[GameDev] Google Analytics-Driven Game Development by Example

Otro artículo interesante de GameDev.net:

Hi there. A few months ago my friend and I performed something like a gamedev experiment. It was 'just create a trashy game, update it as frequently as possible and make it better based on users’ and Google Analytics’ feedback'. We didn't want to spend 6 months on developing another game that will get stuck among thousands ofother games on Google Play. Moreover, we didn't have as much as 6 months. It was more like a quick execution of a half-witted idea. We wanted our game to evolve based on the user feedback and our conclusions.

The Beginning - version 1.0


So, what was the idea for the game? Every time I had a browse in Google Play I wondered how a game about tapping an egg can become so popular. ‘Over 5 million downloads of a simple tapping app. Holy f*ck, what is going on here?’ - I asked myself.


RyydrRG.png
s9YNuy5.png


The game average rating was 3.0 and people were still downloading it. Apparently, the rating does not matter... You can clearly see that for Google Play the number of downloads is more important than the rating. This game is the first search result for the ‘egg’ keyword.

So the idea was: Let’s make a simple egg tapping game as fast as possible. Ingredients: free music, 50% of free graphics, 50% graphics made by my friend, 5 hours of programming, 2 hours of making screens and putting the game on Google Play and here we go: Pet Baby Egg.


nATnIWg.gif


Every time you tap the egg, you get closer to obtaining another pet. The next levels obviously become more and more difficult.


2rrKb7b.gif


The only challenge for the player was to collect all pets, nothing more. Boooooring.Now let’s take a look at some stats.


DtgIAQW.png


Damn, somebody has actually downloaded it! :D


giWBbDX.png


611 sessions in 2 weeks. No marketing. Just keywords and the game. Only 46.2% of returning visitors means that after a month or two, nobody will play this game anymore.


xAtq1zP.png


The average session lasted only 3:03. First conclusions: Let's try to interest the player. Let's give more content to get a longer session duration. So when a player gets a new pet, why not let him take care of it?

First pet care - version 1.1


We added basic tamagotchi features: a kitchen to feed the pet, a fun room to play ball with it and a bedroom for sleeping. Moreover, a player could earn money by playing with his pet and by tapping the egg. There was also a shop to spend the earned money on food. It took us maybe a week to implement some new graphics and some new code in our spare time. This is what we got:


nutUfRw.gif


And after 7 days of version 1.1 being available in the Google Play Store we had the following stats:


fssPVvO.png


The number of new installations increased by around 20%!


m5bBu6J.png


Returning visitiors ratio grew to almost 52%!


4VoP39j.png


And... the average session duration almost doubled! That made sense!

So it was time to draw more conclusions: There were a bunch of game events. We knew exactly what users did in the game.


NOS4eYU.png


We analyzed the top events from the game and we came up with these results: the most popular event is ‘touching the ball’ (thanks to the physics engine that made it fun for the player). The users got 1 coin for 1 tap on the ball or the egg. But we were surprised by how few users bought balls in the shop. Our conclusion was that they didn’t know how to earn money. They play ball with their pets but they don’t know that it earns them money.

Earning money more visible - version 1.15


Before


2rrKb7b.gif


After


cOHoz4a.gif


Before


Lpg7BKV.gif


After


fe8INYe.gif


The effect: In the 1.15 version there was over twice as much (6.77% to 14.99%) events related to buying balls!


ZGBEyWg.png


Ok, let’s add more stuff.

Advanced care - version 1.2


It took more than a week to implement it but we added special indicators to show the levels of hunger, happiness and energy. Our goal was to engage the users more to take care of their pet or to return to the game more often. Playing ball with the pet also became more interactive as it could now play with us!


jYGTFfr.gif


The Effect: Over 57% of returning players. 5.5% more than in the previous 1.1 version!


Tbh4KGV.png


The average session duration grew to 7:22 and that is a 142% boost in comparison with the version 1.0 (3:03) and a 24% growth in relation to the version 1.1 (5:56)!


0cFkmbC.png


More content and a mini game to earn coins! - version 1.3


OK. We had a simple pet care simulator, but we wanted to add more items to be available for purchase. They became expensive (there was inflation in our small game because we didn’t pay enough attention to economy in the game). For example, we made wallpaper available to be put in one of the rooms for 5000 coins. We thought we had to introduce a new way to earn more money so we implemented a mini game that made it possible.


1o4qcFI.gif


The mini game made it possible to earn up to 500-1000 coins in a 1-2 minute gameplay. What was the effect?


1Bjcx3i.png


The retention growth up to 70.5%! It is 13% more than in version 1.2! It was definitely worth implementing.

More stuff and features - version 1.5


In this version we fixed some bugs and added more items and features to the game. It was the last update we made to this game.


iTKhApi.gif

YSUxnn9.gif


The effect:


pOnp5F8.png


The average session duration increased to 8:20 which is a 273% growth compared with the starting point from version 1.0!

Summary


  1. By adding more content to the game we got more retention (from 46.2% to 70.5%).
  2. By adding more content to the game we got longer average session duration (from 3:03 to 8:20).
  3. We do not only gather the Google Analytics events but we also analyze them!

We don’t encourage you to make a poor game and update it, because in the long run it is not cost-effective. But we strongly encourage you to use Google Analytics and analyze the data you get!

Article Update Log


23 July 2015: Initial release

Creative - Articles http://ift.tt/1MUlvJj

Comentarios