Direct answer: Start in that order. Configuration is usually the least disruptive path, integration preserves systems that already work, and a custom build is justified only when a commercially valuable workflow fits nothing adequate on the market.
Start with the workflow gap, not a replacement project
When a business says its systems are not working, several different problems may be hiding underneath. A useful feature may be switched off. Two good systems may not pass information between them. A team may be forcing a real workflow into a product that was never designed for it. Those situations look similar from the outside, but they call for different responses.
Replacing everything too early creates disruption and throws away working capability. Configuring forever can also be wasteful when the product simply cannot express the workflow. The job is to identify the smallest intervention that closes a commercially valuable gap.
The paths and their trade-offs
Configure the system you have
Configuration is the first test because it preserves the current product, data and team habits. It suits a missing approval rule, field, permission, template or report that the product already supports. Its limit is structural: configuration cannot make a product understand a workflow it was not built to represent.
Use an existing product
If another product adequately covers the workflow, buying it is often more sensible than commissioning a build. The trade-off is adaptation. Your team may need to change some practices, accept product constraints and manage a migration.
Integrate systems that already work
Integration is appropriate when each system performs its own job but a hand-off between them is manual. It can remove re-entry without replacing either system. It still creates an ongoing dependency: APIs change, failures need monitoring and somebody must own the connection.
Build the valuable gap
A focused custom build makes sense when the workflow matters commercially and adequate software does not exist. The gain is operational fit. The obligations are equally real: hosting, maintenance, security attention, change management and long-term ownership.
| Path | Best fit | Disruption | Ongoing obligation |
|---|---|---|---|
| Configure | The current product already supports the needed behaviour | Usually low because systems and data remain | Keep configuration documented through product changes |
| Integrate | Useful systems are separated by a manual hand-off | Focused around the connection | Monitor the connection and handle upstream changes |
| Build | A valuable workflow fits no adequate product | Higher because new software and rollout are involved | Host, secure, maintain and improve the software |
A practical decision framework
- 01Is the gap commercially valuable? Name the actor, the repeated workflow and the consequence. If the problem is irritating but not material, leave it alone.
- 02Can the current system be configured? Check product capability before assuming a build. Separate ‘we have not set it up’ from ‘it cannot do this’.
- 03Does an adequate product exist? Adequate does not mean perfect. Compare the cost of adapting the workflow with the ongoing cost of owning custom software.
- 04Is this really a connection problem? If both ends work, focus on the information that has to cross the hand-off and who owns failures.
- 05What valuable gap remains? Define only that gap. A custom layer should complement useful systems, not become an excuse to replace them all.
Worked example: engineered equipment quoting
A water-treatment equipment distributor can issue basic quotes from its accounting package, but its multi-supplier configuration rules live in a spreadsheet maintained by a senior estimator.
Configuration is tested first and cannot express the rules. Generic quoting products are assessed but require the business to abandon important equipment logic. The accounting package still handles orders and invoices well, so replacing it would add disruption without solving the real gap.
The focused answer is to build the quoting workflow and integrate approved information into the existing accounting process. The decision is not ‘build a platform’. It is ‘keep the working system and build the commercially valuable gap’.