Game Economy Project.pdf

Mobile Game Analysis - Liad Traube | LinkedIn
1
Mobile Game Analysis
By Liad Traube
TL; DR
This project goal is to simulate a mobile game database and make an
analysis based on data from that database, including KPI's and game
economy.
1. Introduction
For many days I wandered around web-sites like "Kaggle" and "Maven
Analytics" searching for interesting datasets that I could use to create a
big and impressive project, downloading files, loading the datasets in
MSSQL and Tableau, preparing queries in SQL but It was all for vain.

2
none of the datasets were good enough for my needs from my point of
view and it seemed to me a lot the datasets in Kaggle were tailored for
data-science and machine learning projects and some of the datasets
were simply not "rich" enough in data or did not have what I needed for
the kind of project that I had in mind.
As a result of my frustration in finding a proper dataset I decided that I
have to address this challenge and build a database myself.
As a consequence of having the creative freedom with the mission of
building a database by myself I decided to choose a subject that I find
interesting and I want to study more from the "developers" side.

3
1.1 Mobile Game Introduction
As you will see in this project, I tried to simulate a real mobile game
database as best as I could, using randomly generated data thanks to
websites like "generate data" and various functions in Microsoft Excel.
The database contains data about a mobile game called "Empire Of
Crabs" (I tasked chat-GPT to generate me 10 names for mobile games
and this was my favorite name).
*This image was randomly found in google and was added to illustrate the idea, this image doesn’t belong to me
Game description:
Embark on a crustacean adventure like no other in "Empire of Crabs"! Dive into a
captivating underwater world where you take command of a vibrant and resourceful
crab colony. Your mission? Build, strategize, and conquer to create the ultimate crab
empire!
As the wise and ambitious leader of your crab civilization, you'll navigate the
challenges of the ocean depths, from treacherous currents to rival crab factions.
Harness the unique abilities of your crab subjects, each with their own strengths, to
gather resources, construct intricate habitats, and defend your territory against
predators.
Explore a visually stunning and immersive aquatic realm, where vibrant coral reefs,
mysterious shipwrecks, and ancient ruins await your discovery. Collect rare
treasures, unlock hidden secrets, and uncover the history of your crab civilization as
you expand your dominion.
But it's not all peaceful tides and clear waters. Engage in epic battles against rival
crab colonies for dominance over coveted underwater territories. Strategize your
tactics, deploy your crab forces, and outmaneuver opponents in real-time battles
that will test your leadership skills and wit.

4
1.2 Monetization Strategy
Our product, the mobile app is free-to-play with in-app
microtransactions. These transactions allow players to access premium
content, customizations or enhancements within the game.
in our mobile-app the users have the ability to purchase:
cosmetic Items like skins
power-ups and boosts
time-saving mechanisms
premium content (DLC)

5
2. Data Base Structure
2.1 Database Tables
The data is stored in 9 different tables
1. GameInstall
Holds data about all the users that installed the game at least once

6
2. Users
Holds data about all the users that also opened an account (the other
option is to sign in as "Guest")
3. Log_In
Holds data about when different users logged in and used the app and
for how long (in minutes)

7
4. GameRatings
Holds data about what rating each user gave to the app (if the user
chose to do so)
5. Marketing
Holds data about all the money that the company use to advertise the
game and platform fees.

8
6. PremiumItems
Holds data about the special items or features that the user can
purchase with real life money.
7. Purchases
Holds data about the purchases users made (using real life money).

9
7. game economy
Holds data about the digital in app currency inflows and outflows.
the digital currency we can call "CrabCoins", so the "CrabCoins" that
were generated by various actions made and milestone achieved by the
users actions in game and the "CrabCoins" that were used by the users
and disappeared into the void (or sink, whatever term you like).

10
2.2 Data Cleaning and Validation
As consequence of the data being randomly generated, I had to build
queries in SQL to make sure that data make sense,
for example:
a user cannot have data in the log-in table on a date that is earlier than
the date he installed the app for the first time.
Installs table -
Log_in table -
Query to validate the data
*There were 8 other queries to validate the data but I did not include them in this file

11
3. Analysis
The analysis will be based on data from the year 2020 only, a yearly
analysis you could say.
3.1 Descriptive Statics
I will start the analysis section with some simple analysis to give a little
snapshot about the mobile app performance in 2020.
just to clarify I will add and say that the app was made by 1 person and
the only expenses were spent on marketing.

