Social Data Analytics - Consumer Segmentation


Published on

My presentation at #BWELA 2011. Main points are:

1. Learn about the different social data types
2. Learn how to identify social customer segments
3. Learn how to integrate social consumer data with internal customer data

Published in: Technology, Business
1 Comment
No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Example: psychographic data from spotify auto-sharing on facebook or intent data on amazon wishlist.2 ways: explicitly provided by user, self-reported or automatically generated
  • Implies it’s a scaling issue, and it is, but not JUST a scaling data collection (technology) it’s also a talent/skills/people problem
  • Adapted from Altimeter seven data types - Not all required. To be used as a guide in understanding which data may be needed to answer questions, understand an audience segment, or consumer behavior
  • Why FB is so interesting to brands. They have a treasure trove of self reported psychographic data. It’s why Facebook advertising has grown to 2billion dollar biz in 2011
  • Sources: Plancast, Wishlists (Amazon), Eventbrite, FB Events
  • Sources: CRM databases, Sales databases, web analytics/clickstream analytics that reveal site behaviorsExamples: Amazon, Hunch, recommendation engines
  • LBS services, geo-tagged objects like Photos on Flickr (hundreds of millions)
  • Sources: Yelp, Yahoo Groups, Facebook, etc…
  • There is a precedent in web analytics foundations for segmentation, transfer this idea to social media analytics. It’s a universal construct
  • 4.and expands with other publicly available data sources such as address, phone numbers, etc…7. which customers require tier 1 support (cos don't and shouldn't treat all customers equally). Hilton (Diamond VIP), when I tweet, they should know it's coming from a Diamond VIP, not a general traveler 2 nights a week (worth 14x more annually)7.which customers amplify brand content/actions/promotions the most (make sure you're not missing these people) who shares the brand content the most? Brand Amplification Index/Score?
  • Rapleaf has data on 2 billion+ email addressesabout 400+ million people
  • Disclaimer: I am on the Advisory Board of this company
  • Source: HP study
  • Source: HP study
  • Source: HP study
  • Source: HP study
  • Source: HP study
  • Source: HP study
  • Social Data Analytics - Consumer Segmentation

