The Best Way to Optimize Using Mediation

The main issue I’m having with using a mediation network like MoPub is the issue of assigning eCPM values for all the various ad networks I have mediating.

Looking through some docs, it says to keep the eCPM updated anywhere from weekly to monthly. Finding the true eCPM for all the various networks is difficult if I’ve already setup the waterfall order. What is the best way to calculate eCPM values while still maintaining revenue?

The best way I could think of was this:

  1. Every week, go back 7 days for each ad network filtering by Country (US - Most Popular)
  2. Calculate the average eCPM for the US, and use this value for the general eCPM values in Mopub

Now I know this won’t be totally accurate as the US eCPMs might not match overall eCPM, but is seems I need to use only one country. If I’ve already setup the order of the ad networks, the first ad network is going to get the majority of all the US traffic and get the highest eCPM, while the other networks get the subpar traffic and hence lower eCPM.

Using one country would hopefully level the eCPM values for all networks.

The other option is to turn mopub marketplace off, then set all the networks to the same eCPM values, and see which is the best after a few days. Doing this though will result in lost revenue, and would be very annoying to do on a weekly basis.

What’s everyone’s opinion, is the way I’m describing the best way to do it, or is there a better way?

Thanks.

Hello AndroidDev36, I am glad you started this thread as this is something that concerns many app developers, and I find proper waterfall ordering as one of the most important things when monetizing with ads.

From the experience we have gained with the publishers that are using our platform (I’m working at mediation company AdTapsy), we have definitely learned that there is no simple answer to your issue. What i keep repeating in the conversations with app developers is that every single app is unique and if something works in a certain way for one app, may not work that well for another. What I want to emphasize here (although I am aware I am not saying anything new here) is the importance of testing different scenarios in order to find the best working one.

Finding the best solution to this issue means to test and find:
[ul]
[li]Optimal frequency of eCPM updating (daily, weekly, every 3rd day etc.);[/li][li]How many days in backwards to be used in calculating avg. eCPM?[/li][/ul]

In order to accomplish this, you will need to pick few different scenarios and run them one by one while you are measuring the most important metric: eCPM. Example:

First we need to define 2 scenarios for testing.
Scenario 1:
Update eCPMs every day, while using 7 days in backwards to calculate avg. values.
Scenario 2:
Update eCPMs every day, while using 1 day in backwards to calculate avg. values.

In order to see which one performs better, we need to run Scenario 1 for some time (let’s say a week) and write down eCPMs you are achieving. Then we compare those results with the results from the next week where we have tested Scenario 2. The one that generates higher eCPM is a winner here.

The more tests you can run, the better results you will gain. And it might not work the same for your other apps, so you should repeat the process for every single one of them.

Regarding calculating average eCPMs I must mention here that using only single eCPM (US average eCPM) to rearange ad networks in a waterfall for all countries may impact your overall eCPM and revenue very badly. Ad networks have one weakness: they perform differently in different countries, but this weakness of ad networks is actually a big opportunity for revenue optimization when you are using multiple ad networks in a dynamic waterfall. By dynamic waterfall i mean different waterfall for each country. This is a truly money maker and should not be underestimated.

Now, we all know that calculating average eCPMs for each ad network and for each one of the total of 196 countries will take couple of hours and it will be extremely boring while doing it. Now imagine doing it every day! That is why we at AdTapsy have built unique in it’s kind autopilot where you manage such tests with only few clicks. Autopilot takes care of the rest. We do everything automatically:
[ul]
[li]Calculate average eCPMs;[/li][li]Update eCPMs;[/li][li]Pre rearage ad networks in a dynamic waterfalls for each country.[/li][/ul]

Screen Shot 2015-10-01 at 2.57.04 PM.jpg

Screen Shot 2015-10-01 at 2.58.07 PM.jpg

So, I am strongly recommending you to give it a try to our eCPM Autopilot and find the best working scenario for your app. I will be happy to assist you with the integration and testing, just email me at [email protected]. For more information regarding our mediation platform, take a look at www.adtapsy.com

Happy Monetizing!
Branimir

When are you guys going to integrate with Facebook audience network, adcolony, fyber etc?

We are actually supporting AdColony, although it is not mentioned on our home page. We are considering adding FAN soon. Currently we support:
AdMob, AdColony, AppLovin, RevMob, InMobi, Vungle and Chartboost.

If you add FAN, do let us know. I would be more than happy to give it a try. FAN, admob, vungle and adcolony combo