12
3.2 User Engagement
DAU
daily active users graph shows the number of distinct users that logged
in for each day in 2020.
MAU
monthly active users graph shows the number of distinct users that
logged in at least once for each month.
From the DAU graph we can notice that there is high scattering in term
of daily active users across different days.
Unlike the DAU graph in the MAU graph the number of total distinct

13
users each month is steadier and mostly balanced, with around 3,100 –
3300 distinct users per month.
Session Duration
The amount of time a user spends actively using the app during a single
session.
From these graphs we can note that are no significant changes in the
average amount of time users spent in-app in different months, although
there is one noticeable decline in total minutes between March and
April, while the average minutes in-app did not change at all between
these months we can assume from that a decrease in the total number
of distinct users between March and April (which is also noticeable in
the MAU graph).

14
Frequency
How often users return to the app over a given period of time (in my
project the period is one year).
From the first graph we can observe a slight trend – higher number of
log in's (right side of the graph) result in more total minutes spent in-app
which make sense but perhaps the difference is not significant enough.
Contrary to the first graph, looking at the second graph we can observe a
clear pattern – users with higher number of log-in's has lower average
amount of minutes in-app.
I wanted to check further if there is correlation between the number of
times a user logged in to the app and his average minutes in-app so I ran
code in python to calculate correlation and got the following results:
The results suggests a very weak positive linear correlation but since the
correlation coefficient (0.014) is extremely close to zero, there is almost

15
no linear relationship between the number of times user logged-in and
his avreage minuts in-app.
Retention
Retention rate refers to the percentage of customers or users who
continue to engage with a product, service, or platform over a specific
period of time. It is commonly used as a key performance indicator (KPI)
to measure the effectiveness of an organization's efforts in retaining its
existing customer base.
Retention rate can be calcuated based on user segments in terms of user
activity patterns, for example:
a segment for daily log-in user, a segment for weekly log-in user and so
on..
the higher number of days a "daily log-in" user does not log in to the app
the higher chance he will become " churned customer " (Churn is the
opposite of retention). For a weekly log in user it’s the same idea but
with weeks.
I decided to calculate without dividing the users for segments.
I calcuated the retnetion percentage on weekly basis for all users - how
many users logged in and logged in again the following week.

16
Observing the graph we can notice high scattering in the number of
distinct users every week, but a balanced percentage in the weekly
retention rate which may indicate a percentage of loyal customers
between 5% and 7%.
Game Installs
The following graph shows the number of distinct users that installed
our app every month.
The number of installs over the year is inconsistent, the goal should be
positive trend over time.

17
Rating
Mobile apps and online platforms often allow users to rate and review
apps, products, or services. These user ratings help others decide
whether to use the app or service.
The following graph shows the how users rated the app.
The graph is mostly balanced with one expectation which is the number of users that
rated the app "1", clearly some users have a serious problem with the app and that
should be looked into.
The following graph shows the rating each user gave compared to the total duration
he spent in-app. bigger circles indicate more based opinions.

18
The rational of these graphs is that more minutes in-app equals more
based opinions.
I decided to analyze if there is correlation between user total minutes in-
game and the rating that user gave.
The correlation result was -0.02744
the correlation coefficient is very close to 0, indicating an extremely
weak correlation. This means that there is almost no linear relationship
between the two variables.
Having said that, if usage is high but ratings are consistently low, there
might be an issue affecting overall user satisfaction.

19
3.3 Game Economy
Effective mobile game economy design aims to create a cycle where
players are motivated to engage, generate the game currencies /
resources and enjoy spending it.
For better user engagement and monetization, the currency in game
should have meaning so we don’t want to have it inflated too much, but
at the same time we also should consider to not frustrate the users by
making the currencies / resources too scarce. The goal is to strike the
right balance.

20
"Crab Coins" is the name of the in-game currency.
The following graph shows the total amount of "Crab Coins" that was in
game: total Crab Coins generated – total Crab Coins spent running total
over the year.
There seems to be an increase in total amount of "Crab Coins" over time.
Not necessarily a bad thing or a worrying sign, it is affected by number of
total players and over the year there was an increase in the number of
new users that joined our game world, but for better monetization as
our mobile game is free-to-play with microtransactions we should
monitor the game currency to make sure it's not inflated too much.
There are few things I would like to analyze in the game economy:
1. Analyzing the relationship between the number of active users and
the total amount of "Crab Coins" that is in the game and not spent.
2. Analyzing the number of actions a user make in game and the amount
of "Crab Coins" he has left (how effective are our "sinks").

