2 of 27

INTRODUCTION

Success of games begins with understanding your customer.

Review systems great way to get feedback.

Sheer volume of reviews makes evaluating in depth difficult.

Problems customers have must be addressed.

3 of 27

INTRODUCTION

Patterns exist that can potentially give important insight.

Each customer is unique.

Not all games genres have the same patterns.

Many variables exist within reviews.

Correlations between the variables could be monumental.

4 of 27

DATA COLLECTION

All data came from game reviews left by players

Data files were converted to CSV

Free to play (F2P) and buy to play (B2P) game models in the MMO market.

Main research question intends to answer difference between these models, and their impact on the frequency of highly enthusiastic, as well as negative reviews.

5 of 27

DATA COLLECTION

Using total word count to measure positive and negative net response.

Hypothesis being that higher word count means the reviewer is more sincere.

Each MMO has its own sub-genre

Each sub-genre has its own unique player base

F2P and B2P MMO sub-genres are important in measuring customer response

6 of 27

DATA COLLECTION

Anticipating F2P reviews to be shorter, or lower enthusiasm

Anticipating B2P reviews to contain more enthusiasm

Understanding the relationship between these variables can help developers market more effectively.

See the 12 variables we tracked.

7 of 27

DATA COLLECTION - VARIABLES

Recent_Played_Time Numeric�Review Character �Achievement_Percentage Numeric�Total_Played_Time Numeric�Rating Factor Factor�Friends Integer Numeric�Gamed_Owned Integer�Monetization Factor�Game_Name Character�Genre Integer Character�Adjusted_Time_Played Integer�Review_Char_Count Integer

Tracks last 2/wks played time�Raw review words�% of game completion�Tracks total playtime by hours�Pos/Neg value�Tracks total # of friends�Tracks total # of owned games�Paid/Free value�Name of game�Name of genre game belongs�Weighted total playtime by hours�Count of review characters

8 of 27

DATA ANALYSIS - OVERVIEW

First step in understanding the relationships of the variables.

What affects the time people spend playing a game?

Best indicator, the variable Adjusted_Time_Played.

Regressing variables on Adjusted_Time_Played will show notable values.

9 of 27

DATA ANALYSIS - TIME PLAYED

Notable Values:

Rating
Recent_Played_Time
Friends
Monetization
Achievement_Percentage

�

10 of 27

DATA ANALYSIS - TIME PLAYED - EFFECTS OF RATING

People play games less if they don’t like them

Fair assumption

Data backs it up strongly.

11 of 27

DATA ANALYSIS - TIME PLAYED - EFFECTS OF MONETIZATION

People tend to play free games more.

Free games are generally designed to keep people playing.

There are ‘holdouts’ who will continue playing free games even if they leave a negative review.

12 of 27

DATA ANALYSIS - TIME PLAYED - EFFECTS OF GAME GENRE

The MOBA genre has been well known to attract players for long periods of time.

This is using the adjusted time played value.

13 of 27

DATA ANALYSIS - TIME PLAYED - COMMITTED PLAYERS

Players who had less than 10 hours played removed.

Greatly affects the total average of some games.

14 of 27

DATA ANALYSIS - TIME PLAYED - FRIENDS

Average number of hours categorized by groupings of number of friends.

Fair linear relationship between the two variables.

15 of 27

DATA ANALYSIS - TIME PLAYED - FRIENDS

Average number of friends categorized by time played.

Linear relationship remains, with a slight dip at (700,800).

16 of 27

DATA ANALYSIS - TIME PLAYED - FRIENDS

Outliers severely affect the data.

Several players with many friends at 0 hours played.

Best-fit line still shows minor correlation even with outliers included.

17 of 27

DATA ANALYSIS - RATING - FRIENDS

Number of friends has an effect on rating as well.

Players that have more friends may ‘enjoy’ the game more.

18 of 27

DATA ANALYSIS - RATING - GAMES OWNED

The number of games someone owns seems to slightly affect the rating.

19 of 27

DATA ANALYSIS - RATING - GAMES OWNED

## ## Welch Two Sample t-test 21 ## ## �data: test_neg$Games_Owned and test_pos$Games_Owned

## t = 1.8134, df = 41.784, p-value = 0.03848 �## alternative hypothesis: true difference in means is greater than 0 �

Passes hypothesis test using normal, random samples from the data.

Hard to explain why this is affecting the data, but it is.

20 of 27

DATA ANALYSIS - RATING - MONETIZATION

Paid games have about 2% more negative reviews than free games.

That is effectively 25% less negative reviews for free games.

21 of 27

DATA ANALYSIS - RATING - MONETIZATION

Players are harsher on MMORPG games over MOBA games by a difference of 5%.

Small 2% difference between MOBA and Shooter genres.

22 of 27

DATA ANALYSIS - RATING - COMPLETION

Players who leave negative reviews still complete a similar chunk of the game even though they play less.

23 of 27

DATA ANALYSIS - NEGATIVE RATINGS

How many reviews come from players who have never loaded the game?

17/38,000 - An effective 0%.

However, 5% of all reviews come from players who have played less than 4 hours. (The mean hours played is 511 hours)

How many of these reviews are negative?

20% of ALL negative reviews are from users who played the game 4 hours or less.

Found something they didn’t like

Simply didn’t enjoy the game

24 of 27

SUMMARY

Do people with more friends play multiplayer games longer than those with fewer friends?

Yes.

Correlation between number of friends and time spent in game.

Is there anything that has a noticeable effect on whether a review will be positive or negative?

Inconclusive.

Fewer friends, and more games owned, correlates to more likely negative reviews.

25 of 27

SUMMARY

How do free games compare to paid games in terms of their positive/negative reviews and time played?

Paid games slightly, by 2%, more harshly criticized.
Free games had higher total play time. (Dota 2 skew)

How do game genres compare in terms of positive/negative reviews and time played?

MOBA genre highest total playtime, and number of positive ratings.

26 of 27

SUMMARY

Do people that leave negative reviews play those games as much as those that leave positive reviews? Do they complete as much of the game?

No, and yes.

Negative reviews have less total playtime, but do complete as much of the game.

27 of 27

CONCLUSION

Still no clear way to determine if customer will leave negative or positive review.
Minor trends discovered can help with prediction.
Key trend #1 : Players leaving negative reviews do play the game less, while still completing the same amount.
Key trend #2 : 20% of negative reviewers barely played the game.
If number of negative reviewers grow, focus on improving the beginning of their game.

Overall, companies should focus on solutions, not prediction.