The F-shaped differences
- Licensing: F SKUs price for deskless reality — Teams included with caps/omissions vs E (smaller mailboxes, no/limited PSTN paths by default, device-first assumptions). Fit the capability list to the JOB (does a barista need Copilot?) rather than defaulting everyone to E3.
- Shared device mode: one device, many humans — sign-in/out flows hardened for shift handover (and MAM/Intune shared-device enrolment underneath); the mobile module's per-user assumptions invert here.
- The frontline app set: Shifts (schedules as data — with connectors to real WFM systems), Walkie Talkie (PTT over the network — a headset decision as much as software), Tasks/checklists pushed FROM managers, Praise. The governing insight: frontline Teams is consumed through 3-4 apps on a home screen, not through channels-and-meetings culture.
- Targeting at scale: hundreds of near-identical stores want dynamic
groups (store attribute -> policies, teams, app setup) + deployment
tooling for team-per-location creation — hand-curation dies at 50 sites
(
TeamsTargetingPolicyin your scanned surface plays here).
Rollout pattern that works
Pilot ONE region with the store-manager workflow (shifts published, PTT channels mapped to zones, handover checklist) -> template it (lifecycle module's templates + provisioning automation) -> stamp per site with dynamic membership -> measure through the frontline lens (shift-app engagement, PTT usage), not office metrics (meeting minutes are meaningless here).
What to watch (proofs)
- Licence-fit audit: F vs E assignment against role catalogue — the delta IS money; re-run at renewal.
- Shared-device hygiene: sign-out-at-handover actually happening (device sign-in logs: user succession on shared devices) — lingering sessions are both a privacy and an attribution problem.
- Shifts adoption: schedules published per site per week (Shifts/Graph reporting) — the one metric that says the deployment is real.
- Policy reach at scale: the targeting/dynamic-group memberships vs store roster — a store missing from the dynamic group gets NOTHING; reconcile monthly (and the app-setup policy census in the scan shows what each population actually receives).