Post-Mortem: Discover O - Design with Data

After the failure of our first 2 games at BYOF Studios, we decided once more to go after the global market. We wanted to build something simple, but beautiful, and something that would put us on the map globally. The goal was to keep our timelines short, our quality high, and get noticed. 

Discover O (iOS | Android) was a game conceptualised off many game designer's desperate fascination with the Google Chrome logo.

The concept was simple to grasp.

Swipe the ball to the same-coloured quadrant of the outer circle. The ball shrinks, so the faster you swipe, the more you score. The circle around the ball rotates, as the colors of the quadrants start converging. This is when it gets really tricky. If you don't swipe before the ball shrinks, you lose.


This seemed like a simple mechanic to play. But after building out a prototype and testing it with some friends and family members, the feedback we got made us feel like there was still something missing...

Now, in early 2015, the mobile games industry was riding high on the wave of a new design movement. The minimalist movement. The movement was centred around form and function, but the artistic output of it was a flat style in combination with an analogous colour scheme. 

Google's material design principles were beginning to make waves as well, and a general evolution in game design towards cleaner, simpler games had already begun.

The wave was just beginning to mature around the time we were building the game, and it seemed like the right time to ride it.

The feature we added to the game design was a simple levelling system. You gradually uncover parts of the stunning animated backgrounds while you score more points within each level. When you're done exploring a single level, you can unlock the next one using a few diamonds.

Diamonds are earned every couple minutes of play. Diamonds are used to unlock new levels (i.e. new backgrounds to explore), or to continue without restarting if the player loses.

We launched the game with 5 levels - around 2 weeks worth of content.

As part of the launch strategy, we entered a contest, and earned ourselves a platform to announce our launch on stage by qualifying as part of the Top 5 games at Pocket Gamer Connect's Indie Game Hack event. 

We launched on iOS only because the game fit better with the content Apple was featuring at the time, and they had made it a habit of supporting indies with good products. The game was free, with ads and in-app purchases.

As I've already covered before, the game was featured by Apple on iPad in 20 countries around the world for its launch week, and that gave us a quick surge of 10,000 users. 

But what happened next? Did the users play the game? Did they buy Diamonds? Did they hate it? Love it? Despise it? Did the game go viral?

Let's find out.

At BYOF Studios, we employ a data centric approach to building and servicing our products. To understand the successes and failures of Discover O, we first need to look at what the contributing factors to a successful game are.

Mobile gaming is commonly broken up into 4 simple (yea, right!) facets, namely:

Our Engagement was covered by an addictive core loop and progressive game logic that users could keep coming back to in their free time. This seemed like something that would keep the user replaying for short spurts of a few minutes each.

Retention was covered by unlockable content with the backgrounds and the levels, that would keep the user coming back to explore more.

Monetisation was taken care of by in-app purchases of diamonds in combination with the levels, high scores and the leaderboard. Along with an ad-based strategy that showed the user an interstitial ad every 3 losses heavily tied in with the engagement we were expecting.

The art style, game design, all the PR from entering the Pocket Gamer Connects Indie Game Hack, obsessive networking with game journalists, getting noticed by and then featured by Apple, ability to share high scores and images of levels, rewarded invites, App Store Optimisation and many more were all part of the Distribution strategy.


So let's break those down with some data now:

Our session lengths here give us a good indication of what our Engagement numbers were like:
 

For a game that was heavily reliant on addictive gameplay, constant replayability and an ad-based strategy that relied on the player's willingness to keep playing, a 100 second average session length is definitely not the best place to start. You can see that a significant majority of our users were not playing beyond the 1 minute mark, and that was a big, big problem. The average session lengths we were expecting for a super-casual game like Discover O were closer to the 180 second mark.

Our analysis of this was that the game design itself didn't meet the standards set by other competing games out there, i.e. it wasn't addictive enough. More importantly, we hadn't optimised it for iPad either - so things were a little blurry, out of proportion and hard to play.

Now let's take a look at what that did to our Retention numbers:

Industry standards and our own targets at the time indicated that 40% Day 1 Retention was an absolute necessity for the game design to be deemed successful - ours is pictured above at 20.24%. Day 7 Retention needed to be at 20% at the min - ours is 4.05%. We were churning out our users way quicker than we should have been. 

Our two-pronged approach with Monetisation was suffering as a result of our low engagement and retention numbers. 

IAP Monetisation suffered for the following reasons: 

Ad-Revenue suffered because our engagement and retention were way too low. Had we hit our goals on session lengths, session frequency and D1-D7 retention, we'd have been doing pretty well on this front. Our eCPM's of $2.51 on iOS and $1.39 on Android from AdMob could possibly have been optimised had we used a mediation partner like HeyZap.

And finally, the elephant in the room with every indie mobile game developer - Distribution. After the week we were featured on the App Store, our download numbers plummeted.




The clear drop after the week of the feature, was something we didn't anticipate. We believed everything we'd baked into the game to help with the distribution problem would've made the growth curve an actual 'growth curve'. But unfortunately, it didn't. Since we weren't making any significant revenue from the game to pump back into marketing, and we didn't have a marketing budget to begin with (as most indie studios don't), our download numbers became negligible over the course of a few months. 

And so we reached a point where the engine we created to engage, retain, monetise and grow our users was devoid of exactly that - users.

The key takeaways we're hoping that anyone reading this will get when publishing their first products on the stores are these: 

This isn't a bible or a rule-set. It's meant to be more of a framework on how to think about launching your products if you're lost, whether they're games or apps. It's just the way we think, and we're hoping it can help somebody apply it to their own work.

The science of building for and around the 4 facets or 'problems' (to solve) of publishing are infinitely more complex than what's covered here, of course. We'll go a lot deeper with the next post. 

You can check out Discover O on iOS and Android to have a look at the game and experience all the changes we made to solve the 4 problems of engagement, retention, monetisation and distribution. I'm sure you'll spot a thing or two in there (Discover Thrones anyone? ;) iOS | Android ).


Next post:
Our next deconstruction will be that of the soft launch of our game Bluff Party. We'll be going through each of the 4 problems, the various solutions we employed, the changes we made after careful data analysis, and how those changes affected the data within each of the 4 problems.

Links to previous posts:
Part 2: Strike 3 or third time lucky?
Part 1: Two Strikes - How we ploughed on in the face of failure