When a spreadsheet stops being the right workflow

A spreadsheet that supports a useful calculation is doing its job. A spreadsheet that also assigns work, controls approvals, reconciles conflicting copies, and tells customers what happens next deserves a closer look.

The useful question is not how many rows it contains. Ask what people must do around the file to keep the business moving. This guide offers a practical way to investigate that work before commissioning new software.

Follow a record through a working day

Choose one recurring process: preparing a quote, assigning a job, or updating stock. Watch a record move from its first entry to its final use. Write down each handoff, including messages and phone calls that never appear in the sheet.

For an illustrative service business, a coordinator might enter a request, email an engineer, update a status, and separately tell a customer the appointment time. The spreadsheet may be perfectly readable while the handoffs remain difficult to track.

Record where someone asks, “Is this the latest version?” or “Who is handling this?” Those questions suggest requirements worth investigating. They do not, on their own, prove that a custom application is needed.

The Government Digital Service's discovery guidance recommends understanding the problem, users, and constraints before committing to build. It explicitly allows discovery to conclude that a project should stop. We use that principle here as a decision aid, not as a required business procurement process.

Separate a file problem from a process problem

Use a short observation log with evidence from actual work.

Try the smallest reasonable improvement first. A clearer form, a named owner, or a better use of an existing tool may resolve the problem. Record what the trial changes and what remains awkward.

Describe the replacement as a task

Avoid a brief that only says “turn this spreadsheet into an app.” It risks preserving confusing columns while overlooking the reasons people use them.

Instead, describe a result: “A coordinator can assign a request, the engineer can accept it, and everyone can see the confirmed status.” Add the important exception: “If the engineer declines, the coordinator sees it in the unassigned queue.”

Keep calculations, exports, and unusual cases in the discussion. Ask experienced staff which columns look redundant but support a monthly reconciliation or an external obligation. Plan what happens to historical records before moving active work.

Run a bounded comparison

Test your improved spreadsheet, a suitable existing product, and any proposed custom approach against the same task. Let the people who perform the work try representative sample records.

Agree what you will observe: time spent chasing a status, duplicate entry, unresolved requests, or corrections. Establish your own baseline; an external percentage is not evidence of your likely savings.

End with a decision record: the problem, evidence, options tried, and responsibility for the next step. If a focused connection is enough, start with our guide to choosing a first automation. If a broader replacement is justified, compare custom software and SaaS against the workflow you have now documented.