Making Retention Visible — A Dashboard Built to Surface What's Actually Happening
Hero — dual laptop mockup showing the dashboard, full bleed
Most teams had the data. They just couldn't see it.
Product usage data at most tech companies lives across multiple tools, spreadsheets, and departments. Executives can't easily track which features are driving retention. Analysts spend hours assembling what should be obvious. Rootly was designed to close that gap — a lightweight dashboard that consolidates retention trends, cohort behavior, and churn signals into one surface built for lean product teams who need answers fast.
Research, a fake company, and Python-generated data.
Over 10 weeks I mapped user journeys, analyzed survey responses in weekly check-ins with industry professionals, wireframed layouts iteratively, and benchmarked competitor platforms to identify gaps. To make the work concrete and testable, I created a persona company — StreamFlix, a fictional streaming service facing realistic retention challenges.
To populate the dashboard with believable data, I used the Faker Python library to generate synthetic user behavior — simulating engagement volatility, onboarding drop-off, and cohort-level churn patterns. This made the design genuinely testable and gave stakeholders a tangible, data-rich story to engage with rather than empty wireframes.
StreamFlix data simulation — dashboard populated with Faker-generated cohort data
I also compiled a dedicated research document on effective data visualization principles to ground my layout and chart decisions — read it here.
A layout that answers questions, not just displays numbers.
Low-fidelity wireframes tested hierarchy before any visual decisions were made. The core question driving them: what does a retention-focused team need to see the moment they open the dashboard? The answer became four primary views — onboarding funnel, feature engagement over time, cohort retention heat map, and churn signal indicators.
Low-fidelity wireframes — dashboard layout explorations
High-fidelity dashboard — full view with all four primary panels populated with StreamFlix data
Two StreamFlix scenarios showed exactly where the dashboard changed outcomes.
Scenario 1 — Onboarding drop-off
StreamFlix had been increasing ad spend and lowering pricing to fight stagnating retention. Rootly surfaced the real problem: a broken email verification step was causing 50% of new users to drop before activation. Fix the bug, not the budget. Onboarding success went from 50% to 83%.
Scenario 2 — Feature underutilization
Key features weren't being discovered. Rootly detected region-specific drop-off patterns, enabling targeted testing in low-performing areas. Result: 17% increase in feature engagement within 30 days, 9% decrease in early-stage churn attributed to feature confusion.
Before / after impact screens — onboarding funnel view and feature engagement comparison
The wireframe phase mattered more here than on any other project.
You can't figure out data hierarchy in high fidelity. Rootly pushed me hardest on one specific question: how do you make a dense, data-heavy interface feel navigable rather than overwhelming? The answer was deciding what was primary, what was secondary, and what required user initiation — before touching a single visual.
The StreamFlix simulation was also a lesson in how much more convincing designed work becomes when it's populated with real-feeling data. Empty charts don't tell a story.
Next time I'd prioritize user testing in a less controlled environment earlier, and explore what a lighter-weight setup looks like for teams that don't need the full system. Not every company is StreamFlix.