How to define reasonable KPIs for mobile games

How � to � Define � Just � Right � KPIs � for � Game � Operation � 

�  �  �  � 

TalkingData � Leo � Cui �  � 

�  � Ta � 

� “TalkingData � is � a � professional � data � service � platform � for � mobile � applications, � serving � 2,500+ � active � apps � presently，with � almost � 1,000 � apps � are � mobile � games” � 

• Third-party � mobile � app � statistical � analysis � tools � 

• Professional � mobile � app � data � analysis/consulting � service � 

• Specialized � product � and � methodology � for � mobile � games � 

• Personalized � recommendation � engine/data � mining � service � 

• Forecast � model � and � emulation � service � 

• User � attributes � tagging � and � preferences � mining � service � 

TalkingData � Analytics � 

TalkingData � Insight � 

TalkingData � Product � Line � 

TalkingData � Campaign � 

• Mobile � app � campaign � monitoring � and � assessment � platform � 

•  iOS � monitoring � tool � released � in � Jun, � 2012 � 

• Currently � tracking � about � 1 � million � valid � app � activations � per � day � in � App � Store � 

•  TalkingData � Analytics � is � the � fastest � growing � mobile � data � analysis � platform, � already � covering � over � 5 � millions � devices � 6 � months � after � the � official � release. � Monthly � growth � rate � holds � at � above � 100% � 

2012.5 �  2012.6 �  2012.7 �  2012.8 �  2012.9 � 

3M � 7M � 

14M � 

30M � 

TalkingData � analytics � official � release � published � in � May, � 2012, � the � right � timing � to � witness � the � high � growth � of � mobile � Internet � in � China. � 

1M � 

Game � developers � need � continuous � data � analysis � to � enhance � products �  � Most � developers � don't � have � professional � knowledge � to � analyze � data � systematically, � and � in � the � mean � time, � facing � the � pressure � of � tight � schedule, � high � costs � of � man � power � and � hardware � 

�  � 

App � store � tracking � 

� Paying � player � conversion �  � 

Game � balance � 

Game � improvement � 

Mobile � game � developer � 

Operation � based � on � data � through � out � the � whole � game � life � cycle � 

Player � levels/progresses � players � classified � by � their �  � activeness � 

How � many � registrations? � How � about � DAU、MAU � ? � ……. � 

Game � release � 

Marketing � campaign � 

Campaign � result � tracking � 

Player � conversion/retention � 

Whale � users � 

Props � purchase � stats � 

Virtual � Economy � 

� Joint � operation/release � 

� ... �  � Data � Platform � 

Mobile � game � developers � facing � "data � dilemma" � 

0%  

25%  

50%  

75%  

100%  

Apr   May   Jun   Jul   Aug  

Casual   RPG   Strategy  

Casual � games � are � gradually � falling � out � of � favor � evidenced � by � the � number � 

of � games, � while � strategy � and � RPG � games � are � becoming � hotter � since � they � 

are � more � suitable � for � revenue � generating � through � IAP � 

Game � vendors' � favorite � game � types � are � changing � 

Data � source: � “Mobile � App � Data � Analysis � Q3 � Report � 2012 � ", � joint � released � by � TalkingData � and � NetEase � 

0.0%  

25.0%  

50.0%  

75.0%  

100.0%  

New  players  

+1   +2   +3   +4   +5   +6   +7   +8   +9   +10  

Strategy   AcEon   RPG   Casual   Puzzle   Desktop  

Desktop � games � excelling � at � retention, � with � more � well � polished � 

products; � more � rough � ones � in � other � types � of � games, � especially � RPG, � 

dragging � down � the � whole � average � rate. � 

New � player � day � 10 � retention � of � different � types � of � games � 

Data � source: � “Mobile � App � Data � Analysis � Q3 � Report � 2012 � ", � joint � released � by � TalkingData � and � NetEase � 

0.00% � 

10.00% � 

20.00% � 

30.00% � 

40.00% � 

50.00% � 

60.00% � 

Strategy � 

Action � 

RPG � 

Casual � 

Puzzle � 

Desktop � 

Life � time � distribution � of � diff. � types � of � games � Hard � core � players � usually � have � shorter � life � time, � therefore � need � to � be � motivated � by � 

continuous � flow � of � new � contents. � Desktop � games � have � more � well � defined � playing � routes, � 

but � with � lots � of � variations, � and � have � higher � player � stickiness. � 

Data � source: � “Mobile � App � Data � Analysis � Q3 � Report � 2012 � ", � joint � released � by � TalkingData � and � NetEase � 

User � preferences � aggregation � distribution � research � – � hard � core � player � 

Data � source: � TalkingData � data � mining � research � team � 

0.00% � 

5.00% � 

10.00% � 

15.00% � 

20.00% � 

25.00% � 

30.00% � 

35.00% � 

40.00% � 

角色扮演 �  射击游戏 �  动作游戏 �  战略游戏 � 

战略 � 

休闲 � 

射击 � 

体育 � 

动作 � 

益智 � 

角色扮演 � 

冒险 � 

棋牌 � 

养成 � 

经营 � 

模拟器 � 

网游 � 

KPI！=Superficial � metrics � •  Superficial � metrics � 

–  Cannot � be � changed � –  Non-executable � –  Lack � of � benchmark � 

•  KPI � –  Focusing � on � commercial � 

purpose � –  Approved � by � 

management � –  Executable � –  Benchmark � available � 

Fine � operation � 

Superficial � metric � 

DAU � 

Registration � 

Downloads � 

Conversion � rate � 

Cost/Revenue � 

CAC  VS  LTV？

Mone%za%on  (LTV)  

