Big Data in Marketing
April 2013
tip-off
INTRODUCTION
•    Doug Denton
•    Level Seven
•    Interactive marketing for 6 years
•    eCommerce brought me to marketing
•    Technical education and background
•    Practice Lead – Big Data



       NOW YOU KNOW WHERE I AM COMING FROM
AGENDA
•    Define Big Data
•    Open vs closed loop in social media marketing
•    Demonstrate simple example of closing the loop
•    Consider a couple of example cases
•    Look at the technologies that enable the capabilities
•    Talk about smart ways to move forward



          B2B & B2C UTILIZE THE SAME PRINCIPALS
BIG DATA
•    Data that IT historically ignores
•    Too fast, too different, too much to handle
•    Represents 80% of all data
•    Revolutionary tools now available
•    Very different approach to data processing
•    Very different way of thinking about data
•    A VERY BIG DEAL

            YOU WERE BLIND, BUT NOW YOU SEE
THE OPEN LOOP
•    Social media campaign
•    Participants are engaged and are spreading the word
•    Building good “buzz” – message is viral
•    Conversion rate is low

•  Typical “brand building” scenario



            LOTS OF BANG, NOT SO MUCH BANK
CLOSING THE LOOP
•    Allow the viral nature continue
•    Identify active (and positive) participants
•    Engage and reward desired behavior (1 to 1)
•    Drive a second wave of buzz using response to reward
•    “Constructive interference” gives buzz a double-peak
•    Reward has conversion with measurable value



            MAKE THE INTERACTION PERSONAL
the mechanics
MAKE THE CONNECTIONS
•  Monitor social channels for mentions of campaign,
   products, and brand
•  Confirm positive sentiment
•  Reward individuals for their participation in real time
•  Track and measure conversion




          IMMEDIATE POSITIVE REINFORCEMENT
NOW YOU TRY IT!
•  Step 1: In Twitter, follow @L7BigData
•  Step 2: Tweet using hashtag #L7Rocks
•  Options to engage you:
   •    Direct message to your account
   •    Tweet @you
   •    Retweet you
   •    Favorite your tweet
   •    Advertise to you (promote tweets)

                    THAT IS HOW IT WORKS
WHAT JUST HAPPENED?
•    Custom application monitoring Twitter for #L7Rocks
•    You Tweeted and were detected via direct Twitter API
•    @L7BigData favorited your tweet
•    You following @L7BigData -> Direct Message response
•    Otherwise Tweet @you
•    The exchange is logged for future use



                  THAT IS HOW IT WORKS
Oreo Super Bowl win
OREO AT THE SUPER BOWL
•    “Whisper Fight” commercial ran during first half
•    55,000 tweets in 10 minutes (110/second)
•    20,000 Instagram followers in 2 minutes
•    Cost: $$$ millions $$$

•  Traditional advertising was a success



                THEN THE LIGHTS WENT OUT!
PREPARATION WAS KEY
•    Oreo brand team is social media sophisticated
•    Creative, technical, management all in a war room
•    Ready to respond to an opportunity to amplify impact
•    Ready to compete (Audi, Tide, Walgreens, et al)

•  Within minutes: “Power out? No problem. You can still
   dunk in the dark.”

     FORTUNE FAVORS THE PREPARED MIND – LOUIS PASTEUR
DUNKING IN THE DARK
•  15,000 Twitter retweets
•  8,000 Twitter follows
•  6,200 Twitter favorites
•  5,500 Facebook shares
•  19,000 Facebook likes
•  14,000 Instagram follows



   MULTIPLIED VALUE OF ADVERTISING INVESTMENT
WHAT WAS MISSING?
•    Engaging directly with the participants in real time
•    Highlighting involvement of participants (retweets, likes)
•    Collecting participants for future campaigns
•    One-to-one incentives to monetize the buzz




        BIG DATA IN MOTION & SOCIAL MEDIA FEEDS
customer service
REGIONAL UTILITIES
•    Monitor for mentions of problems like service outages
•    Limit social media feeds to geography of interest
•    Analyze message sentiment
•    Select appropriate response
•    Engage in a conversation




      PERSONAL ATTENTION WINS CUSTOMER LOYALTY
UTILITY EXAMPLE
•  Power goes out in Mentor
•  Wife (Barb) calls me at office in Independence
•  I tweet to see what others are saying about a power
   outage in Mentor
•  Barb posts about the outage in Facebook
•  I Twitter search for #power #outage #firstenergy



