Insights

It usually begins with a perfectly sensible spreadsheet.

Someone needs to track enquiries, schedule work, calculate pricing or produce a monthly report. A few headings are added, some formulas save time and the file does exactly what is needed.

Then the business grows around it.

More columns appear. Several tabs begin feeding one another. Colours acquire meanings that are not written down anywhere. One person knows why a particular row must never be deleted, and another keeps a separate copy because they do not quite trust the shared one.

At some point, the spreadsheet stops being a document and becomes software. Nobody made a formal decision for that to happen, but important work now depends on it.

Spreadsheets are good at what they were built for

This is not an argument against spreadsheets. They are flexible, familiar and extremely useful for calculations, analysis and relatively contained sets of information.

They are also one of the fastest ways to test an idea. A process that is still changing can often be explored in a spreadsheet before anybody commits to building a system around it.

The problem begins when the file is expected to manage responsibilities it was never designed to carry: permissions, complex workflows, simultaneous users, customer communication, audit history, integrations and a dependable source of truth.

The spreadsheet may continue to work, but the effort required to keep it working starts to spread through the organisation.

The warning signs are usually operational

A complicated formula does not automatically mean a business needs custom software. The stronger warning signs are found in what people must do around the spreadsheet.

It may have outgrown its role when:

• staff regularly create copies to avoid overwriting one another;
• only one person understands how key formulas or macros work;
• information is copied manually between the spreadsheet and other platforms;
• rows represent different stages but there is no reliable workflow or history;
• sensitive information is visible to everyone with access to the file;
• customers wait while staff find, check and re-enter information;
• reports require hours of cleaning before they can be trusted; or
• a small mistake can alter pricing, scheduling or customer records without warning.

These are not really spreadsheet problems. They are signs that an informal tool is supporting a formal business process.

Do not automate the mess

The obvious next thought is often to reproduce the spreadsheet as an application. Every column becomes a field, every tab becomes a screen and every formula is rebuilt in code.

That can preserve years of accidental complexity.

Before considering custom web application development, the process needs to be understood independently of the existing file. Which information is genuinely required? Who creates it? Who approves it? What changes over time? Which exceptions are legitimate, and which workarounds only exist because the current tool is limited?

Sometimes the most valuable part of development is not replacing the spreadsheet. It is discovering that several steps no longer need to exist.

Off-the-shelf, integrated or custom?

Once the requirement is clear, there are usually several possible paths.

An established platform may already solve the problem. Customer relationship management, project management, inventory and booking systems exist because many organisations share similar requirements. Adopting one can be faster and less expensive than developing a bespoke application.

If the main problem is duplicated data, integration may be enough. Connecting two existing systems can remove hours of re-entry without replacing either of them.

Custom development becomes more compelling when the workflow is distinctive, creates a genuine competitive advantage, involves several user groups or cannot be represented cleanly in available software.

The right choice is the least complicated one that supports the real process reliably. Owning custom software is valuable when the custom part matters, not simply because it is possible to build.

Where AI and automation fit

Modern automation creates opportunities beyond moving information from one box to another.

AI and automation can help classify incoming information, extract structured details from documents, draft communication, identify unusual records or guide staff through the next step. Those capabilities are particularly useful where part of the process involves interpreting messy human input.

They still need boundaries.

A pricing calculation, permission check or status transition may be better handled by deterministic rules that produce the same outcome every time. AI can support judgement without being placed in charge of every decision. The design should make it clear which actions are automatic, which are suggested and which require a person to approve them.

A system should reduce dependence, not relocate it

Replacing a spreadsheet with an application is not an improvement if the new system is undocumented, impossible to maintain or understood by only one developer.

A dependable system needs appropriate permissions, backups, logging and recovery. Important rules should be documented. Data needs a clear owner, and staff need to understand what happens when something goes wrong.

It may also need to connect with the public-facing website. A customer could submit information once, follow progress or receive timely updates while staff manage the operational workflow behind the scenes.

The objective is not to turn every internal process into a large software project. It is to reduce the invisible risk and repeated effort surrounding work the business already performs.

Start with the spreadsheet everyone is afraid to touch

Most organisations have at least one file that inspires a slightly nervous kind of respect.

People know it is important. They know it is fragile. They also know replacing it sounds disruptive, so another column is added and everybody carries on.

That does not mean the spreadsheet should immediately be retired. It means the process deserves to be documented while the people who understand it are still available to explain why it works the way it does.

The best time to examine business-critical software is before anybody discovers, at precisely the wrong moment, that it was only ever a spreadsheet.