Jan 1, 2023
Most mobile game companies make their money by selling ads.
Normally, 30-60% of that ad revenue comes from waterfall ad auctions, with the rest coming from bidding, a parallel process. The highest bid from either the waterfall or the bidding buys the impression.
Waterfalls are a form of Dutch auction; rather than participants bidding prices up, the auctioneer brings prices down. In a waterfall, an impression is sold by offering it to specific ad networks at specific prices, in a pre-determined order. So for instance, you might auction an impression by offering it to Google at a $300 CPM, then Unity at $280, then Chartboost at $250, and so on. Each of these offers are 'placements', and there can be around a hundred placements in a waterfall auction.
Auction processes are handled for mobile games companies by a mediation provider, like Applovin MAX or Unity LevelPlay; their software runs the waterfall auctions.
Through one mediation provider, mobile game companies sell ads to 10-15 ad networks. Each ad network is a market-maker: they compete to be granted waterfall placements, but they also need to acquire demand from advertisers. You'd think this would all be terribly efficient; but things aren't that simple.
Waterfalls have to be set up by hand. Analysts decide which prices to offer which ad networks, creating the rules by which the mediation software sells impressions. But how are they meant to do this? Why offer first to Google at $300? Why not $310? Why not another network? And if that setup was right before Christmas, will it still be right in January?
Solving this problem is a key for mobile gaming companies. The optimal waterfall setup could generate 10-20% higher ARPDAU (average revenue per daily active user), but demand from mobile advertisers is constantly shifting, and analysts have thousands of waterfalls to optimise. It's an incredibly inefficient part of the mobile gaming value chain - and hence a huge opportunity.