     CLEAR CONCERN ABOUT FIRST ENERGY SERVICE
UTILITY EXAMPLE (2)
•  Twitter Ad from First Energy shows up at the top of my
   search results
•  Lots of tweets about the outage are in my search
   results (but none from First Energy)
•  I get a Direct Message from First Energy with a link to
   report or get information regarding an outage
•  I click the link

 FIRST ENERGY FOUND ME IN THE CHANNEL OF CHOICE
UTILITY EXAMPLE (3)
•  Link takes me to a website where
   •    I see a dashboard of reported outages
   •    I can login using Google, Facebook or Twitter credentials
   •    I can register for updates regarding future outages
   •    I can download a mobile application for FE customers
•  I see that Barb just registered against our account, too



                       FULLY CONNECTED
customer sentiment
WHAT IS IT?
•  Identify and extract subjective information
•  Text analytics, natural language processing,
   computational linguistics
•  Determine the attitude of the writer
•  Classify the polarity of the text (pos, neg, neutral)




          HUMANS DISAGREE 21% OF THE TIME!
HOW DO I DO IT?
•    Mine message boards and social media networks
•    Find mentions of brand, products, campaigns
•    Determine positive, negative, neutral attitudes
•    Compare to competition, historical trends




              BIG DATA MAKES THIS POSSIBLE
WHAT CAN IT TELL ME?




BRAND HEALTH, BRAND DIFFERENTIATION
making it work
GETTING THE DATA
•  Social media channels – direct connections (APIs)
•  Social media channels – GNIP (www.gnip.com)
•  Blogs and boards – Boardreader
   (www.boardreader.com)




    MASSIVE AMOUNTS OF DATA, CAN BE REAL-TIME
PROCESSING THE DATA
•  Massive amounts of loosely-structured data
   •    Big Data tools (Hadoop, Map-Reduce, Hive, JACL, etc)
   •    Enterprise tools (IBM BigInsights)
   •    BI tools (Cognos, SAS, SPSS)
•  Massive amounts of data in motion
   •    IBM Streams
   •    StreamSQL



               BIG DATA MAKES THIS POSSIBLE
closing thoughts
MOVING FORWARD
•  Tools are ready for market
•  Now is the time to take initial steps
   •    Proofs of concept, prototypes
   •    Quick wins
   •    Gain understanding
•  Define your strategy
•  Create a roadmap for adoption


