Two-week sprints, one SIT deployment per release, and around 40% of SIT defects that should never have left QA. The cadence is agile; the integration backbone underneath it is not.
An add-on was missing from the service-change list only when a particular plan was in the original order. Visibility is a function of the existing subscription, not of the product.
A stuck orchestration item offers you a Complete Item button. On a self-completing integration, pressing it means you just passed a broken sync and called it done.
One formula, three artefacts, three different answers. The acceptance criterion divided by a value the workbook and the Apex ignored — and the prose was the wrong one.
One feature, three spellings, twelve artefacts written the same week — and no search on one spelling returns the other two. Three of the pages are screenshots, so they cannot be searched at all.
Six concrete test management needs were run against the tool we already had. Five turned out to be configuration. The honest migration case shrank to one architectural gap.
A regression suite without a documented selection rule is a playlist. Five criteria — Financial, Contractual, Integration, Non-obvious, Gateway — gave every test case a defensible reason to be in or out.
A deploy left a required picklist with only "-- Clear --" in it and took the whole lead-capture journey down. The response was not another bug report. It was a log with hours in it.
A discount field whose validation changes with a sibling selector, a discount that must be absent before it appears, and a rule enforced only in the UI — three controls where the state change is the test.
Four "free for one month" promotions look identical in the cart. The field that makes them temporary exists only in the billing payload — and a device field that is always populated was always populated with the wrong device.