Uken Games and Publishers Clearing House A/B Tested MoPub’s Advanced Bidding

Twitter’s mobile ad platform delivered higher ARPDAU for both publishers

MoPub will conduct similar A/B testing with more publishers - Credit by Twitter
Headshot of David Cohen

Twitter’s MoPub mobile ad platform provided an update on its advanced bidding process, which it debuted in alpha in December 2017.

MoPub is connected to over 180 demand-side mobile platforms, including the Twitter Audience Platform, and participants in its advanced bidding initiative include Facebook Audience Network, Tapjoy and AdColony, with another 10 ad network bidders in active implementation and testing phases.

A/B tests with two publishers resulted in average revenue per daily active user rising between 5% and 15%, as well as increases in filled supply, supply access for all programmatic buyers and an increased share of wallet for all programmatic partners.

Toronto-based game publisher Uken Games—developer of Who Wants to Be a Millionaire? and Jeopardy! World Tour—activated Facebook Audience Network and Tapjoy as advanced bidding networks, testing them against non-bidding networks and the MoPub Marketplace. It was aiming for revenue uplift and greater efficiency.

And direct-to-consumer entertainment, commerce and marketing company Publishers Clearing House, pitted Facebook Audience Network, AdColony and Tapjoy against non-bidding networks and the MoPub Marketplace with the aim of simplifying complicated “waterfall” management and maximizing revenue from each ad opportunity.

Twitter said it always recommends that publishers look at metrics on a per-user basis, rather than by total payout, calling ARPDAU “the north star metric for publisher optimization.”

The social network added, “Even when measuring ARPDAU, it can be difficult to measure the effects of advanced bidding by simply tracking ARPDAU before and after a publisher has switched to bidding. Network bidding algorithms can take time to calibrate, and there are plenty of other variables that can lead to under- or overstating the true impact of mobile app bidding.”

Twitter also outlined the six steps that went into the advanced bidding test by Uken Games and Publishers Clearing House:

  • Identify ad units to opt into the test: Twitter and its partners looked for slices of inventory with unique waterfall configurations. They considered ad units with at least two advanced bidding networks, MoPub Marketplace and enough volume to arrive at a conclusive result. They also sought ad units that were well optimized to see how they stacked up against the best waterfall competition.
  • Duplicate the ad units: The waterfalls of the control group ad units were duplicated into the new test group.
  • Create new placements for remaining networks in the waterfall: It was necessary to collect performance data from all demand sources in the waterfalls, as well as to distinguish which revenue came from the control group and which revenue came from the test group. Identically configured placements were created for each non-bidding network, and those placements were not used in any other ad units.
  • Divert 50% of users making ad requests into the test ad unit: All users in the publishers’ apps were making requests to the control group, so MoPub had its servers divert requests by one-half of the users to the test group.
  • Make changes to the test group to enable advanced bidding: New advanced bidding entries were created for each bidding network, while non-bidding networks were left unchanged.
  • Run the test: The test ran for approximately two weeks, allowing bidding networks to calibrate their models and allowing non-bidding networks to shift their performance according to the different supply mix across the two groups. All non-bidding networks had MoPub’s auto CPM (cost per thousand impressions) enabled to automatically update network performance. Once everything stabilized, the test ran for another two weeks.

Publishers Clearing House saw a 15.3% leap in ARPDAU, while Uken went up 4.8%.

MoPub also summarized its findings from the test with the two publishers:

  • Advanced bidding drives positive results, including a less complex waterfall, incremental ARPDAU gains and a platform set up for further success when new networks are onboarded.
  • Demand competition maximizes revenue: MoPub wrote, “Advanced bidders are able to compete with a real-time price for every single impression, leveling the playing field for all real-time buyers and maximizing competition. Publishers should sit down with their top demand sources that aren’t yet bidding and ask for their timelines to make the transition.”
  • Bidding partnerships are crucial, and advanced bidding gives bidders the first look at all ad requests, upping their ability to meet their advertisers’ key performance indicators. They can also compete with a real-time price for every single impression, leveling the playing field.
  • Time is a currency, and bidding can provide significant savings. Compared with traditional waterfall mediation, advanced bidding provides the opportunity to reduce time spent on ad network management and ad trafficking operations.
  • In-application bidding is a long-term investment, not a short-term chance to boost revenue. Different publishers may see positive results more quickly or more slowly, so it’s important to give your demand partners time to adjust.

MoPub will conduct similar A/B testing with more publishers and work to measure changes in time-to-fill for each ad request.

Twitter added, “In-app bidding is a game changer for mobile app publishers and ad buyers, and at MoPub, we’re excited to continue investing in our advanced bidding product. We’ll be using publisher feedback to iterate on the setup and management flows, and actively working with additional network partners to onboard them to our platform as bidders.” David Cohen is editor of Adweek's Social Pro Daily.