Customer  Acquisi%on  

cost  (CAC)  

•  Free  •  Virally  •  campaign  

KPI �  � needs � standardized � metrics � definition � 

•  Install  /  Sign-­‐ups  By  campaign/channel    CAC（Channel）  Conversion  （Channel）  

•  Organic  Users  •  MarkeEng  Users  •  Click  -­‐to-­‐  Install  -­‐to-­‐  Sync  •  Fake  Users  •  New  User  PercepEon  

PercepEon  by  Channel    

•  DAU  •  MAU  •  Next  Day  AcEviEes  •  Usage  

Login  Emes  Login  length  

•  Monthly  AcEve  Days    

•  DAU/MAU  •  RetenEon  

1  day/  7day  30day  

•  Engagement  Monthly  Logins  per  User  LifeEme  sessions  1～10-­‐day  acEvity  a\er  Install  

•  User  lifeEme    

Retention � Activation � Acquisition � 

ACQ = F(Campaign,channel, Users，CAC, Conv%)

ACT = F(First time Experience，Usage,Design/UX)

RET = F(User guide，operation，task，alert)

•  ARPU（Monthly）  •  ARPPU（Avg.  Revenue  per  Paying  

User  •  LTV  （lifeEme  value）  •  Virtual currency purchased/spent  

By  level/By  date  By  types  purchased  

•  Paying  users（%）  •  New  paying  users  •  Time/level  of  first  charge  •  Whale  

Revenue �  Refer � 

REV = F(Charge trap,whale, Conv%)

•  K-­‐factor  •  Invites  

Per  DAU  Per  who  send  invite  

•  Invite  accept（%）  •  Times  By  type  

Massages  E-­‐mail  

•  Cohort  by  invitee  Revenue  ARPU  

REF = F(Excitation,UX)

KPI � for � AARRR � 

Guide � 、 � first � time � 

experience � 

ACQUISITION � 

•  Social � Networks � •  Apps � Store(New/update) � •  Lowest � price � promotion, � 

Limited � Free � 

Retention � 

Emails � & � Alerts � 

•  User � guide � 

•  Task � 

•  Excitation � 

•  Viral � 

KPI � Model- � needs � professional � methodology � 

•  Ads/Campaigns � •  PR/Forum/Download � sites � •  SEO/SEM � •  EDM � 

•  Publisher � •  Traffic � exchange � 

Ads、IAP、Freemium � 

Degree � of � difficulty, � time, � 

interests � 

Red  Infinity  was  established  in  2010  and  has  rich  experience  in  product  development  and  publishing.  The  company  has  become  one  of  the  industry  leaders  in  just  one  year.    

SNS,，SLG，Poker，Puzzle

MAU 4,000,000

Make money？ Metrics-driven design

Descrip%on  ★  TouchArcade.com  HOT  NEW  GAME  ★  

Versions  IniEal  release：Sep  14,  2012  Current  version：Nov  1,  2012

Puzzle  +  BaCle    +  Collec%on  +  EDU  

 Test  in  App  store，Without  MarkeEng.

n  DAU n  Day 1 retention n  Day 7 retention n  Virtual income n  Marketing Users � 

How  to  op%mize? KPI Dashboards and Alerts – Why did it Happen? – Advantage - Dashboards

The most Important Metric between Initial release.

n The first experience � 

n WHY？ It’s been relatively Low � 

Day 1 Retention

18.2% Alert

Avg.

Bad Day 1 retention，but kind of okay after day 2. Cause: new players � 

Why？

Loading Connect

Sign up

Abruptly � lost � introduction � 

Bad Connetion

n  Move to more reliable data center n  Consider domestic and overseas server distribution n  Simplify introduction, less steps n  Embellish introduction n  Polish pet UI to make it more attractive � 

What can I do？

Advantage

Difficulty？ Operation.

n Daily awards � 

n Know players progress � 

n Pay attention to degree of difficulty of early levels � 

n  Daily task n  Login bonus n  Scheduled copy n  Bonus pet n  Friends aid

Only 9 %，WHY？

Difficulty？

Make more money.

n A/B test � 

n Drill down � 

n Whale � 

Extra  gems?

Bonus  pets?

What’s your opinions?

Compared � to � revenue � graph, � purchase � is � 

distributed � more � balanced � and � forwarding � 

Is � it � possible � to � let � more � players � make � 

purchase � earlier? � 

Immense � visual � appeal � – �  � Premade � PowerPoint � Templates � 

Standardized � metrics � Alarm � monitoring � 

Talkingdata � Game � Analytics � 

Data � statistics � 

Data � analysis � 

Data � mining � 

•  � Standard � reports � (something � wrong � with � the � game?) � 

•  � Custom � reports � (find � problem: � when? � where? � who?) � 

•  � Metrics � monitoring � (need � any � action?) � 

•  � user � segmentation � (specific � user � group � analysis) � 

•  � multi-dimensional � analysis � (multiple � dimensions � combo � analysis) � 

•  � cohort � analysis � (time � slice � analysis) � 

•  � A/B � Test � (functional � analysis) � 

•  � Statistics � (regression � analysis, � association � analysis) � 

•  � Forecast � model � (revenue/active � users) � 

•  � Pattern � recognition � (Probability � of � loss, � probability � of � paying) � 

数据运营之路依然漫长！ � TalkingData � Game � Analytics � 

� Web: � www.talkingdata.net � 

Sina � micro-blog: � Leo_Cui � 

Thank � You �