What Are Your Customers Saying about You? - RKG Summit May 2012
•Download as PPTX, PDF•
0 likes•856 views
Social media usage is exploding. All the cool kids are doing it. But how do you keep up with what people are saying about your brand? How do you know if they are positive or negative? Do you know how computers "listen" to social media? To answer those questions, and many more, you need a social media listening program. You could start with a basic tool, such as Google Alerts, but many businesses find that they need more. At this event, see real-world case studies that show social listening in action. Discover what tools are available today (and what their shortcomings are) and learn how social media listening tools are evolving. Find out how to keep up with all the conversation about your company and their industry in social media, so that you can respond to protect your reputation.
Recommended
Recommended
More Related Content
Recently uploaded
Recently uploaded (20)
Featured
Featured (20)
What Are Your Customers Saying about You? - RKG Summit May 2012
- 1. Mike Moran What Are Your Customers Saying About You? Mike Moran RKG Summit May 2012 www.mikemoran.com © 2012 Mike Moran Group LLC
- 2. Mike Moran What are they saying about you? FORUMS/NEWSGROUPS MICROBLOGS VIDEO SHARING SOCIAL NETWORKS WIKIS The Conversation SOCIAL MEDIA NEWS PHOTO SHARING AGGREGATORS MAINSTREAM MEDIA BLOGS 2 2 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 3. Mike Moran Message boards have long been complaint centers Would you have spotted this comment? Would someone know how to respond? What if that is one my competitors? 3 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 4. Mike Moran But now all you need is a phone Your customers look at reviews before going into your restaurant Or writing a review while they wait for the check 4 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 5. Mike Moran Social media now makes marketing a conversation Readers comment on your blogs They change your wikis The create blogs of their own They create “hate” sites if they don’t like you Web 1.0 users were consumers Web 2.0 users are participants 5 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 6. Mike Moran Marketers do not have message control We don’t control the message Maybe we never did The message is changed, rebutted, and misconstrued by our audience We must modify what we say in response 6 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 7. Mike Moran Social media is exploding 164 million blogs (over 1M new posts each day) Source: Invesp, July 2011 250 million tweets per day (almost tripled in one year) Source: Twiter, October 2011 800 million Facebook users (half are daily users) Source: Facebook, December 2011 YouTube serves three billion videos per day Source: YouTube, December 2011 695,000 status updates on Facebook—every second! Source: Barry Ritholtz, December 2011 7 © 2012 Mike 7 © 2012 Mike©MoranMike Moran 2008 Group LLC
- 8. Mike Moran The blueprint for social media marketing Listen Mobilize Engage Measure 8 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 9. Mike Moran The blueprint for social media marketing Listen Mobilize Engage Measure 9 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 10. Mike Moran Ford contains a potential disaster A Ford fan site goes public with a Ford cease and desist order that goes viral on the Web Scott Monty of Ford tweets “not good” when he first hears the story Later, his legal team explains that the fan site was selling counterfeit Ford items and he tweets that Within 24 hours, the story is dead, with Ford’s reputation intact 10 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 11. Mike Moran Graco handles a recall online Imagine the nightmare of baby strollers recalled for safety reasons Graco jumped on Twitter and responded to every nervous tweet, requesting serial numbers and providing advice Afterwards, as many stories praised Graco as slammed them 11 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 12. Mike Moran Comcast’s rude introduction Comcast’s first exposure to social media came from a YouTube video of a service man sleeping on the customer’s couch They later became one of the first companies to pioneer customer service on Twitter 12 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 13. Mike Moran USAA has a social voice of the member program Executives Functional Leaders Regional Leaders Real-time Trend Reports Analysis Unified Information Access Product Websites Social Media Surveys and Focus Groups • Product/Service Ratings & Reviews Customer Comments on: • Customer Discussion Forums • Facebook Fan pages • Blogs • Customer Article/Blog Comments • Twitter • Flickr 13 13 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 14. Mike Moran Listening tells you the lay of the land Learn what customers think Decide what you’d like to change about that 14 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 15. Mike Moran How do you start listening? Find your friends Use search Follow your favorite bloggers Approach 1: Listen by person 15 15 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 16. Mike Moran Influencers of the conversation can be scored Reach » Site traffic • Influencers vary by » Followers/Friends conversation » Number of social media venues • Quantitative scoring Engagement Authority can reveal the people » # of relevant » Online rank who make the messages » Number of back- » Frequency of links conversation conversation » Respect or standing within community • Human analysts can Connectivity » Bloggers blogroll provide qualitative » Cross-topic scoring, also connectivity » Influence flow 16 © 2012 Mike Moran © 2008 Mike
