I’m not sure what it is that predisposes marketers towards creating complexity. But while peers in operations, supply chain, and product development strive daily to simplify the complex, marketers sure seem immensely prone to creating absolutely epic Rube Goldberg machines in response to even the simplest problems. For evidence of this, look no further than a recent article seeking to persuade marketers that “multiplatform” buying works – that is, running multiple DSPs and programmatic buying platforms on a single campaign.
To be clear, it’s true that each DSP has its own methodology, optimization approach, and technology. But this simply means each DSP has a different way of showing the same customers the same ad creative (and deciding on the right combination of the two). Some DSPs may be more reliable and efficient, manage frequency more consistently or be better at differentiating between real humans and web bots. These things matter – a lot – but they don’t improve as you add more platforms. Ultimately, the author’s assertion of multiplatform superiority rests on the “more is more,” “these ones go to eleven” logic that costs marketers millions in tech fees and wasted impressions while creating no incremental value. So what’s the reality when it comes to running multiple DSPs?
- You might get more impressions by adding another DSP, but not more reach. The author argues that buying at scale for targeted campaigns requires multiple platforms. The logic behind this theory is that adding another platform somehow enables the advertiser to find more people, even in the most niche of segments. The problem is that what additional DSPs allow you to add is impression volume (likely in the form of more frequency), not necessarily reach.
Want to actually reach more people? Then you have a few options: ask your DSP to add more supply partners (although most major DSPs pull from exactly the same pool of SSPs and exchanges), get more data (either through first-party, third-party, or partner sources), or expand the definition of whom you’re trying to reach. In some exceptional cases adding a new DSP may allow you to access new inventory providers, thus finding more reach (for instance, Google Bid Manager can’t access Facebook’s exchange), but for the most part there are better, simpler paths to finding more people to target.
- You’re not going to win trying to optimize the optimizers. The author contends that a multiplatform operation can drive incremental performance by optimizing the optimizers – finding the optimal allocation of budget across platforms by analyzing each DSP’s individual algorithm while also reverse engineering how SSPs optimize against those DSP algorithms. If it sounds complicated, it is. And the author admits as much, noting that managing continually evolving technical relationships and optimization methods requires “highly-skilled traders and analysts,” essentially using humans to try to out-optimize the machines. It’s also at best a zero-sum game. Sometimes the optimizing the optimizer approach will win and sometimes it will lose, but with their vastly superior processing capabilities, my money is on the machines over the humans.
Ultimately the case for “multiplatform” rests on a few faulty assumptions that, even if true, still don’t outweigh the benefit of doing what only a single integrated platform can do: ensure that ads are shown to real people your brand cares about, at a reasonable frequency, and in a place where those ads can actually be seen. The fact is, by simplifying the technical complexity and focusing on the basics, marketers can save millions, achieve massive global reach and ultimately focus on the stuff that really makes a difference – creating messages and experiences that resonate with their target customers. That’s something no machine, no matter how complex, can do.
