Overview
A subscription software company serving logistics clients needed near-real-time visibility into activation, churn risk, and revenue metrics. Their existing BI stack required analysts to write SQL for every new question, creating a backlog measured in weeks.
Mobintix implemented a Next.js dashboard backed by Python ETL jobs orchestrated with Airflow and BigQuery as the analytical store. Role-based views expose KPIs for executives, customer success, and product teams without sharing raw table access.
Interactive D3 charts support drill-down from company-level summaries to individual account timelines with export to CSV for board packs.
Challenge
Source data arrived from product telemetry, billing, and support tickets with inconsistent identifiers. Stakeholders distrusted numbers that disagreed between finance and product spreadsheets. Page load times on the legacy tool exceeded thirty seconds for common filters.
Approach
We built a unified account dimension with survivorship rules agreed by finance and product. Streaming ingestion lands events within fifteen minutes; nightly jobs rebuild slowly changing attributes. Cached aggregates power landing pages while detail queries run on demand against partitioned tables. A metrics catalog documents definitions so everyone uses the same churn and ARR formulas.
Outcomes
- Manual weekly report assembly dropped sharply as teams adopted self-serve views.
- Executive reviews use same-day data instead of five-day-old exports.
- Customer success identifies at-risk accounts earlier through health score tiles.
Lessons learned
Agree metric definitions before drawing charts — politics kills BI projects faster than technology. Invest in a metrics catalog early. Partition BigQuery tables by date to control cost as event volume grows.
Review dashboard usage monthly and retire unused tiles. Cluttered BI homepages erode trust the same way cluttered mobile home screens do.