21
3. Are there enough sinks to control inflation? do users engage with
those sinks?
The logic is that if our "Crab Coins" currency is inflated users will have
less incentive to purchase microtransactions. If our currency is worthless
users will have less incentive to engage with currency generating
activates – less user engagement.
The next graph tries to answer the first questions, what's the
relationship between the number of active players and the total "Crab
Coins" in-game.
The graph shows the total amount of crab coins that active users
generated each month in-game and did not spend (total remaining game
currency) as well as the number of active users each month.
every circle is different month.

22
In order to analyze to connection between total currency in-game and
the number of active players I decided to run a liner-regression
calculation when the number of distinct players is the explanatory
variable (x) and total "Crab Coins" in-game is the response variable (y), in
words - how the number of distinct players effect the total-currency not
spent in-game.
*Each circle is different month.
the results of linear regression are the following:
Slope: 61351.623661431004
R-squared: 0.5158445387651833
Slope - A positive slope (like in our results) means that as the number of
distinct active users increases "TotalCrabCoins" tends to increase and the
number 61,351 is the rate of change of "TotalCrabCoins" for a one-unit
change in number distinct active users.

23
R-squared – R-squared provides a measure of how well the linear
regression model fits the data. A higher R-squared value indicates that
the model's predictions are closer to the actual data points
the results put into words say that 51% of the variance in
"TotalCrabCoins" can be predicted by the number of active users.
Moving forward, next I tried to analyze the relationship between
number of actions a user made and that user's total "crab coins"
balance.
*we will consider actions as currency generating actions or purchasing
game items using in-game currency and not clicking on different menus /
switching screens.
Observing the graph the relationship between the variables is not clear, I
decided to run regression calculations.

24
Slope: 3620.8862488051304
R-squared: 0.018975813281307322
Based on these results, the linear regression model does not appear to
provide a strong fit for the data, and the relationship between the
variables is quite weak.
Carefully interpreting the outcome of the results there is a positive but
very weak relationship between the total "CrabCoins" and the number
actions taken.
The goal should be based on the game strategy, if the game strategy is
based on progression so the more actions a user make the more total
currency he should have (positive relationship between those variables).

25
The next graph shows -
total "Crab Coins" user generated, from lower to higher;
compared to total "Crab Coins" user spent from right to left.
I would like to analyze the relationship between currency generated &
currency used, the goal is that there will be a positive correlation
between generating our game currency and spending it.
to put in words – if there is negative correlation then perhaps the users
don't value the game currency enough / hoard it / don’t have enough
incentive to keep engaging with resource generating actions.
The results indicate a negative correlation meaning the higher total
currency the user generated (x) the less total currency they spent (y).
*Correlation does not imply causation and there could be other hidden
variables that influence the relationship between those variables.
With that being said, negative correlation is opposite of our goal.
This may indicate that the more user generated game currency the
greater his "Crab Coins" balance will be, danger of inflation and not
effective resources sinks in game.

26
3.4 Expenses
UA Cost
User acquisition is a metric used in marketing and business to measure
the cost of acquiring new customers, users, or clients for a product,
service, or platform.
User Acquisition Cost = Total Marketing and Advertising Expenses /
Number of New Customers Acquired
Judging the graph there is a big jump in UA cost from February onward
that should be looked into.
*As I mentioned earlier in this project, the game was developed by a
single developer and all the expenses are marketing and platform fees
that were included in the marketing table.

27
3.5 Monetization
The most important metric.
Essentially, this is how game developers make money from the games
they created.
Our product, the mobile app is free-to-play with in-app
microtransactions. These transactions allow players to access premium
content, customizations or enhancements within the game.
in our mobile-app the users have the ability to purchase:
cosmetic Items like skins
power-ups and boosts
time-saving mechanisms
premium content (DLC)
Category analysis

28
First, I would like to analyze how different microtransactions categories perform.

29
The most profitable category is power-ups and the least profitable is the
housing category.
In order to improve monetization perheps its advisable to get back to the
"drawing board" for the weak perfoming categories.
The graph below shows how different microtransactions item
performed.

30
ARPDAU
Average Revenue Per Daily Active User, pretty much describe the
relationship between the number of daily active users and monetization.
ARPDAU = Total Revenue / Daily Active Users (DAU)
The rational is that our ARPDAU can hint if the daily active users in our
app are the right audience in term of monetization.
if the ARPDAU is low then maybe we need to accurate our marketing
campaigns.

