Waterfall vs. Unified Auction Metrics: How To See The Whole Picture [VIDEO]

Posted on June 07, 2018
By Matt Kaplan, Content StrategistContent Strategist

As noted business consultant Peter Drucker once famously said, “If you can’t measure it, you can’t improve it.” App publishers today now have dozens of metrics and key performance indicators to track, but are they the right ones? What if they aren’t getting a full and complete picture of monetization performance?

In our latest Whiteboard Wednesday video, Andrew Gerhart, our VP of Publisher Platforms, highlights the limitations of standard waterfall metrics and how unified auctions provide you with a more complete picture of your monetization ecosystem. Not only do unified auctions yield higher CPMs and improve user experience, but they give you a wider and deeper view of your monetization landscape!

Video Transcript:

Welcome to another Whiteboard Wednesday. My name is Andrew Gerhart. I’m the VP of Publisher Platforms at InMobi. Today we're gonna be talking about metrics, the lifeblood of ad monetization. And the importance of transparent metrics versus standard metrics.

So, to start, let's take a look at the standard ad call. You start off, the app makes a call to the SDK, the SDK makes a call to the ad server. Different platforms label this differently. It could be an ad opportunity. It could be an ad request. At InMobi, we call it an auction. So now we have one auction. Depending on the way that the buyer is integrated, where it’s server-to-server, TAG or SDK, where the buyer is called from will vary. But either the ad server or the SDK makes the call to the buyer. So with a standard waterfall model, one ad calls the buyer, and now we have an ad request.

Let's assume, because things don’t always work perfectly, that the buyer doesn't return an ad or something doesn’t happen, and now we don’t have anything there. The waterfall model will go down to the next buyer and make another ad request. So now we now have two ad requests.

Let’s assume that they return an ad. We don’t have anything logged here, so we don’t have another event. The ad is sent back to the SDK, back to the device, we try to serve the ad. Unfortunately, the SDK times out. So then we go back to the waterfall, we go to buyer #3. Again we have an ad request. That ad is sent back to the ad server, back to the SDK, back to the app. We attempt to serve that ad again. Luckily now, we serve that ad, and we have a delivered impression.

Because we have an impression, we can calculate our fill rate, we can calculate our revenue, and we can calculate our eCPM. Unfortunately, most platforms do not highlight the issues that happen with failures and timeouts. We can’t see if there were any other bids, bid prices, wins, win prices, timeouts and failures to show, and because we don’t know who bid within the waterfall of the auction, we can’t calculate our use rate. And so you see, we’re running blind with regards to our monetization stack.

Now, let’s take a look at what transparent metrics look like. Same ad call, so the app makes a call to the SDK, the SDK makes a call to the ad server, and so we have an auction. Now, with InMobi’s unified auction, an ad call is made to all buyers. And so now we have multiple ad requests going out to the buyers in the auction.

Unfortunately, buyer one in the same scenario has a connection error, and so now we have a pre-bid failure. They didn’t get to bid within the auction because they timed out, and so now we know what happened with that particular buyer, so now we can have an informed conversation with that buyer and resolve the issues. So they’re out.

Luckily, buyer 2, 3 and 4 all return a bid. Now we have three bids as a result of this auction. Along with those bids, we know the CPM, we know their bid prices. Now we can see the bid landscape for the buyer’s we’re working with.

But, three buyers or three ads can’t serve, so the ad server needs to do some work to determine the highest paying ad based on the logic that the publisher has set, as well as the ad server has set. And so now we have to choose a buyer, we have to choose a winner. Now we have one win, and now we know the win price. This becomes important in having conversations with your buyers around CPMs, bid pricing, and how to get them to increase their price in serving ads. So, we now have one winner.

Let’s assume that buyer #3 won the auction based on having the highest CPM. That ad is sent back to the device, but unfortunately, similar to scenario #1, it times out. Now, we have a failure to show, what we logged out of that. And, we fail over to the next highest bidding ad. So now we go to buyer #4. Buyer #4 serves the impression, and now we can calculate our fill rate, and we can also calculate our use rate, revenue and CPM.

You’ll see here there’s a distinct difference between the standard metrics that are found in most mediation and ad serving platforms, and the transparent metrics that are found within auctions and InMobi’s unified auction.

Really, really important to dive deeper into complex monetization stacks. At scale, publishers are working with lots of third-party technologies, networks, exchanges, tags, SDKs, and so it’s really important to have great insights into what’s happening so that you can diagnose and grow your revenue.

Thank you for your time, and we’ll see you next week.