Don't miss
  • 1,898
  • 5,500
  • 5,760
  • 107

Think tracking metrics is hard? Well, it gets harder

By on June 20, 2013
playspace-avatar-EN

Should you always track metrics using a third-party solution, or are there some times when it is worth the investment building a home-brew solution? In this guest post, Alfonso Villar shares his experience at PlaySpace of managing metrics tracking across multiple platforms and territories. He found that once their games had reached a certain scale, traditional metrics solutions were not able to handle the multiple funnels at work in their business — so they built their own. Do you think he made the right decision?


All developers calculate, or should calculate, their game metrics to optimize the conversion funnel. When you launch a game in a single market, you can concentrate on analyzing different metric cohorts. For example, age, genre, whales, etc. However, the analysis becomes more complicated when this involves different countries, games and platforms.

Measuring real-time multiplayer game metrics across different audiences is not an easy task. At PlaySpace we have come across several problems and challenges that we have had to learn from and resolve as we go.

Problems and Challenges:

  • Multidimensional game metrics: Every piece of data that is manually added to a spreadsheet is a query into the database. Therefore, if we have to analyze 23 to 40 countries by 12 metrics per country and 9 different games per platform, we are faced with over 3.000 game metrics that have to be analyzed and updated by hand; an impossible figure to manage.

  • Timing is crucial: It is really important to be able to analyze any piece of critical data to make important decisions efficiently, so that we don’t repeat the same errors on a monthly basis. To improve the conversion rates you need to be able to make a quick analysis.

  • System saturation: Having all of the information solely in one database server, means that any queries to calculate game metrics could saturate the system. At the beginning we could manage this situation, however as soon as our games became more popular and we had a 24-hour user audience, we realized that we were not able to make queries to the system.

  • No information about return of investment: We have all the information about how much time users spend playing, how much they spend, and what they do while playing, etc. The problem occurs when you want to calculate if a specific player has been profitable or not as you need to know how much they have cost you. The acquisition team has  all of this information which is spread across different marketing channels. Therefore, another big challenge is to integrate the cost of user acquisition with the user monetization into the game to calculate the return of investment per user.

Solution:

To face these challenges, we had two options: hiring an external game metrics services tool, or developing our in-house metric system. Due to the fact we already had all this information on our platform we decided to develop our own system that could respond to any needs that arose as time went on:

  • As we needed the information on demand and accessible from any location, we opted to use a Web development framework with responsive design. This framework was directly integrated with our system and with different Graphic APIs such as Google Charts. This allows us to view real-time game metrics from any device and with any web browser.

  • At the same time, as this tool was directly integrated with our system, we had to make a copy of the database exclusively for queries and analysis. Any heavy query that is done does not affect the user experience.

  • Before developing our own metrics system, it was necessary to analyze and modify the data structure. All queries from different platforms, social networks, etc. were standardized and adding new queries would not take any additional development.

  • To calculate the return of investment we needed to develop a tracking system to identify which marketing channel each user had come from. Also we needed to integrate the cost of user acquisition to the system. To do that, it was necessary to integrate the Facebook API or even be able to add the acquisition costs manually on the system in case the marketing channel does not support an API integration.

We created our own game metrics dashboard (more than three dashboards per game in some cases), where we can breakdown the data by game, marketing channel, language, country, etc.

This solution means that we are able to make decisions the day before we launch a new marketing campaign or a new game feature. Because of this system we have seen that  “not all that glitters is gold”. For example, we found that some countries or audiences where cost of user acquisition was really good but the monetization was extremely bad. In the same way, we also detected that different game features have different user engagement, depending on the country or the audience.

As a conclusion, I would recommend not to be tricked by metrics that are evaluated individually. If you are able to make the most of the multidimensional game metrics and understand them in their entirety; these metrics will be a powerful tool that will not only improve your game but also improve your conversion funnel.

About Alfonso Villar

Alfonso Villar is CEO and Founder of Playspace, the leading developer and operator of evergreen leisure online games for the Spanish and Portuguese speaking world.
  • Eric Seufert

    Nice post. To add to the requirement that game data be available “on demand and accessible from any location”, I’d say that increasingly “from any device” is important. Open-source visualizations libraries like d3.js allow for this: graphs render the same on an iPad as on the desktop, and authentication remains centralized (the article astutely pointed out the necessity of this) as it’s on the web.

  • John Griffin

    Good article. I appreciate the pain you went through. However, I don’t necessarily think that game developers should build this themselves. There are 3rd party solutions that are based on exactly the things you had to do (and presumably learn) yourself. A lot of the things you describe above are actually quite basic. Business Intelligence is a specialist field and very very well developed. I would seriously encourage developers to focus on building great games and don’t go building tools for which there are already great solutions.

  • Can Bakir

    Dear John,
    What 3rd solutions would you recommend? Can you list a few pls.

  • John Griffin

    Sure … best known are probably Kontagent, Flurry and Playnomics. Ones I recommend are Conker.io and Swrve. I can’t reply to this without also pushing you towards my own company GameSparks which may be what you need depending on your requirements. We are more about providing a full end-to-end dashboard as part of a fuller Backend-as-a-Service platform rather than hardcore behavioural analysis.

  • Can Bakir

    Thanks for the insight all of this was very useful. Also I had a chance to review your website. I on the otherhand work at Peak Games. Our games focus is Facebook at the moment. We have a few mobile extensions of our games, but we havent truly experienced the mobile yet. Is it possible to continue this conversation further? Whats the best way? my linkedin is http://tr.linkedin.com/in/canbakir

  • Pingback: 如何使用参数优化游戏的转换率 | GamerBoom.com 游戏邦

  • Pingback: 如何使用参数优化游戏的转换率 - 爪游控