Jonathan Mendez is CEO of Yieldbot, an intent-based ad platform. Follow him on Twitter at @jonathanmendez.
Since the first ad exchange, the idea has propagated in the industry that as the buying and selling of digital advertising becomes automated — as machines replace the people — the industry will enter a glorious period of growth and liquidity.
The reality is much different. Agencies have now lost their ad server margins and ad tech companies built to exploit this media shift have razor-thin to non-existent margins. Personal relationships still rule deal-making. Just take a look at the size of some representative businesses through this programmatic “value” chain.
- Cadreon (agency trading desk): over 300 employees
- MediaMath (demand-side platform): 140 employees with 53 open positions
- AppNexus (buying platform): 1,000 employees within two years
- PubMatic (supply-side platform): 325 employees
It seems these automated media buying and selling companies need a shitload of people to make them run. Ideally, none of these people are salespeople, though we know that is not the case. I would argue on a dollar-for-dollar basis there were fewer people needed to support the industry back in the ad network heyday.
So if the new pipes VCs have spent $6 billion building require even more people to operate them than the old ones, where does that leave the automated future of online advertising? It leaves it with the one thing that can truly be automated, eliminate headcount and improve margin: performance optimization.
Algorithms are going to do a much better job optimizing than a person could ever do and will do it at scale. Performance optimization can be automated because it truly has all the ingredients necessary to benefit from algorithms that can take empirical data to quantify complex relationships, identify patterns, capture probability distribution, make intelligent decisions and learn from them.
Ad tech companies stacked with 25-year-old account managers and reporting analysts are not going to stand a chance in the next wave of intelligent optimization agents backed by terabytes of daily data fueling complex self-learning algorithms. Nor should they for our industry to reach its full potential and deserve the name ad “tech.” In the coming wave of auto-optimization and artificial intelligence, companies in our industry will not boast about how many employees they have but how few.
More in Media
Why Amazon and YouTube pitched operating systems, not just TV inventory at this year’s upfront
Negotiations over identity, infrastructure, AI-driven buying take place as much as programing.
The Economist prepares for a two‑track internet: one for humans and one for AI agents
The Economist is testing agent-readable versions of content that already sits outside its paywall, as it prepares for “two versions of the web.”
Amazon bets creator video podcasts can be the next TV network – if it can fix measurement
Amazon’s Upfront presentation leaned into its podcast offerings, which the company believes are the next generation of TV networks.