Scaling AdTech Engineering: Building a Culture of Ownership, Optimization, and Impact

In my last three posts, we explored the foundational pillars of adtech engineering: managing trillions of real-time ad auctions, optimizing networks for efficiency at scale, and designing software architecture that withstands the demands of billions of users. But driving success in adtech isn’t just about technical prowess; it’s about fostering an engineering culture that thrives on ownership, performance, and relentless optimization.

At Quantcast, we handle over 5 million transactions per second with sub-50ms latency — a blink of an eye. These numbers underscore the stakes and highlight why an engineering culture centered on scale, cost, and customer obsession is crucial. We optimize continuously, balancing immediate performance with long-term impact.

Guiding Principles in a Performance-Driven Culture

The reality of large-scale adtech? It’s all about trade-offs. Depending on the systems we build, we make different optimizations. In bidding, we prioritize scale, cost, and latency while being willing to sacrifice completeness. Our pixel infrastructure, by contrast, is optimized for completeness and speed, while keeping costs as low as possible. For backend compute clusters, we have two configurations: one that optimizes for high throughput at low cost and another that focuses on high speed.

What’s unique about our approach is that many of our best ideas come from the ground up. Instead of architects designing solutions in isolation, our most successful designs evolve through innovation from all levels of product and engineering starting at grad engineers. Multiple product lines we operate today started as Hackathon projects.

This pattern of innovation is intentional. We encourage our engineering teams to deeply understand our domain and their areas of ownership and then create tailored solutions. Our guiding principles — Win with Customers, Own it, and Velocity — shape our approach to problem-solving and empower engineers to think independently.

The Culture of Empowerment and Ownership

An empowered engineering team is a high-performing one. At Quantcast, engineers are encouraged to take on challenges as owners of their projects. Here’s what that looks like in action:

• Prudent Risk-Taking, Backed by Support – Innovation in a high-stakes environment like adtech requires calculated risks. We foster a culture that encourages experimentation, and with regular 1-1s, performance reviews, and promotion discussions, we create a feedback loop where engineers align their ambitions with actionable guidance. This support helps them navigate challenges thoughtfully. And when things inevitably go wrong, we have a reliable incident review process to learn the right lessons and improve continuously.
• Optimizing for Velocity and Quality – Speed and quality aren’t mutually exclusive. We constantly iterate on our processes to shorten our feature cycle time. Our architecture is designed for rapid testing in production, and most projects include rollback options, so we can quickly validate and adapt as needed. A rigorous code review process also ensures we maintain high code quality and consistency.
• Clear Pathways for Growth – Our growth framework aligns with the engineering culture we aim to cultivate. In promotions, we look for candidates who demonstrate capability at the next level, both in “what” they achieve and “how” they do it. For engineers who work on large-scale systems, this often comes down to optimizations, cost efficiency, and quality trade-offs. This alignment allows our Career Growth process to reinforce the values we strive for as an engineering team.

Measuring Success Through Iteration

Building at internet scale is as much about what we measure as how we deliver. To optimize effectively, we rely on data-driven decisions, backed by rigorous measurements that guide our improvements. We upload trillions of measurements per data to time-series databases (think Datadog/Prometheus) on a daily basis. Every enhancement, from leveraging our network backbone to optimizing load balancing and UDP, is validated by data, to ensure that we are truly moving forward with every step. We don't get it right 100%, but when we do, we are confident it is making a positive difference.

Optimization isn’t a one-time effort; it’s a mindset that keeps our systems responsive and resilient. In adtech, perfection is elusive, but we believe that relentless improvement drives lasting success. We measure, optimize, and iterate — and keep moving forward, because at internet scale, adaptability is our greatest asset.

In Summary

• Pick your trade-offs wisely
• Empower Teams, Embrace Imperfections
• Measure, Optimize, Iterate

Thoughts?