When building modern SaaS infrastructure, sending raw billing data, transaction logs, and customer Stripe exports to third-party analytics vendors introduces unnecessary compliance overhead. GDPR, CCPA, and enterprise security reviews often stall out the moment telemetry pipelines touch external servers.
To solve this, I built CohortLens - Client-Side SaaS Retention & MRR Engine, a fully client-side analytics engine that processes raw CSV exports locally in the browser using Web Workers and modern canvas rendering, keeping your financial metrics completely isolated from external tracking.
Architectural Overview
CohortLens runs entirely inside the browser sandbox. When you drop a subscription billing CSV into the interface, the execution pipeline handles parsing, cohort transformation, and matrix calculation without a single HTTP request hitting an external database.
[ Raw Billing CSV ] │ ▼ [ Web Worker Sandbox ] ──► [ Local Memory State ] ──► [ Canvas / SVG Heatmap ] │ ▼ [ Zero Network Requests / Air-Gapped ] - Client-Side Parsing: Large CSV files are ingested via streaming chunk readers to prevent UI thread blocking.
- Deterministic Cohort Matrix: Subscriptions are mapped by subscription start month and tracked through subsequent billing intervals to compute true net revenue retention (NRR).
- MRR Waterfall Decomposition: Tracks New, Expansion, Churn, and Reactivation MRR changes on a monthly cadence.
Key Technical Capabilities
- Air-Gapped Processing: Zero cloud telemetry or external logging dependencies. Ideal for strict financial audits and private equity portfolio reviews.
- High-Performance Rendering: Handles large transaction histories smoothly via optimized client-side array aggregations.
- Investor-Ready Outputs: Instant generation of retention heatmaps and waterfall charts formatted for board decks.
Try It Out
You can test the engine immediately with sample data or your own billing exports:
If you are tired of routing sensitive financial metrics through bloated cloud dashboards just to check your monthly churn rate, give CohortLens a spin and let me know your thoughts on the client-side performance benchmarks.
