Twin ChiefsLoading
Software planning

Custom software or an existing tool? Start with the workflow

The useful question is not whether custom software is better. It is whether the task you need to support fits an existing tool well enough—and whether the remaining gaps justify owning another system.

A custom application can give you control over a distinctive workflow. It also creates responsibility for maintenance, permissions, integrations, and support. An existing product removes some of that work, while introducing its own limits. Make the tradeoff concrete before choosing.

Describe the work before comparing products

Follow one real task through the people and tools involved. Record what starts it, who supplies information, which approvals happen, and where the finished result goes. Screenshots or sample records with private details removed are more useful than a general request for ‘automation.’

Look for repeated copying, delayed handoffs, and places where people keep a private spreadsheet to repair a shared process. These are clues, not automatic reasons to build. Sometimes changing ownership or simplifying a rule solves the problem without new software.

Compare three paths, not just two

There is a useful middle option between buying a product and replacing everything. You may be able to keep an existing tool as the system of record and add a small integration or focused interface around the missing step.

For example, a team could collect project enquiries through a dedicated form while using an established submission service behind it. The public experience can be specific to the company without making the team responsible for operating every part of the delivery infrastructure.

  • Configure an existing tool when its permissions, records, and main workflow already fit.
  • Connect existing systems when the pain is concentrated in a handoff or a repeated transfer.
  • Build a focused application when the core task itself is distinctive and existing products require persistent workarounds.

Test the constraints that would be hard to change

Try a representative task with realistic sample data before committing to an approach. Include a correction, a failed delivery, and a user with limited permissions. A successful happy-path demonstration does not show how the system behaves during normal mistakes.

Check data export, access control, API availability, service limits, and account ownership. If a tool is central to the workflow, identify what would happen if you changed providers. You do not need a complex migration plan for every experiment, but you should know whether the information can leave the system in a usable format.

Include the work after launch

Compare ongoing responsibilities as well as the initial implementation. An integration needs someone to notice expired credentials and failed jobs. A custom app needs someone to apply updates, investigate errors, and restore data. A purchased product needs someone to manage accounts and configuration.

Write down who handles each responsibility. If the answer is ‘the developer,’ clarify how that person receives a failure report, what information is available, and how another maintainer could take over. These details belong in the decision, not in an emergency conversation later.

Choose the smallest useful test

Build or configure enough to complete one important workflow, then test it with the people who actually do the work. Keep the inputs and success criteria the same across options so the comparison is fair.

A good outcome may be an existing tool, a modest connection between systems, or a custom product. The decision should follow the observed task and the team's capacity to maintain the result. If you cannot yet describe that task clearly, start with a product brief before commissioning a build.

See the products behind these notes in our work, or start a conversation about a project of your own.