 TOOLS ARE READY – NEED TO UNDERSTAND THE VALUE
questions?
thank you

Level Seven - Big data in interactive marketing

  • 1.
    Big Data inMarketing April 2013
  • 2.
  • 3.
    INTRODUCTION •  Doug Denton •  Level Seven •  Interactive marketing for 6 years •  eCommerce brought me to marketing •  Technical education and background •  Practice Lead – Big Data NOW YOU KNOW WHERE I AM COMING FROM
  • 4.
    AGENDA •  Define Big Data •  Open vs closed loop in social media marketing •  Demonstrate simple example of closing the loop •  Consider a couple of example cases •  Look at the technologies that enable the capabilities •  Talk about smart ways to move forward B2B & B2C UTILIZE THE SAME PRINCIPALS
  • 5.
    BIG DATA •  Data that IT historically ignores •  Too fast, too different, too much to handle •  Represents 80% of all data •  Revolutionary tools now available •  Very different approach to data processing •  Very different way of thinking about data •  A VERY BIG DEAL YOU WERE BLIND, BUT NOW YOU SEE
  • 6.
    THE OPEN LOOP •  Social media campaign •  Participants are engaged and are spreading the word •  Building good “buzz” – message is viral •  Conversion rate is low •  Typical “brand building” scenario LOTS OF BANG, NOT SO MUCH BANK
  • 7.
    CLOSING THE LOOP •  Allow the viral nature continue •  Identify active (and positive) participants •  Engage and reward desired behavior (1 to 1) •  Drive a second wave of buzz using response to reward •  “Constructive interference” gives buzz a double-peak •  Reward has conversion with measurable value MAKE THE INTERACTION PERSONAL
  • 8.
  • 9.
    MAKE THE CONNECTIONS • Monitor social channels for mentions of campaign, products, and brand •  Confirm positive sentiment •  Reward individuals for their participation in real time •  Track and measure conversion IMMEDIATE POSITIVE REINFORCEMENT
  • 10.
    NOW YOU TRYIT! •  Step 1: In Twitter, follow @L7BigData •  Step 2: Tweet using hashtag #L7Rocks •  Options to engage you: •  Direct message to your account •  Tweet @you •  Retweet you •  Favorite your tweet •  Advertise to you (promote tweets) THAT IS HOW IT WORKS
  • 11.
    WHAT JUST HAPPENED? •  Custom application monitoring Twitter for #L7Rocks •  You Tweeted and were detected via direct Twitter API •  @L7BigData favorited your tweet •  You following @L7BigData -> Direct Message response •  Otherwise Tweet @you •  The exchange is logged for future use THAT IS HOW IT WORKS
  • 12.
  • 13.
    OREO AT THESUPER BOWL •  “Whisper Fight” commercial ran during first half •  55,000 tweets in 10 minutes (110/second) •  20,000 Instagram followers in 2 minutes •  Cost: $$$ millions $$$ •  Traditional advertising was a success THEN THE LIGHTS WENT OUT!
  • 14.
    PREPARATION WAS KEY •  Oreo brand team is social media sophisticated •  Creative, technical, management all in a war room •  Ready to respond to an opportunity to amplify impact •  Ready to compete (Audi, Tide, Walgreens, et al) •  Within minutes: “Power out? No problem. You can still dunk in the dark.” FORTUNE FAVORS THE PREPARED MIND – LOUIS PASTEUR
  • 15.
    DUNKING IN THEDARK •  15,000 Twitter retweets •  8,000 Twitter follows •  6,200 Twitter favorites •  5,500 Facebook shares •  19,000 Facebook likes •  14,000 Instagram follows MULTIPLIED VALUE OF ADVERTISING INVESTMENT
  • 16.
    WHAT WAS MISSING? •  Engaging directly with the participants in real time •  Highlighting involvement of participants (retweets, likes) •  Collecting participants for future campaigns •  One-to-one incentives to monetize the buzz BIG DATA IN MOTION & SOCIAL MEDIA FEEDS
  • 17.
  • 18.
    REGIONAL UTILITIES •  Monitor for mentions of problems like service outages •  Limit social media feeds to geography of interest •  Analyze message sentiment •  Select appropriate response •  Engage in a conversation PERSONAL ATTENTION WINS CUSTOMER LOYALTY
  • 19.
    UTILITY EXAMPLE •  Powergoes out in Mentor •  Wife (Barb) calls me at office in Independence •  I tweet to see what others are saying about a power outage in Mentor •  Barb posts about the outage in Facebook •  I Twitter search for #power #outage #firstenergy CLEAR CONCERN ABOUT FIRST ENERGY SERVICE
  • 20.
    UTILITY EXAMPLE (2) • Twitter Ad from First Energy shows up at the top of my search results •  Lots of tweets about the outage are in my search results (but none from First Energy) •  I get a Direct Message from First Energy with a link to report or get information regarding an outage •  I click the link FIRST ENERGY FOUND ME IN THE CHANNEL OF CHOICE
  • 21.
    UTILITY EXAMPLE (3) • Link takes me to a website where •  I see a dashboard of reported outages •  I can login using Google, Facebook or Twitter credentials •  I can register for updates regarding future outages •  I can download a mobile application for FE customers •  I see that Barb just registered against our account, too FULLY CONNECTED
  • 22.
  • 23.
    WHAT IS IT? • Identify and extract subjective information •  Text analytics, natural language processing, computational linguistics •  Determine the attitude of the writer •  Classify the polarity of the text (pos, neg, neutral) HUMANS DISAGREE 21% OF THE TIME!
  • 24.
    HOW DO IDO IT? •  Mine message boards and social media networks •  Find mentions of brand, products, campaigns •  Determine positive, negative, neutral attitudes •  Compare to competition, historical trends BIG DATA MAKES THIS POSSIBLE
  • 25.
    WHAT CAN ITTELL ME? BRAND HEALTH, BRAND DIFFERENTIATION
  • 26.
  • 27.
    GETTING THE DATA • Social media channels – direct connections (APIs) •  Social media channels – GNIP (www.gnip.com) •  Blogs and boards – Boardreader (www.boardreader.com) MASSIVE AMOUNTS OF DATA, CAN BE REAL-TIME
  • 28.
    PROCESSING THE DATA • Massive amounts of loosely-structured data •  Big Data tools (Hadoop, Map-Reduce, Hive, JACL, etc) •  Enterprise tools (IBM BigInsights) •  BI tools (Cognos, SAS, SPSS) •  Massive amounts of data in motion •  IBM Streams •  StreamSQL BIG DATA MAKES THIS POSSIBLE
  • 29.
  • 30.
    MOVING FORWARD •  Toolsare ready for market •  Now is the time to take initial steps •  Proofs of concept, prototypes •  Quick wins •  Gain understanding •  Define your strategy •  Create a roadmap for adoption TOOLS ARE READY – NEED TO UNDERSTAND THE VALUE
  • 31.
  • 32.