Case Study · Journey Analytics

Cross-Device
Journey
Analysis

A journey reconstruction case study that corrected inflated dashboard conversion, exposed UTM loss, and separated tracking gaps from real product friction.

Focus
User behavior · Funnel reconstruction · Attribution quality
Stack
GA4-style events · BigQuery · SQL · Python
Type
Portfolio case study
Session journey · reconstructed
loop 3×UTM lostMobileDesktopPricingSubscribe
tracked path
attribution loss / loop
42%31%
Conversion corrected
28%
Paid sessions with UTM loss
19%
Users looping 3+ times on pricing
44% lower
Conversion for looping users
The problem

The funnel
looked stable.
The journey wasn't.

A subscription product was reporting a 42% conversion rate. The number looked stable, but growth targets were being missed. Product saw unexplained pricing drop-off, Marketing suspected attribution issues, and the dashboard was hiding how users actually moved across sessions, devices, and SPA routes.

What looked wrong
ProductHigh pricing-page drop-off without a clear cause
MarketingPaid campaigns looked strong on last-click attribution
DataDashboard conversion did not match reconstructed behavior
Findings
Four signals changed the decision.
01

Conversion was overstated

Reliable conversion was 31%, while the dashboard reported 42%.

02

UTM loss distorted paid performance

28% of paid sessions lost UTM attribution after SPA route changes.

03

Pricing loops signaled confusion

19% of pricing visitors looped 3+ times.

04

Platform gaps changed interpretation

Around 18% of Android subscriptions were untracked.

Role & takeaways

What I owned.

I designed the journey analysis logic, audited tracking quality, reconstructed sessions from event-level data, analyzed pricing loops and attribution loss, compared platform behavior, and translated the findings into prioritized recommendations.

What this shows
Reconstruct user journeys beyond default dashboard funnels.
Separate measurement bias from real product friction.
Turn fragmented event data into decisions across product, marketing, and data quality.
Next.

Explore the rest of the work.

Maïssa Bounar© 2026

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