Pre-Launch Checklist

This is an honest checklist of open items before wider rollout. Some are security/permissions gaps, some are naming/process cleanup, some are pure infrastructure. None of these are hidden — check them off as they're resolved, don't skip them because the app "mostly works."

Access & permissions

• Gate the ungated routes. /bookings/new, /estimate, /workbench/:proposalId, most /admin/* pages (beyond the ones already in the page registry), and /csr-intake can currently be reached by any signed-in user just by typing the URL, regardless of role. Add these to the page registry with sensible role lists. See /help/permissions/known-gaps.
• Decide: should Communications and Training really be visible to everyone? They currently have no role restriction. Confirm this is intentional, or scope them down.
• Document the Admin vs. Executive distinction for end users. The two roles look identical in the UI (both bypass all page gates), but only Admin can actually invite users or edit permission sets — enforced server-side. Executives who click those buttons will be blocked. Make sure Executives know this in advance rather than discovering it.
• Add /approvals to the sidebar for the roles that need it (at minimum manager, likely executive), or explicitly decide it should stay a direct-link-only page.
• Retire or redirect /csr-intake. It's a legacy page superseded by /bookings/new and is currently ungated. Decide whether to remove it, redirect it, or restrict it.

Naming & terminology consistency

• Confirm all customer- and team-facing copy says Necessary Components, never "adders."
• Confirm SAD is consistently spelled out as Site Assessment Deposit on first use in any given screen/document.
• Reconcile pipeline stage labels shown in the UI with the internal database keys used in code/conversation, so training materials and support conversations use consistent language. See /help/rollout/naming-conventions.

Permission sets

• Permission sets are currently underused — most users only have their role's default access. Define one standard permission set per role as a starting template, rather than leaving grants ad hoc. See /help/rollout/open-decisions for the specific role recommendations.

Data & environment

• Confirm the demo-data flag is OFF for all production user accounts before go-live.
• Stand up a staging environment. There is currently no separate staging deployment — all testing happens in the same environment used in the field. This should be resolved before scaling up usage.

Integrations

• Fix the ServiceTitan "Door / slab" mapping. SKU 39723877 was deleted on the ServiceTitan side, so pushes for that material bucket currently fail. An active replacement SKU needs to be mapped in /admin/servicetitan-mapping before launch, or Door/slab jobs need a documented manual workaround.
• Spot-check the Framer webhook → RentCast property lookup pipeline for a few real leads to confirm it's reliably populating property data (see /help/troubleshooting/property-lookup-missing).

Role segmentation review

Recommended cleanup before broadening rollout — see full rationale in /help/rollout/open-decisions:

• Keep Phil on door_expert (don't promote to manager unless he genuinely needs manager-level visibility).
• Keep Danielle on csr; only add dispatcher if she is actually scheduling site assessments herself.
• Keep Sergio on project_manager; only add reconciler if he's doing scope-validation work, not just project management.
• Keep installer strictly read-only, separate from any PM write access — don't blend the two.
• Prefer permission sets for one-off expansions over stacking additional roles onto someone's account.

Sign-off

Before calling launch "done," a manager or admin should be able to check every box above, or explicitly note why an item is deferred (with a target date) in /help/rollout/open-decisions.