Why this is your problem, not Oracle's
Oracle tests the standard product. It does not test the SuiteScript written for your go-live, the workflow built around your approval chain, or the integration your team maintains against a third-party system.
That gap is the whole reason release testing exists as a task. The platform being fine and your account being fine are different statements, and only one of them is somebody else's responsibility.
Step one: get the preview account early
Release Preview is requested from Setup, Company, Release Preview, and it becomes available on a published schedule ahead of each release. The preview account mirrors your live environment, so requesting it early gives you the maximum testing window.
One practical trap: preview accounts are purged after a stretch of consecutive inactivity. Requesting one and then not logging in for two weeks means starting over, so put a recurring reminder in the calendar for the testing window.
Step two: test your customizations, not the feature list
Reading the release notes tells you what changed. It does not tell you what broke. The only way to learn that is to run your own configuration against the preview.
Work outward from the things most likely to be affected by platform changes, then cover the flows the business cannot operate without.
- SuiteScript: user event, client, scheduled, Map/Reduce, Suitelet, and RESTlet scripts.
- Anything touching authentication or tokens — integrations fail here first.
- Workflows and approval chains, especially ones with conditional branching.
- Saved searches and the dashboards and reports built on them.
- Third-party integrations end to end, not just a connection test.
- Custom record types and any field behaviour flagged as changed in the notes.
Step three: test the flows the business runs on
After the technical surface, walk the processes people actually use: order to cash, procure to pay, and the month-end close. A release that leaves every script running but changes a form's behaviour can still stop an order being entered correctly.
Use real scenarios rather than happy paths. The failures worth catching are usually in the exception handling — the credit hold, the partial fulfillment, the multi-subsidiary transaction.
Step four: write the plan down once
The single highest-value output of a release cycle is not the testing itself; it is a written test plan you can run again in six months. NetSuite publishes sample test plan templates alongside the release notes, which is a reasonable starting structure.
Record what you tested, what you expected, what happened, and what you fixed. The second cycle then costs a fraction of the first, and it survives the person who ran it leaving.
- One row per scenario: what you ran, expected result, actual result.
- Note the owner for each area so nothing falls between finance and IT.
- Keep the fixes and their reasoning — the next person will ask why.
If nobody owns this in your company
Release testing is the clearest example of recurring work that has no natural owner. It has no ticket, no requester, and no deadline until the release lands — so in an account with no dedicated administrator, it is almost always the first thing that stops happening.
That is why it is included at every tier of a fractional administration retainer rather than sold as a project. Work that happens on a fixed schedule needs an owner on a fixed arrangement.
Shahin Zakizadeh
Founder & Principal NetSuite Consultant at SZnetsuite — SuiteScript development, automation, integrations, and billing for growing NetSuite teams.
About the practiceRelated services
Keep reading
NetSuite 2026.2 Release: AI-Fortified Financial Management, Explained
A breakdown of the AI-driven bank reconciliation, close management, and payment automation coming in NetSuite 2026.2 — and what to test before it lands.
Read articlePerformanceNetSuite Performance Optimization Checklist
A practical checklist for finding bottlenecks in searches, scripts, workflows, integrations, and user experience.
Read article