“The Sell Sider” is a column written by the sell side of the digital media community.
Today’s column is written by AJ Okereke, head of revenue technology at Graphiq.
Header bidding is a sell-side technology that allows advertisers to use real-time pricing to compete for a given impression within a publisher’s programmatic auction.
This influx of real-time advertiser price competition leads to upward pressure on pricing for a given ad impression. The upward price pressure helps close the gap between the winning bid and closing bid in a second-price auction (see graphic below).
This is a stark difference from how ads were previously sold programmatically, as it reduces the prevalence of an information asymmetry and infrastructure advantage that once privileged Google’s Ad Exchange advertiser demand over all other ad exchanges.
But as publishers add more header bidding partners, they must answer one critical question: Is there an optimal number of exchange partners to work with via header bidding before experiencing diminishing returns?
I believe that additional exchange partners provide incremental gains in revenue, and I know other publishers share the same sentiment.
Source: PubNation
Measuring Partners
Many publishers not only work with numerous exchanges, but based on my industry observations, some work with more than 10 exchange partners.
Does working with multiple exchange partners provide sustainable revenue uplift? Yes. Every publisher is different in size and content type but I’ve learned, anecdotally, of many publishers seeing revenue gains of 30% to more than 50%.
For publishers unsure about the optimal number of header bidding partners to integrate, it’s possible to track individual partners’ impact on revenue with A/B testing. Publishers should first establish a control group, consisting of their ad stack as is, without any additional partners, and a variation group, comprised of their ad stack with one additional partner.
An A/B framework can independently test the effect of each additional partner on overall revenue per visit, holding traffic constant in each group. Thus, with a statistically significant sample size, it is possible to measure average gains in revenue per visit for each additional partner. It is important to keep in mind that it takes time for partners to learn and optimize toward a particular publisher’s audience, so gains in revenue per visit can be realized over varying periods of time.
The Results
Depending on the findings, publishers should be able to calculate the revenue attributable to header bidding as new partners are added and determine when a point of diminishing return has been reached.
This insight can also help inform auction optimization strategies, including competitive price floor strategy, partner latency improvements and auction timeout efficiencies.
These actions may set the stage for even more advanced header bidding strategies, such as exchange throttling, where an exchange may be cut from the auction if no bid is received on the first page view.
Another advanced tactic is exchange staging, where historical bid trend data is used to dynamically insert an exchange at a certain point within a user session, perhaps the second or third page view instead of the first. This can lead to overall increases in yield for both publishers and exchange partners.
Publishers can also employ discrepancy multipliers, where an exchange’s bid is discounted by its respective discrepancy amount so that each partner competes at its true price within the ad server. This will help publishers achieve true and fair auction where all exchange partners compete on a level playing field. This will also increase accuracy of revenue reporting.
Finally, publishers should experiment with Google Ad Exchange Dynamic Flooring, taking the highest bid returned by a header exchange, multiplying it by an appropriate percentage based on historical bid data and using that amount as a floor for each individual impression in Google’s Ad Exchange.
Any combination of these tactics will help publishers embark on a path to achieve the fairest auction possible, with a goal of having advertisers pay the true value for a given impression.
Follow Graphiq (@GraphiqHQ) and AdExchanger (@adexchanger) on Twitter.
