Betting on HTML5 for game development – Part 3

html5In the previous article we talked about the portability benefits of HTML5; its ubiquity in the native mobile world and on the web is a value that no other technology can claim without the help of proprietary plugins.

But HTML5 also offers intrinsic advantages particularly during the development cycle – advantages that may lower the development costs of your game projects.

Play testing is an important component of quality control during the game development phase. Improving the gameplay and balancing the difficulty of a game is a tricky task that requires a lot of iterations and input. For this reason, it may be crucial to provide early access to a number of players during the development cycle.

However, unless the project has the chance to have high anticipation, it can be hard to find players willing to install the alpha or beta version of an upcoming project. For mobile apps the process is even harder as the app may be confined to ad hoc beta areas with limited access requiring the tester to provide an email address or even a testing device ID.

Paying for crowd sourcing services can also be expensive and these users may not behave exactly like real players (some important key metrics such as the dropout rate may be biased).

This is where HTML5 can help. HTML5 applications do not require installation and giving access to the game is a simple matter of sharing a link online. This gives us many opportunities to reach out to real players and leverage their play sessions with appropriate analytics.

Let’s illustrate with a concrete example. A link to the demo of Jelly Slice was shared on several forums as well as directly sent to interested players during development. By doing this, I was able to analyze data on performance, behavior, level completion, failure and scores collected from hours of active gameplay with over 2650 levels completed by real players. The whole process highlighted interesting facts about the game and allowed me to better balance the levels and improve the design of the game in specific areas.


The chart above identifies the level 8 of Jelly Slice as being too difficult. The standard deviation values for completion time were also puzzling at first but most likely indicate a flawed design due to the high availability of “hints” allowing a number of players to rush through the levels.

Based on these observations the following could be done:

  1.  The level 8 could either be introduced later or be simplified.
  2.  The hints, currently available for every level, could be scarcer or be made available only after a few tries or the penalty for using them be more important.

The average score on all 2650 completed levels was 2.2/3 which indicates a well balanced difficulty overall. However, I found the figure below quite interesting.


It highlights the lack of incentive for the players to spend more time improving their score and can potentially give me important insights on the direction to take for monetizing the app: booster iAPs would only work if the social integration is strong while premium content may work well on this game even without the social component.

As we can see, through a minimal and costless setup, the versatility of HTML5 can be used efficiently to iterate the design and quality process of our mobile games allowing us to make important adjustments prior to an official release.

Thanks and feel free to get in touch.