How a test plan is laid out
Test plans use these sections, in this order. A plan leaves out a section it does not need, and some plans add their own, such as API only or Edge cases worth a run.
Refer to a check by its ID in every bug report and test run, for example “
TC-14 fails on iPhone 15, iOS 26.2”.
Pick the right environment
Each surface talks to one backend. Pair a web URL with a mobile build that uses the same backend, or a message sent on one will not appear on the other.
A TestFlight build carries the mobile code from the
qa branch, but it calls the production API. A mobile check that also needs a new API change passes in TestFlight only after that API change reaches main. What’s ready shows, for each feature, whether it has reached qa and main.
Which mobile build to use
- TestFlight (iOS). Use this for manual QA. A new iOS build reaches TestFlight on every push to
qa. Install the newest build before each test run. - Play internal (Android). An Android build reaches Play internal only when someone builds and submits one by hand, so the newest Play internal build can be older than
qa. Before you test a recent change on Android, ask the QA lead whichqacommit the build came from. - App Store or Play Store. Use this only to compare against what customers have today. A store build gets JavaScript changes over the air once they reach
main. A change that needs a native build waits for the next store release. See Older builds. - Development, Revyl, or simulator builds. Engineers and agents use these against the dev backend. See the agent runbook.
Accounts, roles, and orgs
Most checks name a role. Use a separate account for each role, because Soundcheck shows different screens to each one:
The QA lead gives each tester their accounts and the test org name. Do not create orgs or invite people outside that org.
Feature flags decide what you see
Some features are on only for selected orgs. Soundcheck reads its feature flags from PostHog, and one PostHog project serves dev, QA, and production. A flag change therefore applies to every environment at once.
If a feature is missing, confirm with the QA lead that your org is on the list before you file a bug. A missing feature in an org that is not on the list is expected.
Test data rules
- Name everything you create. Start each gig, chat, and note title with
QAand the check ID, for exampleQA TC-03 hourly shift. - Do not publish a draft gig that has invitations. Publishing a draft sends its pending invitations by push, text, and email to real people. Publish only gigs whose crew are all test accounts.
- Do not message or @mention real people. Chat sends pushes and texts. Keep test chats to test accounts.
- Archive, do not delete, gigs with clocked hours. The API refuses to delete a gig that has time clock punches.
- List what you leave behind. End every test run with the gigs, chats, and records you created, so the QA lead can clean them up.
Report a bug
File every bug in the Linear QA team. Add the Bug label, plus mobile, web, or backend for the surface where it happens. One bug per issue.Test plans
What to test
Every feature ready for QA, where it is deployed, and what is out of scope.
Chat
Gig chats on web and mobile, the SMS bridge, and message actions.
Time clock
Crew clock in and out, crew hours, approval, and timesheets.
Travel and venue
Crew flights and rides, the ride card, and the venue map.
Mobile gig screens
The redesigned gig screen, staffing, offers, and agreements.
Invites and onboarding
Text and email invites, sign-up paths, and where invitees land.
iPad, Performance mode, and platform
Wide layouts, Performance mode, widgets, and calendar subscribe.
Agent runbook
How an agent runs these plans and reports results.