Overview
Self-hosted Apache Superset analytics platform replacing manual actuarial Excel models — giving actuaries, product managers, and the board live loss-ratio, combined-ratio, and premium-velocity dashboards with drill-down to individual policy cohorts.
The challenge
Monthly actuarial reporting required five days of manual data extraction from the policy administration system, pivot-table modelling in Excel, and PDF distribution. Loss-ratio signals on the motor book arrived 45 days late — after the pricing cycle had already locked — meaning pricing errors were not corrected until the following quarter. The board received loss ratios at product-line granularity with no ability to drill to geography, channel, or vintage.
The solution
We deployed Apache Superset on the client's Qatar-region infrastructure, using Airbyte to sync the policy administration system's Oracle database into a PostgreSQL analytics warehouse. We built dbt models for loss ratios, combined ratios, premium velocity, and reserve adequacy — with policy vintage, channel, geography, and agent as standard dimensions. Custom Superset dashboards used the drill-down feature to let actuaries move from board-level combined ratio to individual cohort loss triangles in three clicks.
Outcomes
Actuarial reporting cycle from 5 working days to 4 hours
Loss-ratio signal latency reduced from 45 days to 48 hours
Pricing re-calibration time cut by 68% — motor book re-priced 3 weeks earlier in Q2
Board now drills to geography-level loss ratios independently during meetings
Key metrics
Tech stack
Services
Client testimonial
“We used to approve a motor pricing change based on data that was 45 days old. Now we make the decision on data from yesterday. That shift alone is worth everything we paid for this.”
— Chief Actuary, Safa Insurance Group · Doha
Related work
SME lending platform · Riyadh
Daily 07:30 AST credit-portfolio briefing: 12-rule anomaly engine flags early-delinquency signals, auto-generates SAMA Fintech Unit-compliant risk summary PDF, and pushes WhatsApp alert to credit committee. Portfolio review prep 4 h → 19 min; delinquency early-detection rate rises from 41% to 89%.
Result:Portfolio prep 4h → 19min; early-delinquency detection 41% → 89%
Full-service digital marketing agency · Riyadh
Automated weekly client performance report: aggregates Google Ads, Meta, TikTok, and GA4 data, applies 9-rule anomaly engine, and delivers a branded Arabic PDF plus WhatsApp voice-note summary to each client. Report prep 3.5 h → 14 min per client; client churn drops 22%.
Result:Report prep 3.5h → 14min/client; client churn −22%
Long-haul trucking company (42 trucks) · Dammam–Riyadh corridor
Daily fleet-pulse report: GPS telematics anomaly engine (harsh braking, idle-excess, route deviations), automated SFDA waybill-compliance checks, and WhatsApp driver-alert system. Fuel cost per km drops 14%; SFDA compliance violations fall from 22/month to 1.
Result:Fuel cost/km −14%; SFDA violations 22 → 1/month
Book a free 30-minute audit, or message us on WhatsApp. We'll send a fixed-price proposal in two days.