Leveraging Analytics In Gaming - Tiny Mogul Games
Post on 29-Jul-2015
466 Views
Preview:
Transcript
Analy&cs in Gaming Rajdeep Gumaste
Product Manager -‐ TMG
What is Analy;cs? • Wiki says – “The process of discovery and communica;on of paBerns in data” – We agree
• Metrics to measure different quan;;es • Some metrics considered KPIs for benchmarking • Metrics captured directly by third party services (like GA) and by tracking in-‐game ac;vity of users on servers
GeKng Started – What happens if you do not ‘Analy;cs’ ?
Who decides what goes in ?
No ..really .. Why Do We need it?
• To be aware of what works and what doesn’t • To target users in a smarter way – for promo;ons,
acquisi;ons etc. • To roll out features that maBer
• Anybody involved in the project can contribute to the process of seKng up an in-‐game analy;cs system
• The design and product teams work in tandem as they define the
iden0ty of the game and need a set of specific metrics to validate the same
How should it behave (or ideally should behave)?
Be Accurate
Be Quick
How does the system work?
Product Goals
Translate into data
requirements
Incorporate into the system
Analyze results
Pillars of Life ( in this scenario)
• Acquisi&on – New installs • Reten&on – % people installing on day X ,ac&ve on your app today • Engagement – Time based % of returning users (weekly or Monthly) • Mone&za&on – Show me the money !
A Case Study
Some thumb rules
• Always have a hypothesis (or hypotheses) for a problem – Don’t fret about geKng this (them) wrong. Its what we are going to
prove/not prove using the analysis
• Make sure you use the right amount of data for your analysis – Too Much : Needlessly cumbersome and maybe inconclusive – Too liBle : Inconclusive
• Explore all possible reasons for a par;cular finding in the data • There may not always be a problem to solve. It can be used to
do usual tasks in a beBer way.
Situa;on 1
111 129 120 125
157 136 130 122
105
148
121 141
154 154 134
122 126 127 150
0 20 40 60 80 100 120 140 160 180
New installs
New installs
• In talks with vendors for acquisi;on drive • Very important to acquire the right kind of users if you want to make the game
have stable numbers ajer the acquisi;on has happened
Hypothesis 1 : Acquisi;on : A discovery driven approach I think a geographical skew exists in my user base. I need to target my acquisi;ons in a smarter way
Approaches : • Look at Geographical skew in, Ac;ve users, New users, Engagement • Depending on whether we want the maximum people to download, or s;ck around
ajer downloading ,we can choose the target loca;on
• Further Analyses : • Deep dive into the data to see if a carrier skew exists
City DAU WAU Installs Weekly Eng.
Delhi 500 3500 300 14%
Pune 200 2000 200 10%
Mumbai 150 2500 180 6%
Bangalore 550 1000 50 55%
Chennai 400 1500 50 27%
1800 10500 17%
Problem 2 : Reten;on is dipping very low from D3 to D4 resul;ng in very low D7 – mul;ple reasons – content, Game features, player aspira;on
45
40
35
22 20
18 16
0
5
10
15
20
25
30
35
40
45
50
D1 D2 D3 D4 D5 D6 D7
Dx Reten;on
• The D1 , D7 numbers are respectable on their own
• This is not a problem that will break the game
• Do we s;ll want to go ahead and run stats? – Of Course!
Hypothesis : My users are exhaus;ng the content very fast
Findings/insights : • Ini;al hypothesis was wrong – ques;ons unanswered • Further Analyses :
• CIR also low for certain topics – ques;on difficulty problem • # of people playing each topic – Is nature of the topic a problem?
Topic Name Total
ques;ons Ques;ons answered CIR
Ques;ons unanswered
Topic 1 200 78 1 122
topic 2 200 165 0.8 35
topic 3 200 50 0.5 150
topic 4 200 44 0.3 156
topic 5 200 150 1 50
Problem 3 : Low Weekly Engagement : % of weekly ac;ve users playing today (DAU/WAU)
20
25 28
15
26
21 20
0
5
10
15
20
25
30
06/02/15 06/01/15 5/31/2015 5/30/2015 5/29/2015 5/28/2015 5/27/2015
Weekly Engagement
Weekly Engagement
• Avg. Weekly engagement is around 20%, meaning only 1/5 of the users in the last week have returned to play the game
• Problem : In the presence of content and adequate player aspira;on, the features in the game are the likely cause
Hypothesis : Users are only interested in playing with friends as opponents
Findings: • The finding seems to be inline with the data extracted. Hypothesis confirmed
• Dis;nct difference in the behavior of users playing challenges with friends and random players
• Further Analyses : • Look into why people are sending such few challenges to friends • Check if a loca;on skew exists for players matchmaking
Random Topic Name
Challenges sent
Challenges completed
Comple;on Rate
Topic 1 250 78 31%
topic 2 125 65 52%
topic 3 450 150 33%
topic 4 500 44 9%
topic 5 650 150 23%
Friends Topic Name
Challenges sent
Challenges completed
Comple;on Rate
Topic 1 35 34 97%
topic 2 60 50 83%
topic 3 55 50 91%
topic 4 20 16 80%
topic 5 80 65 81%
Is that all? – Not by a long shot
Ques;ons?
top related