Case Study · Tracking Quality

Mobile Analytics
Reliability
Framework

An end-to-end mobile tracking audit and monitoring framework that fixed Firebase data quality, reduced a 34% revenue gap to 1.2%, and restored reliable product and sales reporting.

Focus
Mobile analytics · Tracking quality · Data reliability
Stack
Firebase · Commanders Act · BigQuery · SQL · dbt
Type
Portfolio case study
Tracking health
96%
cross-platform parity
Daily
automated QA
Live
34%1.2%
Revenue gap reduced
12.4%0%
Missing transaction IDs
7.9%0.2%
Android duplicate conversions
3–5 days<4h
Incident resolution time
The problem

The dashboard
looked fine.
The data wasn't.

A high-volume mobile app running Firebase Analytics and BigQuery had accumulated tracking failures across revenue events, consent logic, funnel steps, and iOS / Android parity. Product, Sales, and DPO teams were seeing symptoms, but the issues were hidden behind dashboards that appeared clean.

Who flagged what
ProductOptimizing a funnel built on duplicated events for 8 months
Sales34% gap between analytics revenue and Stripe actuals
DPOPre-consent tracking violations in event dispatch
Failures found
Six issues.
Duplicate conversions on Android
claim_submitted firing 2× per session — 7.9% duplication rate
Missing revenue keys
transaction_id null on 12.4% of purchase events
Broken Android funnel events
Checkout steps out of sequence, invalidating funnel analysis
iOS / Android event divergence
Only 63% event parity across platforms
Pre-consent tracking violations
Events fired before the user consent signal was confirmed
Naming inconsistencies
Same actions mapped to different event names per platform
What I built
Three layers.

A reusable audit flow from event collection to decision readiness.

I

Audit and taxonomy

Firebase DebugView sessions, BigQuery event-level analysis, standardized taxonomy, and parameter dictionary.

II

Tracking implementation

Commanders Act dispatch logic with GDPR consent gating, parameter validation, server-confirmed triggers, and consistent event structure across iOS and Android.

III

Validation and monitoring

BigQuery and dbt validation suite with daily checks for nulls, duplicates, naming compliance, platform parity, sequence integrity, and rolling anomaly detection.

Results

Results that changed trust
in the data.

Based on event-level analysis in BigQuery and validation checks built for the framework.

MetricBeforeAfter
Revenue discrepancy34%1.2%
Monthly revenue undercount€41200€1480
Transaction ID null rate12.4%0.0%
Android duplicate conversions7.9%0.2%
Cross-platform event parity63%96%
Pre-consent violations4 event types0
Incident resolution time3–5 days<4 hours
Role & takeaways

Trust the data.

Turns tracking quality from reactive debugging into a repeatable analytics process.

My role

I owned the end-to-end analytics reliability work: auditing Firebase events, defining the tracking taxonomy, rebuilding Commanders Act event dispatch with consent gating, writing BigQuery validation checks, and structuring the monitoring framework for long-term data quality.

What this shows
Diagnose analytics reliability issues that dashboards hide.
Connect tracking failures to product, sales, privacy, and business risk.
Build reusable monitoring systems that make analytics quality repeatable.
Next.

Explore the rest of the work.

The repository includes the event audit log, approved taxonomy, measurement plans, Commanders Act tracking logic, SQL validation suite, anomaly detection, GDPR checks, and results summary.

Maïssa Bounar© 2026

Create a free website with Framer, the website builder loved by startups, designers and agencies.