A real-time intelligence dashboard that transformed how a fast-growing SaaS company monitors, interprets, and acts on its data — reducing analyst time by 60% and surfacing anomalies before they become problems.
The client — a mid-market SaaS company with 12,000+ users — was drowning in data but starving for insight. Their analytics team spent 3+ hours daily pulling reports from five disconnected tools: Stripe, Mixpanel, HubSpot, Postgres, and a custom event database. By the time a report was ready, the data was already stale.
Anomalies — revenue spikes, churn indicators, usage drops — were being spotted days late, after customers had already churned or opportunities had been missed. They needed a single source of truth that could not only display data, but interpret it.
Nexus is a real-time analytics command centre built on a React frontend backed by a Python FastAPI layer that ingests, normalises, and streams data from all 12 of their data sources into a single PostgreSQL warehouse — updated every 90 seconds.
On top of the data layer, we built an AI anomaly detection engine using a combination of statistical modelling and GPT-4 function calls. When an anomaly is detected, Nexus doesn't just flag it — it explains it, cross-references related metrics, and suggests a course of action in plain English.
Within 30 days of launch, the analytics team eliminated their daily report-pulling entirely. Anomalies are now caught within 3 minutes on average. The AI digest emails replaced two standing weekly meetings. The client calculated an ROI of 6x within the first quarter — factoring in analyst time saved and two major churn events that were intercepted early.