Level Metrics

The December build of Bullet Time Ninja has been online for a few weeks now. I have been using analytics software to track how people have been playing the game. Today, I will share some of my metrics, and show how I am using this information to improve gameplay.

Above is a youtube video of the version of the game where these metrics were tracked.


Measuring Engagement
I use metrics in Bullet Time Ninja to see how many levels people play, and how many times people die in each level. In theory, if the game is extremely fun, most people will play the game all the way through. Otherwise, people will quit and play something else.

Thanks to free services like Playtomic, I can collect large amounts of player data, and organize it into pretty charts.

This chart shows how many people play each level:

Almost everyone plays the first level, but only about one-third of players make it all the way through to the end. Notice that the drop-off is pretty large in the early levels, but is a bit smaller in the later levels as players get more invested.

Discovering a Level Design Problem

There is a particularly large drop-off on Level 7. Why did that happen? After a small bit of investigation, I noticed that a lot of people ended up dying on that level. Let's add how many times players die to the chart.

Whoa! On Level 7, players died an average number of four times! This level must be too difficult for some players to even finish!

Although, looking at the chart, there are a few other places where players die a lot, but don't quit. This probably means that this one particular level gets too hard, too early in the game. If I make it a little easier, it should smooth out the game's difficulty curve, and allow more people to enjoy the Bullet Time Ninja.

Level Design Revisions

The metrics I collected told me a lot of useful information. Let's take a look at the level to figure out what makes it so difficult for new players.

The first part looks pretty straightforward.

This second part seems to be a little harder. When I looked at it again, I realized two problems:

1. The player needs to dash a long distance to reach safety, which is difficult.
2. This is the first scrolling level the player encounters, and the player cannot see the wall on the far right until they get there. This is bad level design.

I chose to extend the cliff out far enough that the player can see it before they jump the gap.

These design changes should make the level more fair and fun to play.

Collecting Metrics on this level helped me discover an important design error, and correct it. As Bullet Time Ninja continues to grow, I expect that these Metrics will become even more interesting and useful.

- Greg