Milano, 04 Dec 2013 1 The Power of Social Data: Transforming Big Data into Decisions Andreas Weigend bit.ly/weigenditalia
1. Data and Decisions Value of Data? Agenda 2. Amazon as Data Refinery Equation of Business 3. Implications of Social Data Revolution Audience Connected Individuals and Context 4. Summary Questions via Twitter, use @aweigend 2
3 15 years ago: Connecting Pages (Google) 10 years ago: Connecting People (FB) 5 years ago: Connecting Apps (Apple) Now: Connecting Data
4 Today, in a single day, we are creating more data than mankind did from its beginning through 2000
Mobile Context: Many sensors Identity: Proxy for person Easy for advertiser to reach user, but high cost of interrupt if inappropriate Easy for user to contribute 5
Social Data: Two Meanings 1. Relationships between people ( social graph, e.g., on Facebook or LinkedIn) 2. Data people share (or socialize, e.g., checkin, purchase, book review, picture) ------------------------------------------------------------------------------------------------------------------------------------------------------------- Note: Social Media differs from Social Data (e.g., GPS) 6
Social Data Revolution Google has changed the way a billion people think about information Facebook has changed the way a billion people think about identity Amazon has changed the way a billion people think about purchases 7
8 1. Transport energy Industrial Revolution Production 2. Transport bits Information Revolution Communication 3. Create (and share) bits Social Data Rev
Data and Decisions Rule #1: Start with a question, not with the data E.g., Which route do I take? E.g., Who do I work with? 9
10 Mindset Skillset Toolset Dataset
Big Data: Mindset to turn Mess into Decisions <when>2013-05-28t00:17:08.341-07:00</when> <gx:coord>11.0955646 47.4944176 0</gx:coord> <when>2013-05-28t00:46:14.410-07:00</when> <gx:coord>11.0894932 47.4880099 0</gx:coord> <when>2013-05-28t00:47:14.425-07:00</when> <gx:coord>11.1069126 47.5154249 0</gx:coord> 11
Berkeley SF Home Facebook Stanford Google 12
Imagine you had your geolocation from the last decade readily available at your fingertips What question would you ask? How would knowing that it is recorded 24/7 change your behavior? 13
14 London 1854
google.com/history 15,317 searches 17
What data would you pay for most? 1. Geolocation: Where did a customer go? 2. Search history: What did she search for? 3. Purchase history: What did she buy? 4. Social graph: Who are her friends? 5. Demographics and similar attributes 18
Big Data = Mindset to turn Mess into Decisions External (facing the outside) Internal (within the company) 19
20 What changed? The Journey of Amazon
What changed? Algorithms Data The Journey of Amazon AI BI CI DI 21
The Journey of Amazon What changed? Algorithms Data AI BI CI DI What did not change? Ask for forgiveness, not for permission True customercentricity Recommendations and Discovery 22
1. Data and Decisions Value of Data? Agenda 2. Amazon as Data Refinery Equation of Business 3. Implications of Social Data Revolution Audience Connected Individuals and Context 4. Summary Questions via Twitter, use @aweigend 23
Amazon as Data Refinery Goal: Help people make better decisions Data strategy: Make it trivially easy to Contribute Connect Collaborate 24
Equation of Business Expresses business strategy, values etc. Needed for evaluation of experiments Rule #2: Base the equation of your business on metrics that matter to your customers 25
26 Equation of Business Rule #3: Focus on decisions and actions, and design for feedback
5 Stages of Amazon Recommendations 1. Manual (Experts) 2. Implicit (Clicks, Searches) 3. Explicit (Reviews, Lists) 4. Situation (Local, Mobile) 5. Social graph (Connections) 27
Amazon s Share the Love Social Commerce
29 The 4 C s Content Context Connection Conversation
30 2000 Markets are Conversations 2013 Conversations are Markets
Where are the Conversations? Company Consumers
1. Data and Decisions Value of Data? Agenda 2. Amazon as Data Refinery Equation of Business 3. Implications of Social Data Revolution Audience Connected Individuals and Context 4. Summary Questions via Twitter, use @aweigend 32
Data sources for marketing a new phone product Segmentation (Demographics, Loyalty) Social Graph (Who called whom?)
Adoption rate 4.8x 1.35% 0.28% Segmentation Social Graph
35 Shift in Mindset Non-Social: Audience Social: Connected Individual
1993 On the Internet, nobody knows you re a dog
2013 On the Internet, everybody knows you re a dog
38 Shift in Identity Non-social: Attributes Social: Relationships
39 Shift in Business Models Non-social: hotels.com, craigslist Social: airbnb, lyft, relay rides,
E, Me, We! 1. Digitize: E-commerce Focus on company and products 2. Share: Me-commerce Focus on consumer and attributes 3. Connect: We-c0mmerce Focus on connection between consumers 40
41 Connected Individuals Rule #4: Embrace transparency: Make it trivially easy for people to connect, contribute, and collaborate
1. Data and Decisions Agenda 2. Amazon as Data Refinery 3. Implications of Social Data Revolution 4. Outlook and Summary Last chance to tweet questions, @aweigend 42
GLΛSS 43
The 4 Data Rules 1. Start with a question, not with the data 2. Base the equation of your business on metrics that matter to your customers 3. Focus on decisions and actions, design for feedback 4. Embrace transparency: Make it trivially easy for people to connect, contribute, and collaborate 45
Some Data Beliefs 1. Let people do what people are good at, and computers do what computers are good at 2. Build stuff that enables a future you want to live in 3. Give data to get data 46
Questions for you 1. Do your customers understand the value they get when they give you data? 2. Does your product or service get better over time and with data, or worse? 47
Questions for me? Andreas Weigend weigend.com Social Data Lab aweigend@stanford.edu 48
Data Scientist Data literate Able to handle large data sets Understands domain and modeling Want to communicate and collaborate Curious with can-do attitude 49
Data Science vs Business Intelligence 50
Data Science vs Business Intelligence 51