31
IAP Conversion Rate
In-App Purchase Conversion Rate refers to the percentage of users who
made purchases in app out of all total users that had the opportunity to
do so.
Improving IAP conversion rates can lead to increased revenue without
necessarily needing to acquire a larger user base.

32
4. Conclusions & Suggestions Based on The Analysis
Opening Remarks
First thing first, I would like to note that metrics change over time, the
important thing is to monitor the KPI's and understand what they
mean.
KPI's can give you a snapshot of the product perfomence.
They can help you to check if you are on track to achieve your pre-
defined goals, can help you make informed choices about resource
allocation and strategy adjustments.
KPI's also help identify areas where processes can be improved. When
you track performance over time, you can pinpoint inefficiencies or
areas with high variability and work to optimize them.
Furthermore KPI's can act as an early warning indicators. If certain KPI's
start to decline, it could signal underlying issues.
User Engegment
DAU (Daily Active Users)
There is high scattering in the number of daily users, meaning users
don’t have enough incentive to log in day after day. A suggestion to
Improve this KPI I would like to make is to set-up a daily log-in reward or
daily quest (perheps even both) giving the users an encourgment to log-
in on different days.

33
Rating & Number of Installs
user ratings help others decide whether to use the app.
There is a high number of users that rated the app "1".
after analyzing the total minutes users spent in-app compared to the
rating they gave there is indication that the ratings are mostly based –
meaning some user have genuine problem with our app that is
affecting overall user satisfaction and that should be examined.
As long the rating won't improve it will be harder to convince potential
users to download the app.
Frequency & Retention
From the analysis results its quite visible that users with high number of
log-in's have fewer average minutes in-app, on the opposite side of the
scale there are some users with low number of log-in's and high average
minutes in-app.
this kind of user activity pattern means that some users log-in and stay
in the app for an extensive amount of time, which can be good but -
a game with monetization strategy of in-app microtransactions should
probably be designed as "GaaS" (Game As a Service).
The rationale is that users won't invest in a game that they don’t expect
to play it again in the future, meaning that if our game monetization
strategy is microtransactions thus our goal should be on encouraging
higher number of log-in's and retention, less about total time in-app.

34
Game Economy
Game Sinks
From the analysis there is negative correlation between the total
currency the user generated and the total currency that user spent.
The interpretation of that outcome is that the higher total currency the
user generated the less total currency they spent citing game currency
sinks don’t work properly as the user progress in game, the more user
progress the higher his currency balance will be – less incentive to
purchase game boosters nor less motivation to keep carrying out actions
that generate game currency.
My suggestion is to design game currency sinks tailored for all levels of
user progression, meaning that users should always have a need to
spend, whether it's paying for utilities (getting stronger, quality of life
upgrades), rare collectibles, crafting or gambling (pay x for a chance to
win y).

35
Monetization
Microtransactions Categories
Stems from the analysis a high variance in different categories
performance in terms of revenue.
Concerning the weaker performing categories, it might be good idea to
get back to "drawing board" after analyzing what is good and bad about
the purchasable items in those categories according to user feedback, as
well as doing A/B test and experimenting with changes in order to try to
improve those categories.
Another possible thing to do is to offer limited-time promotions,
discounts, or special incentives to encourage players to explore these
categories.
As for the best performing categories – more of the same.
Analyze the factors contributing to the success of the best performing
categories and consider expanding the offerings within the best
performing categories. Introduce new content, items, or features that
are aligned with what players are enjoying
Mixing - Offer bundle deals that include items from the best performing
categories along with items from other categories.

36
Closing Remarks
This data-driven approach can help us to identify strengths, weaknesses,
and opportunities for improvement.
The suggestions put forth based on the KPI's analysis are designed to
capitalize on the game's strengths and address its challenges.
Tools Used in This Project
SQL – to view through GitHub Click Here
Python – to view through GitHub Click Here
Tableau – to view through Tableau Public Click Here
Excel – to download files Click Here
My LinkedIn Profile - Liad Traube | LinkedIn

Game Economy Project.pdf

Related slideshows

More Related Content

What's hot

What's hot (20)

Similar to Game Economy Project.pdf

Similar to Game Economy Project.pdf (20)

Recently uploaded

Recently uploaded (6)

Game Economy Project.pdf