In ad tech we use the word ‘optimisation’ a lot. We optimise for performance, for ad spend, and for revenue. But when we optimise, we do so to win. This means that someone else loses. In this piece, Hilit Vainberger (pictured below), CRO, Cignal.io, discusses why, as an industry, we need to start thinking collectively about the greater good of the entire ecosystem and not just what’s good for our organisation this quarter.
In the last few months, there has been a lot written about supply and demand path optimisation. But when the supply and demand sides each try to optimise, we end up with an arms race escalation and no one really wins.
Let’s start by looking at the supply side. For every ad query an SSP sends out, only about 2-3% of DSPs will respond. Among all DSPs connected to an SSP, only 10-25% will actually make a bid. This means that 75-90% of the SSP bandwidth is being wasted on DPSs that will not contribute to the SSP’s (or the publisher’s) revenue. And with header bidding, DSPs are queried for every auction, sometimes 15-20 times, which results in a lot of wasted bandwidth.
Why do so many DSPs connect to an SSP? Perhaps they once bought impressions from that SSP, or they want to have the option of acquiring impressions from that SSP sometime in the future. In today’s era of marketing automation, once a DSP connects to an SSP, it doesn’t feel compelled to disengage.
A second challenge on the supply side is the fact that each publisher works with multiple SSPs. Publishers do this because they’re looking for the greatest total revenue, but it creates a situation where their demand partners end up bidding against one another for the same user. No sales manager would assign multiple sales teams to pitch one client. This would result in a lower price for the company brought on by the competing teams and perhaps even an annoyed client. But, once again, having multiple SSPs selling the same user adds complexity and bandwidth, which also cost money.
Unfortunately, the situation is no better on the demand side. Like publishers, marketers also work with multiple partners (DSPs on the demand side). This creates similar redundancies where many queries are sent out to the same brand for the same user, resulting in the marketer paying more for the impression due to the unnecessary competition.
As an industry, we need to start thinking collectively about the greater good of the entire ecosystem and not just what’s good for our organisation this quarter. We need to start thinking about how we achieve optimal results without relying on optimisation by automation.
The supply side needs to require DSPs to pay to gain access to user queries. This will (1) push out the tyre kickers who don’t even kick the tyres, which will reduce publisher latency and SSP bandwidth while (2) increasing revenue. There also needs to be greater transparency regarding how auctions operate (first price, second price, hidden fees, etc.) so that buyers and sellers completely understand each other.
Beyond this, the supply side needs to consider the pains of the demand side. How are issues like header bidding and fraud negatively impacting demand clients?
The demand side needs to stop optimising by automation and start thinking about optimal digital ad ecosystem performance. Do we need to see every auction bid in order to understand how to bid or can we use technology to bid more intelligently, saving bandwidth and time costs. And is all of this information really empowering smarter decisions in real time? Is the demand side so confident that they have the right publisher and user data in order to win the right auctions?
Ultimately, both the supply and demand side need to consider each other’s needs as we look to create a buying and selling ecosystem which is more efficient. What are each side’s core needs and what can we do without in order to reduce the clutter and increase the efficiency of digital ad marketplace.