Today, browsing my old notes on mobile game performance optimization, I’ve decided to structure them in a single post. I’ve started write my notes from that time, when together with my fellows we launched our first mobile game on production. That was really outstanding experience. At that time most of things were new and unknown for me. But despite this fact, rich work experience at leading global software development and consulting company allowed pretty quickly setup environment and establish required processes. And few months later we achieved first result. It was time when we published our first game on Google Play and Apple Store. Finally, first downloads and installs. We started receiving first comments from real consumers. It was first market feedback on our new game. Later we realized that having all these data was not enough to understand whether our product successful or not. At the end we faced with task to transform “successful product” term into something measurable, something that we can calculate and optimize, something that we can have direct influence on.
Before I will tell you how we measure success of our products I would like to give you simple example from everyday life. Let’s imagine that you’ve opened cafe in your city. Your cafe has nice interior and delicious cuisine. It is located very close to downtown and it’s visited by a huge number of locals and tourists every day. Can we assume that your cafe is successful, based on this information? From the first point of view, yes we can, because of big flow of customers that are coming. But let’s take a look deeper and we will see that many of customers are tourists, who visiting your cafe because of advertising posters in their travel guides. And the rest of customers are locals, who walking along the street and seeing new sign decided to look inside and try something new. After a while you decided to stop all marketing campaigns and remove advertising posters from travel guides. As a result, flow of new visitors significantly reduced. Tourists no longer visit your cafe and locals also skip it another way. Such seemingly successful cafe is now half empty. What actually was the reason for such significant fall of visitors and whether you could respond to this issue in advance. Actually, if we look deeper, your cafe was not successful from the very beginning. All the tourists, who came to your cafe, was result of advertising posters in their travel guides. And locals, after visiting cafe for the first time, almost never come back later, because of unfriendly staff and quite inflated prices. I’m sure having this insights from the very beginning you would execute all possible preventive actions to avoid such terrible situation.
Retention and LTV as Core Metrics
The same behavior we can observe with mobile games. Seemingly large number of downloads and comments does not mean “successful product”. We discovered this with our first mobile game and right away decided to introduce quantitative metrics in order to measure product performance, which eventually impacts product success. As a core metrics we took Retention and LTV.
Let me briefly visualize and explain these two metrics.
Retention – reflects ratio of returning users on specific day that belongs to some group within a defined timespan (let’s say Cohort A), to users from the same Cohort A, who installed and launched game for the first time. By analogy with cafe, those visitors who came to cafe for the first time and continue visiting it next days.
Area under this chart represents all users from Cohort A within specified timespan, who installed and launched game for the first time on Day0 (in our example it is 100 users) and continue returning to game during next seven days Day1 – Day7. The resulted retention for Cohort A by days can be calculated by following formula:
Day(X) Retention = Day(X) Users / Day0 Users
And in our example it will be the following:
D1 = 60% | D2 = 40% | D3 = 40% | D4 = 35% | D5 = 40% | D6 = 40% | D7 = 30%.
To understand whether resulted retention good or bad I propose to use this baseline that defines of 7-day (D7) retention rates in mobile casual games:
30% – amazing | 20% – good | 16% – average | 12% – poor | 8% – oh god, something’s wrong!
LTV (customer lifetime value) – revenue accumulated for single user that staying and interacting with a game during predefined period of time. By analogy with cafe, accumulated income from single visitor who came to cafe for the first time and continue visiting it next days.
Each dot on this chart is single interaction with user that resulted into some revenue. Set of all dots represent accumulated revenue of Cohort A users within specified timespan, which equals $170. So resulted LTV can be calculate by the following formula:
LTV = Accumulated Revenue / D0 users
And in our example it will be the following:
$170 / 100 = $1.7
Game Financial Success
After series of simple mathematical calculations we know average revenue per one user. But know the question arises – how much we spend to bring one new user into the game. In other words, what our CAC (customer acquisition cost). For example, let’s say we pay $1 to bring one new user into the game. And now, knowing our average LTV which equals $1.7, we can say that our game is quite successful, as we can invest into new users engagement and our return of investments will be positive:
ROI = LTV – CAC = $1.7 – $1 = $0.7
To keep things going well we just need to follow one rule: as long as the average customer lifetime value (LTV) exceeds the cost of customer acquisition (CAC), we will have a positive return on investment on all our marketing expenses:
LTV > CAC => SUCCESS
In case we can’t met this condition and our LTV is less than CAC, then we need to take vital preventive measures, like:
- Increase Retention. Increase area that represent users from Cohort A on Retention char
- Optimize Monetization. Increase density and/or size of dots on Accumulated Revenue chart
- Apply Virality. Decrease cost of acquiring new user to the game by distributing total cost among multiple users using viral approach. So having $1 engagement which result in 1 new user and further this user acquire 3 more users for free (e.g. using word of mouth) we can say that final CAC is distributed among 4 users and equals $1 / 4 = $0.25.
After introducing core metrics into our first mobile game (iOS/Android) we understood how it perform. Primary monetization channel was built purely using AdMob advertising network. And there were no specific retention and viral techniques implemented that keep user playing and return to game. So, resulted Retention and LTV (as of November 2014, when we introduced product performance metrics) were following:
Day7 Retention = 11% and LTV = $0.05
Unfortunately, not so good as we originally thought based on downloads and installs. But at the end it was valuable experience that we’ve gained and now continue spreading among our products. Ultimate success of any single product boils down to knowing and understanding of its quantitative and qualitative attributes. That is why introducing Retention and LTV as key performance metrics for your product, will allow continuously check it’s health status and make right and reasonable decisions.
Thanks for taking a peek at my notes. I’m excited to learn more about different techniques and approaches in the area of product design and development. Later I’ll check back in and share with you what I’ve learned. In the meantime, feel free to share your own experience on how you measure mobile game performance and what is your average Retention and LTV. I’d like to hear your thought in comments below.
You may also like my other posts:
- How to calculate Mobile Game Retention and LTV using Google Analytics
- 5 Criteria of Successful Mobile Game
Originally published at measureofluck.com on August 25, 2015.