Tactile

Installs over time

As measured by newPlayer events, there was a total of 102,392 installs over the 31 days.

Installs by Country

The following chart shows the top 20 countries in terms of installations.

Engagement - Daily Active Users

Number of users that recorded any event on a given day.

Engagement - DAU/MAU

The ratio of Daily Active Users to Monthly Active Users (MAU) shows how well an app retains users. DAU/MAU for February 2006 was:

Average DAU: 18,879

February MAU: 96,339

DAU/MAU ratio: 19.6%

Most successful games have DAU/MAU ratios of around 20%.

Engagement - Daily Rounds

Number of times the average user is opening the app on a daily basis.

Retention - Day N Retention

Number of users that installed the game on Day 0 and then recorded an event on Day X.

Retention - Day N Retention - All Devices

Number of users that installed the game on Day 0 and then recorded an event on Day X.

Retention - Day N Retention - WebGL

Number of WebGL users that installed the game on Day 0 and then recorded an event on Day X.

Retention - Day N Retention - All Except WebGL

Number of non-WebGL users that installed the game on Day 0 and then recorded an event on Day X.

Retention - Played for more than a day

What percentage of users played the game for more than one day.

In-App Purchases over time

Total in-app purchases over time.

Revenue Per Daily Active User/Paying User

Revenue per Daily Active User (ARPDAU), and per Paying User (ARPPU).

Spending Types

For users who have made a purchase, how many purchases are they making?

Spending Types

How much revenue is being created by each category of users?

Top Countries

Countries generating the most revenue.

Top and Bottom Countries - Normalized

Highest and lowest countries for revenue per install.

Versions

Has the difficulty changed from version to version?

Versions - Timing

When were users using each version?

Versions - Level Attempts

Which levels were users attempting in each version?

Level Difficulties

Which levels are the most difficult (most failed attempts)?

Player Drop-Off

Which levels cause the most players to give-up (next level never attempted)?

Player Drop-Off by Level

What does user drop-off look like over the first 100 levels?

Encouraging Conversions

Which levels are leading to users' first purchases?

Coin economy over time

What is the net balance of coins (coins received - coins used) each day?

Campaign Summary

Testing Campaign Significance

From the summary data, we might conclude that users coming from campaigns are more likely to make purchases, however, there could be confounding factors. For example, campaigns could be:

• Targetting richer countries, and those users are more likely to make purchases
• Targetting hardcore gamers who progress further and are therefore more likely to make purchases
• Targetting users with specific devices that are more likely to make purchases
• Providing users with coin bonuses, which then encourages users to make purchases

We can build a model to test this.

Testing Campaign Significance - The Setup

The model built attempts to account for all these factors, and has the following inputs:

• Country - Mean revenue per user from each country
• Device - 4 dummy variables
• User skill - Maximum level reached, percentage of failed missions, percentage of abandoned missions, and total mission events
• Coins - Total coins received and coins used
• Campaign - A simple flag if the user came from a campaign (1) or not (0)

We use these data points to create a Logistic Regression Model to that predicts whether a user will make a purchase (1) or not (0).

Testing Campaign Significance - Results

What does this mean?

What conclusions can we draw from this model?

1. Campaign users are more likely to make purchases - Even when we take into account differences in country distribution, device type, level progression and coins, however there could be...
2. Confounding factors - These results do not rule out the existence of other confounding factors that explain the remaining difference, which is why...
3. More research is needed - Identifying all the factors that make campaign users more likely to make purchases can help to target future campaigns.

Model Performance

Model Performance

How accurate was the model?

Campaign Timing

When did the campaigns have the biggest impact?

Most Profitable Campaigns

Campaigns that produced the most revenue.

Case Study 1

Campaign ID: 9s5vVew1HvRvFU+l8FHarl/rh1WBtudlUUe+K5CuudM=

New users: 1,784

Revenue created: 1,065

Device used: Primarily IOS (1,772 of 1,784)

Country of origin: Users come from most developed countries

Appears to be a successful campaign. Attracted a large number of users, with 218 (or 12.2%) of those those users going on to make in-app purchases.

Case Study 2

Campaign ID: 4EPzfhaYBdO1groLb/JBb6j0OAV2bqxyn8fUBBeZ6HM=

New users: 215

Revenue created: 577

Device used: IOS

Country of origin: United Kingdom (212 or 215)

Appears to be a very positive campaign, but in fact, only 5 of the 215 users were paying users. It just happens that one of those users generated 573 of the 577 revenue.