Mobile Monetization Scenario Design & Big Data Arther Wu - PowerPoint PPT Presentation

Mobile Monetization Scenario Design & Big Data —— Arther Wu Senior Director of Monetization and Business Operation

Agenda Quick update of Cheetah Mobile Ad Scenario Design Big Data / Relation with Advertising

#1 #2 1.6B ~500M Publisher Publisher Users MAU of Mobile Tool of Mobile Apps Apps in GP in GP Worldwide

600+M 60+ downloads countries 4.7 No.1 rank Android tool app Phone Boost Junk Files Anti-virus App Manager

Rapid Growth of Global Mobile User Base Mobile MAU MM GLOBAL FOOTPRINT 494 10.7X ∼ 71% Mobile MAU Overseas 1 2015-Jun

Global Presence for Advertisers Berlin London San Francisco Beijing New Delhi Tokyo New York Paris Hong Kong Mexico Taipei City San Paolo Jakarta

Ads Mobile Monetization In-app Purchase Advertising Paid Download

US Total Media Ad Spending Share, by Media 2014 - 2019 45.0%$ 40.0%$ 35.0%$ TV$ 30.0%$ Digital$ 25.0%$ Mobile Digital$6$Mobile$ Print$ 20.0%$ Radio$ 15.0%$ Outdoor$ Dictories$ 10.0%$ 5.0%$ 0.0%$ 2014$ 2015$ 2016$ 2017$ 2018$ 2019$ Resource: eMarketer 2015 Sep

Mobile Advertising Ecosystem Advertiser ! Ad Exchange ! Publisher !

CM Cloud Data Demand Supply Advertisers Cheetah Ad Network Apps Publisher Advertiser

Cheetah Mobile’s Strength on Mobile Advertising

How Does Cheetah Mobile Ensure Ads Performance Appropriate Scenario Big Data Ad Format Design Usage

Multiple Ad Formats in Cheetah Apps

Cheetah Ads Satisfy Marketing Objectives Cheetah Ads Advertisers’ objectives Awareness Consideration Conversion Loyalty CPM / Branding

Cheetah Ads Satisfy Marketing Objectives Cheetah Ads Advertisers’ objectives Awareness Consideration Conversion Loyalty CPC / Brand Cognition

Cheetah Ads Satisfy Marketing Objectives Cheetah Ads Advertisers’ objectives Awareness Consideration Conversion Loyalty CPS / Customer Willingness

Cheetah Ads Satisfy Marketing Objectives Cheetah Ads Advertisers’ objectives Awareness Consideration Conversion Loyalty CPA / Customer Loyalty, Retention

Users’ daily life cycle CM Locker: Charging screen

Users’ daily life cycle CM Locker: Weather flow

Users’ daily life cycle Clean Master: Phone boost result page

Users’ daily life cycle CM Security: AppLock

Users’ daily life cycle PhotoGrid: Result page

Scenario Design User User’s Behavior Current Flow Scenario

Relevant Information User’s scenario Flight Search Hotel Info

User Engagement Non-intrusive Placement CM Locker’s info page Clean Master: Phone boost result

CM Launcher’s News ballon Longer Time on Page Non-interruptive Placement User Engagement

Yet, there’s NO such “law” in Scenario Design

What would people do after editing photos?

People share immediately either on social media or instant messenger.

User behavior flow cannot be interrupted But you can extend it. Take a photo Edit Share

+ = PhotoGrid PhotoGrid Social Media Community

Big data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it … - Dan Ariely, a professor of Duke Univ

Variety 4V of Big Data Mobile Unstructured Sensor Social Audio Photo Video Database Table MB GB TB PB Veracity Volume Noise Biased Anomaly Incomplete Batch Periodic Near Real-time Velocity Real-time

Big Data plays notable role in Mobile Advertising Precise Business Big Data Audience Intelligence User Profile Targeting

Big Data CM Cloud Data Demand Supply Advertisers Ad Network Cheetah Facebook Apps Twitter Publisher Advertiser

Audience Insights How to Use Big Data? Google Play App Category Data Mining Interest Tag Raw Data Profile Machine Learning Many kinds of algorithms Click Preference

Big data mining - Algorithms Classifier (Supervised Learning) Logistic Regression Gradident Boosted Decision Tree SVM Neutral Network Naive Bayes Boosting … Clustering (Un-supervised Learning) Gaussion Mixture Model K-Nearest Neighbour LDA ...

One of algorithms: Gradient Boosting Regression Trees, GBRT Android iOS Open shopping apps No or over 3 times today less than 3 User Type User Type A B Didn’t make Make an order an order Female who like to shop

What is Machine Learning? Learning = Representation Evaluation Optimization Profile Ad Performance A/B Testing

Cheetah App Cloud Cheetah App Cloud App Graph Ad Serving App Usage User Profiling Analytics Analytics 4MM+ Apps Audience Insights 4MM+ Apps Audience Insights Better User Targeting Better Performance of Apps

App Graph Ad Serving - Showroom

Audience Insights from Mobile Behavior and Consumption $$$ Social Data Model Mobile Usage Derived Model

Profile Targeting: 
 Ad Display Based on App Usage Graph In-App Tracking: 
 In-app events like purchases and sign- ups are attributed to the media source Conver'ng)Data)into) DataSync Optimization: 
 Audience)Insight Campaign data gets smarter by the second to increase ROI Retargeting: 
 Re-engage users to return to your app with targeted campaigns

Select the Right Audience 42+ proprietary tags Psychographics/ Demographics/Consumer Behavior Device/OS/Carrier Localization Purchase & Engagement $ Leverage Cheetah Orion for Precise Patterns Audience Targeting Predictive algorithm

Yield Optimization - eCPM Comparison eCPM Ad Network A Ad Network B Ad Network C Advertisers

Thanks. arther@cmcm.com