// Case study / Mobile analytics implementation
From "what should we measure?" to 30+ shipped tickets.
A greenfield mobile analytics build across iOS and Android: workshop-led measurement strategy, Firebase SDK tagging guidance, QA support, GA4 validation, cross-platform parity, consent, and downstream reporting integration.
What happened
A mobile app launched with trustworthy telemetry on day one.
A state-regulated retail platform shipped a brand-new mobile app with a full analytics layer underneath it: 30+ tickets authored, owned, and deployed across iOS and Android, cross-platform schema parity audited and closed, GA4 e-commerce taxonomy live, and consent management integrated through TrustArc before launch.
The reporting surfaces that now run on this stack work because the instrumentation layer was built deliberately rather than retrofitted after release.
Situation
Zero prior mobile instrumentation.
The client was building a mobile app from scratch. There was no mobile taxonomy, no event schema, no SDK integration, and no production telemetry deployment story. Downstream marketing, product, and compliance work all depended on the layer being right the first time.
Task
Stand up the measurement layer end to end.
The goal was full mobile analytics across iOS and Android, from workshop to tracking specification to production deployment, with explicit cross-platform parity and consent governance integrated before launch. In a state-regulated context, privacy review could not be an after-the-fact bolt-on.
SchemaEvent design
Full GA4 e-commerce taxonomy: purchase, add_to_cart, view_item, begin_checkout, items array, value, currency, transaction_id.
TelemetryInteraction events
Bottom navigation, search-query submission, PDP metadata, barcode-scanner interactions, and account-management flows.
QAParity + testing
30+ Jira tickets across iOS and Android, real-device testing, GA4 DebugView validation, regression QA checklists per sprint.
GovernancePrivacy sign-off
Privacy-compliance review before any PII-adjacent property was specified, plus data-governance sign-off from client stakeholders.
Full GA4 e-commerce taxonomy: purchase, add_to_cart, view_item, begin_checkout, view_cart, refund, items array, value, currency, transaction_id.
Bottom navigation, search-query submission with results count, PDP metadata on every event, barcode-scanner interactions, and account-management flows.
Real-device testing across iOS and Android, GA4 DebugView validation, regression QA checklists per sprint, and ticketed parity gaps.
Privacy-compliance review before any PII-adjacent property was specified, plus data-governance sign-off from client stakeholders.
Action
The mobile layer was built as production infrastructure.
I designed the event taxonomy, wrote the implementation tickets, provided Firebase SDK tagging guidance, validated event behaviour in DebugView and on real devices, audited iOS and Android schema parity, and connected the production telemetry layer to the consent setup before launch.
The work also created the data foundation for later analysis: retention cohorts in GA4 Explore, product metadata on interaction events, and a consistent mobile event shape that could be connected to the broader web/app/CRM reporting layer.
Outcome
The analytics team could trust the mobile stream from launch.
- 30+ Jira tickets authored, owned, and shipped to production for mobile greenfield instrumentation.
- Cross-platform schema parity audited across iOS, Android, and web, with gaps ticketed and closed.
- Consent management integrated with tag configuration in production through TrustArc.
- GA4 e-commerce taxonomy and screen/interaction coverage available for downstream product and marketing work.
Design lesson
Instrumentation is cheapest before launch.
When a product team asks for attribution, cohort analysis, or funnel visibility after launch, the answer depends on decisions made months earlier. The practical value of this engagement was not only what it measured, but what it prevented: retrospective cleanup, mismatched platform semantics, and consent ambiguity.