3 Takeaways from Gamecamp 2019 – Man vs. Machine + 2 more learnings

Header banner with the gamecamp logo and the words “3 takeaways from gamecamp.io” written next to a game contoller icon

I recently gave a speech about Ad LTV in Google’s Gamecamp event in Warsaw. It was a great presentation and we already shared it with you in this blog post. I was impressed by the level of content and wanted to share some of my thoughts about the presentations. 

1 – Man vs. Machine – Who is the better frog

The keynote presentation was about the role of Machine Learning in games. RJ Mical was the speaker. If you don’t know him, he is one of the creators of the Amiga computer, Atari and a bunch of other gaming oriented projects.

The presentation was an intro to machine learning (ML) combined with insights about how ML can serve the Game creation side and how to use it for optimizing Game design, monetization, etc. However, for me, one of the interesting moments was when RJ mentioned that he has seen ML train itself playing frogger and totally mastering the game after only 6 hours and getting to a stage where it’s basically unbeatable.

Now, this made me think about betting. This will be a good time to mention a company called Skillz. Skillz is a successful company who serves game publishers. Their offering allows skill-based games which are naturally single players to offer matches in which players can bet money and more importantly make money.

How is that even related? It kind of is because there are many games in the market today that include a completely legitimate way to make money and with the right ML technology you might be able to create a player that never loses. Maybe there is someone already doing that. Probably there is.

Takeaway #1 – never bet inmobile games if you can’t see the person you are betting again and ensure he is human

Segmentation is an obvious win but somehow still under utilized

Michal @ Pixel Federation gave a super interesting presentation about Segmentation and how to use it to drive more revenue from payers. Michal presented 3 case studies about this and discussed aspects like purchase intervals and how to adapt the offering to segments without alienating the community. Consider 2 options:

  1. Segment A – $1 buys 500 coins vs. Segment B – $1 buys 1,000 coins
  2. Segment A – $1 buys 1,000 coins vs. Segment B – $2 buys 1,800 coins

The 1st option is unfair and creates a balancing problem which in turn makes the fan base unhappy. The 2nd option, on the other hand, maintains the value of the coin and focuses on how the offer is presented.

Michel from pixel federation on stage showcasing the segmentation experiments ran by them to optimize monetization

To me – segmentation is such a no brainer tool. Treating all users, in the same way, must be wrong. It’s always surprising for me how many companies are not utilizing it. Specifically, there is a huge opportunity I see in utilizing segmentation for ad-funded apps. There is a big chunk of users who never monetize on install campaigns and should really be segmented and monetized differently. At the same time, there is also a big opportunity in defining inventory packages and working with SSPs to attract demand for these.

Takeaway #2 – segmentation is a must have especially for ad-funded publishers

We will be happy to discuss better segmentation for ad monetization over a call


Keeping ad networks true to their promises

Nate @ Kolibri was talking about the implementation of ads in their Idle tycoon games. Kolibri is using Ironsource for mediation and Looker as a dashboard. One of the interesting views they implemented was a per network promised vs. actual.

The context here is that the industry is moving to stacked waterfalls. Here is a quick explanation of how stacked waterfall normally works:

  • You set up a few price floors per network – these are associated with placements
  • This creates more predictable eCPM levels as each placement is guaranteed to generate a minimal eCPM (the floor)
  • The waterfall calls the floors from high to low to ensure optimized monetization

There is a better explanation of stacked waterfalls as well as some tricks to optimize it in the case study we did with tripledot.


Nate’s point was that this entire system relays on the networks actually paying the price floors. Apparently, there is always some level of the gap between the promised CPM and the actual one. You can basically calculate a “lost revenue” for the entire waterfall as the diff between the hypothetical revenue promised via the CPM floor and the actual revenue you got paid by the network. Tracking this “lost revenue” and working to minimize it can pay big dividends in overall ad monetization.

Takeaway #3 – Auditing ad-network promises is becoming an increasingly important part of monetization measurement.

Feel free to share:
Previous articleTop 10 Hyper Casual Game Mechanics
Next article5 Reasons Why App Devs Should Love Reporting Automation
Raised in the Kibbutz and reborn in the city, Yaniv is a certified entre-parent-neur. When he’s not busy doing SEO, content marketing, administration, QA, fund raising, customer support… [stop to breathe], you can find Yaniv snowboarding down the slopes of France and hiking with his kids. Yaniv holds a B.Sc. in Computer Science and Management from Tel Aviv University. He is also an avid blogger and a speaker at industry events. Before SOOMLA, Yaniv co-founded EyeView


Please enter your comment!
Please enter your name here