Delivery Hero's internal tracking service eats 10x the event load Google Analytics ever saw and recovers 97% of tracking data. That's not a theoretical benchmark - it's what Alina Krasavina's team ships in production after dumping the vendor.
Why a Food Delivery Giant Canned Google Analytics
Third-party analytics aren't just a privacy headache. They're a cost sink and a scalability bottleneck. Krasavina's talk at InfoQ lays out the math: a simplistic, highly scalable architecture beat a mature SaaS product on both throughput and data fidelity. The team captured 97% of tracking data - meaning they lost only 3% of events, a rate competitive with or better than GA's typical sampling under load.
The Architecture That Makes It Work
No microservices spaghetti. No Kafka cluster the size of a small bank. Krasavina describes a deliberately lean stack: minimal moving parts, each component tuned for one job. The result? A system that absorbs 10x the previous load without cracking.
What This Means for Every Engineering Team
You don't need a billion-dollar analytics budget to own your user-data pipeline. Delivery Hero proves that a focused internal build can outperform a generic SaaS tool on the metrics that actually matter - throughput, cost per event, and data completeness. The vendor lock-in premium is dead when a food delivery team can match GA on scale and beat it on capture rate.
Next time your CTO asks why you're still paying for a third-party analytics service that drops 10% of events under peak traffic, send them this slide deck.
Source: Presentation: Challenging Google Analytics: Building a Scalable, Cost-Effective User Tracking Service
Domain: infoq.com
Comments load interactively on the live page.