Building software in the digital advertising world is a uniquely punishing endeavor. Unlike traditional platforms, an ad tech stack doesn’t just serve static content—it handles hundreds of billions of real-time auctions every single day, all clearing in under 200 milliseconds.
When you scale a platform to that degree, the operational, structural, and management challenges multiply exponentially.
In a recent episode of The Yield Doctor, host James Deaker sat down with Nate McCoy, an engineering leader who has spent over 25 years navigating the trenches of ad tech development. Nate has guided engineering teams through critical eras at Aquantive (Atlas), Microsoft, AOL, Verizon, and Yahoo.
From surviving the dot-com crash to decommissioning massive legacy ad stacks, Nate shared his core rules for building, integrating, and maintaining focus in ad tech engineering.
1. Shift from “Waterfall” to Continuous Delivery
In the early days of digital advertising, engineering teams relied on traditional “waterfall” project management: you spend months planning, months coding, months testing, and then you execute a massive release.
But as the internet transitioned into a 24/7, live production environment, the waterfall model became a liability. If your development cycle takes six months, the entire industry could move past you before your code ever goes live.
To survive in modern ad tech, platforms must embrace continuous deployment and small-batch changes. By releasing code incrementally, engineering teams minimize risk. If a minor deployment goes awry, it can be instantly rolled back before it costs the business millions of dollars in downtime.
2. Streamline Your Developer-to-Operations Ratio
Operational efficiency isn’t just a tech metric; it’s a margin driver. In the nascent days of ad tech, platforms were incredibly resource-heavy. It wasn’t uncommon to see a ratio of two software engineers for every one production operations person. That overhead severely ate into business margins.
Today, heavily scaled production environments require automated tooling. A healthy, modern engineering organization should aim for a ratio of 10-to-1, or ideally even 20-to-1, developers to explicit operations staff. Modern operations professionals are software engineers themselves, writing code to build the automated infrastructure that allows the rest of the development team to deploy safely.
Similarly, shifting away from a separate, massive quality assurance (QA) or testing discipline toward a model where developers write and test their own code removes communication bottlenecks and accelerates the release pipeline.
3. The Golden Rule of Tech Mergers: Pick One Stack and Abandon the Rest
When digital media companies go through mergers and acquisitions, they frequently inherit a messy puzzle of overlapping technologies. At one point following the Verizon, AOL, and Yahoo mergers, their combined ecosystem was juggling 14 distinct advertising technology stacks.
When leadership tries to integrate or blend multiple platforms together, they almost always create a Frankenstein’s monster. Merging stacks devours massive amounts of engineering time, introduces extreme opportunity costs, and results in a compromised product that leaves the business lagging behind nimble competitors.
Instead, execute the strict strategy of picking one single stack and abandoning the rest.
- Pick your best platform: Choose the superior technology or the one that aligns with your long-term vision.
- Reallocate your talent: Move all of your incoming engineers onto that single platform to make it better.
- Accept short-term breakage: Expect minor revenue disruption or customer friction from the systems you shut down. Accept it as the price of buying your engineering team’s time back to build features that drive future growth.
Trying to execute a massive migration with a corporate mandate of “zero revenue breakage” forces engineers to cut corners and delays the project for years. Don’t fall victim to the sunk cost fallacy.
4. Solve the Product Bottleneck with Radical Focus
The number one problem in ad tech engineering isn’t a lack of talent or capital—it is a lack of focus.
Ad tech platforms are immensely complex research and experimentation engines. They require PhD-level data scientists and deep systems engineers to constantly interface and run simultaneous machine learning experiments for targeting, forecasting, and auction dynamics.
Because the ecosystem has so many moving parts, businesses often over-hire product managers. When an organization has too many senior product managers, each individual naturally fights for their own feature “win” to advance their career. This dilutes engineering resources, forcing developers to time-slice their focus across dozens of minuscule projects.
True marketplace success comes from radical focus. Users don’t care if an ad platform has a massive library of secondary features if the core auction engine or reporting UI doesn’t make them money. Choose a few major initiatives, execute them flawlessly, and protect your engineering resources from feature creep.