SaaS Customization vs Configuration When Customers Want Different Rules
Compare SaaS customization vs configuration using a rule-change worksheet, compatibility tests and maintenance math for customer-specific requests.
Luka Abramovic3 min read

SaaS customization vs configuration is a question about what changes when a customer’s preference changes. A label or logo can often be a setting. A different approval sequence can alter permissions, saved work and notifications. Classify the request by its effect on decisions before promising a simple customization.
For software built from specialist expertise, also ask whether the request changes the standard the product is meant to preserve. A customer-specific term may be harmless. Removing a required review can change the substance of the service.
Classify the requested difference
| Request | Likely treatment to investigate | Question that changes scope |
|---|---|---|
| Different terminology | Controlled labels | Does the term also change the meaning? |
| Different session allowance | Entitlement setting | What happens to existing allocations? |
| Extra approval step | Supported workflow variant | Who may approve and resume work? |
| Different decision method | Separate product capability | Does the current output still mean the same thing? |
These are starting classifications. Write a concrete before-and-after example for each request. If the example changes the output or who may authorize it, involve the domain owner before treating it as presentation work.
Define what a changed setting does to existing work
Suppose a customer changes a review requirement while ten projects are in progress. Decide whether those projects retain the original rule or adopt the new one. Store enough information to explain which rule applied when an output was approved.
Versioning the relevant configuration can preserve that explanation. A record pointing only to today’s settings cannot tell you what yesterday’s decision meant if those settings have changed. Add an effective date or explicit version where the workflow requires historical interpretation.
Calculate the combinations the team must support
In an illustrative product, three independent on-or-off settings create 2 × 2 × 2 = 8 possible combinations. Adding a fourth doubles that to 16. Not every combination necessarily needs a separate end-to-end test, but the team must decide which interactions matter and which combinations are unsupported.
Write supported combinations down and reject invalid ones at configuration time. Hiding an incompatible option on one settings screen does not protect imports or administrative changes that use another route.
Give every supported variation an owner
Customer request and example
Shared capability it would create
Allowed values and default
Who may change it
Effect on existing records
Supported combinations
Acceptance tests for default and override
Support owner and retirement condition
Use this record to compare configuration with a separate implementation. A separate customer branch needs its own update and compatibility plan. Configuration needs validation and shared tests. Neither route becomes free because the visible change is small.
Before accepting the request, demonstrate the default customer and the overridden customer completing the same core task. If the variation makes the task fundamentally different, name and scope that difference explicitly. The useful outcome is a controlled product capability, with its boundaries visible to both the customer and the team maintaining it.
Bring one customer-specific request and its before-and-after example. We can assess whether it belongs in settings, a supported workflow variant or a separate scope. Discuss the first release with Adamant Code.