systems side by side before the spreadsheet is retired. Never big-bang it.
• The exit rule: whatever you move to, confirm you can export your data out again. You’re escaping lock-in, not changing landlords.

The honest cost comparison

If you’re at the point of pricing the custom rung, our custom business software service covers discovery, build and the parallel-run — and connecting it to everything else you run is what API development and integration is for.

Costing the spreadsheet you already have

The comparison most businesses make is “free versus £8,000”, which guarantees the wrong answer because the left-hand column is not free. Price the current position honestly first. A five-person firm where each person loses an hour a day to re-keying, chasing the current version and fixing broken formulas is spending twenty-five hours a week on admin. At a fully-loaded cost of roughly £22 an hour, that is about £550 a week, or £26,000 a year, before you count a single mispriced quote.

You will not recover all of it — no system removes admin entirely — but recovering half of twenty-five hours pays for a £12,000 build inside a year and keeps paying afterwards. Run the same sum on your own numbers before any supplier conversation:

• Re-keying time: hours per week moving data between the workbook, your accounts package, email and job sheets.
• Chasing time: hours spent finding the current version, asking who changed a cell, or rebuilding something someone overwrote.
• Error cost: one mispriced job a quarter at your average margin is a real, recurring number.
• Opportunity cost: the work you decline or delay because scheduling it manually is painful.
• Key-person risk: what it would cost in disruption if the spreadsheet’s author left with a month’s notice.

Write the total on one line. That figure is your budget ceiling and your business case in the same number, and it is the only thing that will stop the project stalling at the first quote.

A worked example: the quoting spreadsheet

A ten-van maintenance firm quotes from a workbook containing 400 rows of materials pricing and a labour matrix. Symptoms: two engineers keep offline copies, supplier prices are updated quarterly at best, and every quote is retyped into the accounts package. What replacing it costs and returns:

The honest answer for that firm is usually the second row. Custom becomes right when the pricing logic itself is the differentiator — a rule set no product supports, a materials model competitors cannot copy quickly — or when the integration burden across three existing systems is the actual problem, which is where the cost of integration work becomes the deciding line item rather than the build.

The costs that appear after the invoice

Buyers budget the build and forget the operating model. Four items catch people out.

Maintenance. Custom software needs hosting, dependency updates, security patching and small fixes — plan for 15–25% of build cost annually, and treat a supplier who never mentions it as a warning sign. The shape of those bills is set out in what keeping custom software alive really costs.

Training and the productivity dip. Expect four to eight weeks where the team is slower, not faster. Budget the time and say so out loud, because an unannounced dip is what causes staff to quietly retreat to the old workbook.

Change requests. The requirements you discover in month two are real requirements, not scope creep — but they cost money. Hold back 15–20% of the budget for them rather than spending it on features nobody has used yet.

Data migration. Cleaning and importing years of inconsistent records is frequently 10–20% of a project’s effort, and it is the part suppliers most often exclude from fixed quotes. Ask explicitly who owns it.

Questions to put to any supplier

• “What would you build first, and what would you deliberately leave out of version one?” A supplier who wants to build everything has not understood the risk.
• “Where does the data live, who can export it, and in what format?” Get the answer before you sign, not when you leave.
• “What happens to this system if you are unavailable for three months?” Documentation, repository access and a second pair of hands should already exist.
• “How does this connect to the accounts package and the website?” Isolated systems recreate the re-keying you paid to remove — a job form on your own business website that drops straight into the system is often the highest-value integration and the cheapest to build.
• “Show me something similar still running after three years.” Longevity predicts maintainability better than any portfolio screenshot.

Edge cases where the ladder does not apply

Regulated or personal data. If the workbook holds health, financial or extensive customer data, permissions and audit trails stop being a convenience and become a compliance requirement — that pushes you up the ladder sooner than efficiency alone would, for the reasons in what GDPR compliance actually requires.

Seasonal businesses. Migrate in your quiet quarter. A parallel run during peak season is how projects get abandoned halfway.

Genuinely simple lists. A contact list, a rota for four people, a stocktake done twice a year — leave them alone. Not every spreadsheet is a system waiting to happen, and replacing a working list with a login screen is a downgrade.

Businesses about to change shape. If you are merging, selling or restructuring within a year, buy the cheapest thing that holds until the dust settles. Custom software built for a business model you are about to abandon is the most expensive kind of waste.

Sources & Further Reading

• Spreadsheet Error Research & Horror Stories — EuSpRIG
• UK GDPR Guidance — Information Commissioner's Office
• Airtable Pricing — Airtable