    1. 1. Social Data Analytics Understanding social data types, segmenting social consumers and social data integrationNov 2011
    2. 2. Agenda• Learn about the different social data types• Learn how to identify social customer segments• Learn how to integrate social consumer data with internal customer data
    3. 3. Social Data Types (Way Beyond “Likes”)
    4. 4. Social Participation Skyrockets And so does data generation….
    5. 5. Social Data ExplosionMore than 900 million objects that More than 200 millionpeople interact with each day (pages, Tweets per daygroups, events and community pages) These activities generate valuable social data that can reveal NEW knowledge about consumer behavior More than 3 million More than 3 billion checkins per videos viewed per day daySources: Facebook, Twitter, YouTube, Foursquare
    6. 6. Example: Anatomy of a Tweet Tweet Meta-data may contain: -Demographic data -Psychographic data -Location data -Intention data -Referral data -Sharing data
    7. 7. Companies Struggling to Cope Data generation is rapidly outpacing our ability to manage and understand how to use it.Sources: IBM CMO Study
    8. 8. Resources and Needs Not Aligned The talent gap is widening. Current supply is a shallow pool that cannot meet growing demands. Hiring from the outside isn’t the silver bullet (or even a realistic option in many cases). Internal education is critical.Sources: IBM CMO Study
    9. 9. Use the Right Data for the Job Q: Is social contributing to our enterprise goals?A: Well, we have2,834 Facebook That’s not cutting it! Likes…
    10. 10. The Social Data Tree Location Behavioral Sharing Referral Intention Brand/ProductPsychographic Demographic
    11. 11. Demographic Data Age Education Gender Income Race- Used for making broad generalizations about groups of people- Most common data type used- Has limitations. Not all individuals will conform to the profile- Table stakes. Start here and layer on other data types
    12. 12. Psychographic Data Personality Lifestyles Interest Values s Attitude- Combined with demographics, used for highly targeted outreach and/or advertising- Self reported by consumers on social platforms- Gives companies opportunities to better understand and align with consumer needs, wants and expectations. (a.k.a. to be helpful and relevant).
    13. 13. Intention Data Desired Planned State Events Desired Planned Product Activities- Good for understanding what consumers “want” or “want to do”- Least accurate data type, can lack contextual relevant details (ex: in-market)
    14. 14. Behavioral Data Past actions, activities that can be used as a predictor of future intentions- Allows companies to more appropriately target segments for better marketing results- Allows companies to use personal preferences and interests to move closer to a true one-to- one relationship with their customers
    15. 15. Location Data Physical location of a consumer- Provides contextual understanding at a certain point in time- Can trigger contextually relevant promotions and/or rewards- Can be used to identify intent- Provides opportunity for brands to improve offline customer experiences by understanding behavior patterns in location and opinions shared with it
    16. 16. Referral Data Ratings Rewards Non-Verbal Reviews Gestures- Can be used to identify brand promoters, detractors- Provides insight into product/service attributes matter most
    17. 17. Brand/Product Data Brand/Product related conversations Product Attribute Conversation s Product types Measures- Can provide richer understanding of consumer perspective on specific aspects of products and/or brand measures- Represents unfiltered consumer feedback- Helpful in optimizing product launches, product development roadmaps, and content strategies
    18. 18. Sharing Data Reach Clicks Conversion Shares s Generational Sharing- Identify brand advocates that actively spread branded content- Learn how branded content travels through generational sharing- Understand word of mouth, what they share, why they share it
    19. 19. Social Data Segmentation (Segment or Die)
    20. 20. Aggregate is the Enemy “All data in aggregate is crap” –Avinash KaushikAll Visitors?Total Pageviews?Total Likes?People Talking About This? = Gratuitous Data PukingFollowers/Fans/Friends?The list goes on…..
    21. 21. Social Segmentation BenefitsWhy should Analysts care? -Find the answers to the “why” and “how” questions of consumerbehavior -Go beyond data puking! (no more “hits” reporting)Why should Marketers care? -Get closer to the nirvana of personalized, one-to-one relationship w/ consumers. Build value into every engagement! -Experiences and information need the right audience first and foremost Find the right audience. -Improved targeting (outreach/advertising) Ex: Facebook uses a combination of demo, psycho, product, location, and referral data -Identify and convert the most profitable customer segments -Prioritize consumer segments and align marketing investment against it (we can measure their outcomes!)
    22. 22. Social Segmentation Methodology 1. Find a unique ID (email, twitter handle, linked profile, etc…) 2. Search social profiles using unique id 3. Calculate and match to individual4. Expansion. Build out social profiles of specific individual, across data types mentioned above 5. Optional: content consumption/creation mapping, influence analysis 6. Scale across your unique ID list (aka customer) 7. Segment results.
    23. 23. Social Segmentation Tools/Vendors Main CapabilitiesAggregates user informationby resolving email addressesor social networking handles Some offer APIs that can match email addresses toprofile URLs across 125 social networks Can integrate with offline data sources as well(telephone, identity, location)
    24. 24. Social Segmentation Tools/VendorsCapabilities FlipTop FullContact Qwerly PeekYou Rapleaf Rapportive SpokeoSocial affiliations (profiles)DemographicsGeographicInterestsCareer/EducationWealth/FinanceHealthLinked URLsReverse email searchReverse phone searchAPI accessPartner with 3rd party infoInfluencer Measurement Available Available but not comprehensive Unavailable 2 4
    25. 25. Social Segmentation Examples People search by username/profile name Social profiles identified and tied to real identity
    26. 26. Social Segmentation ExamplesSegment Non-Social Visitors fromSocial Visitors‣ Value of FB Fans and socialengagement‣ Leverage messaging tactics forperformance improvement andincreased conversions
    27. 27. Social Segmentation Examples Empirical research suggests that the lifetime value of a promoter is worth at least 4-5x that of passives (neutral customers) or detractors (unhappy customers). Promoters buy more often, spend more, refer more, and cost less to acquire and serve.Net Promoter ScoreTM is a customer loyalty metric developed by FredReichheld, Bain & Company, and Satmetrix. It categorizes customersinto three segments; promoters, passives, and detractors.
    28. 28. Social Data Integration (bridging the gap)
    29. 29. This is just the beginning… “The merger of unstructured and structured data will fuel the nextrevolution in business intelligence” -Prasanna Dhore Global Customer Intelligence, HP
    30. 30. Why Social Data Integration?
    31. 31. Social Data Integration Framework
    32. 32. Social Data Integration Example - HP Unstructured data: product conversations from social media Structured data: purchase and service history from crm and support databases
    33. 33. Social Data Integration Example - HPClassify each product conversation from social according to product attributes
    34. 34. Social Data Integration Example - HPConnect the dots to structured sales and service data. Improve sales forecastingand service/support resource management
    35. 35. Social Data Integration Example - HPHP Social Data Integration Results:-Strong social data signals about a product were aleading indicator for increased sales andregistrations (use cases: marketing optimization by spend, audience and channel)-Strong negative sentiment in media signals abouta product were a leading indicator of increasedsupport tickets/calls (use cases: operations efficiency. Managing customer support resources)
    36. 36. Ken BurbaryVP, Group Director, Social Strategy &