- 17. Mike Moran Influencers have relationships with each other 17 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 18. Mike Moran Suppose you don’t know who to follow? See what’s happening now with Twitter search Use hashtags Approach 2: Listen by topic 18 18 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 19. Mike Moran Engagement tools do simple listening Choose people to follow Enter simple keywords or hashtags for subjects Only finds Twitter data, with possibly some blogs or Facebook Hootsuite is a prime example Seesmic is also popular Tweetdeck being purchased by Twitter 19 19 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 20. Mike Moran Google Alerts are comprehensive, but slower Google Alerts are free and easy, but not realtime Set up a search and follow the e-mails or an RSS feed Perfect for small businesses and unique search keywords 20 20 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 21. Mike Moran But other companies fail with search algorithms “T-Mobile” will be found quite easily “Sprint” not so much “Verizon Wireless” is also not easy to isolate 21 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 22. Mike Moran How can you compare volumes? You want to say, “Our volume of conversation is bigger than our competitors” But… You can’t ensure the accuracy of your dataset The volume goes up all the time anyway 22 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 23. Mike Moran And no technology sees all the data, anyway Public: Twitter, blogs, YouTube, most message boards Private: Most Facebook, most LinkedIn, some message boards In between: Reviews 23 23 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 24. Mike Moran And all mentions of your brand are not good Top Five Tech Pundits in Smart Phone Conversation Blog URL Net Sentiment Rank Traffic (toward Brand) Venue falls in the top 1% of highest Engadget A trafficked, most influential sites Venue falls in the top 1% of highest Gizmodo A trafficked, most influential sites Venue falls in the top 10% of high Electronista B trafficked sites Venue falls in the top 10% of high UberGizmo B trafficked sites Venue falls in the bottom 90% of Switched C trafficked sites 24 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 25. Mike Moran Do you want to measure postings or mentions? [Negative for Treximet Cardio Side Effects] “I did a little research of my own, and was a bit alarmed to come upon a forum of migraine sufferers who had tried the drug and reported tightness in the chest and other indications of cardiac disturbance. I decided to leave the Treximet in the desk, untouched. “ Blog.lazyharpy.com, published on 18-02-2009 [Positive for Cardio Side Effects] “For some reason, today I picked it up. I did a quick web search for adverse side effects of Imitrex, which I have used for years, and felt surprised to see the very same descriptions as those accompanying the Treximet. Since Imitrex has never bothered me,..” Blog.lazyharpy.com, published on 18-02-2009 • Do you care if the whole post is positive or negative? [Positive for Treximet Effectiveness] “…I popped a Treximet, slanted my shades, closed and locked my office door, and put • Or what the specifics are for my head down for fifteen minutes. When my alarm went off, my each issue? head was perfectly clear. That was four hours ago.” Blog.lazyharpy.com, published on 18-02-2009 25 25 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 26. Mike Moran What metrics matter? Trends Sentiment Share of Voice 26 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 27. Mike Moran What do you need to use listening for? Market Research Crisis Management Product Development Sales Leads Reputation Management Recruiting 27 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 28. Mike Moran What do you need to use listening for? Need High Relevance Less Relevance OK Market Research Crisis Management Product Development Sales Leads Reputation Management Recruiting Aggregated data OR Individual postings 28 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 29. Mike Moran Algorithmic sentiment analysis misses sarcasm “Oh, the iPhone is a beautiful girl, no doubt.” The automated sentiment analysis failed to identify the sarcasm and coded the entry as positive for iPhone, while failing to understand the author was actually saying there was no value for the iPhone beneath its flashy exterior 29 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 30. Mike Moran Algorithmic sentiment analysis misses nuance “I waited on line for You know that my entire lunch hour these are negative, at my Wells Fargo but there is no word branch today.” to tell the algorithm These would be marked neutral by “State Farm never most algorithms told me I had no flood coverage.” “Amazon wouldn’t refund my money.” 30 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 31. Mike Moran Algorithmic sentiment misses context The same words mean different things An “unpredictable” movie is good, but “unpredictable” food quality, not so much We like “small” cell phones but not “small” hotel rooms “Faded” jeans are good, but not “faded” photos “Frozen” computers are bad, but “frozen” margaritas are good 31 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 32. Mike Moran The problem: Algorithms alone fall short 70% accuracy on relevance 70% accuracy on sentiment 70% times 70% = 49% “Oh, the iPhone is a beautiful girl, no doubt.” The best algorithms seem to fail half the time 32 © 2012 Mike©2010 Mike Moran © MoranMike LLC 2012 Group 2008
- 33. Mike Moran Human analysts can correct the algorithms If you need the data to be right, you need people to check the machines The machines collect the data and make the easy calls, and they suggest the answers for the tough ones, but humans make the final decision 33 33 © 2012 Mike©2010 Mike Moran © MoranMike LLC 2012 Group 2008
- 34. Mike Moran But that can be expensive You can’t afford to have people look at everything What do you instead? Sampling Machine learning 34 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 35. Mike Moran Machines can learn from the humans Machine learning algorithms can detect patterns where human analysts corrected the machines That feedback can then be used to update the computer algorithms so the computers are more accurate on the first try But…you need accurate training data—sometimes lots of training data 35 35 © 2012 Mike©2010 Mike Moran © MoranMike LLC 2012 Group 2008
- 36. Mike Moran How does machine learning work? Supervised learning corrects the machine Unsupervised finds unknown patterns Semi-supervised corrects when the machine is unsure 36 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 37. Mike Moran Still don’t care what people are saying? You can take this approach No one can force you to listen But you might be interested in knowing who is listening… 37 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 38. Mike Moran Yep, the big G… Google has is listening ears on No one knows everything Google might be listening to, but… …two we have evidence for: Sentiment of links Human ratings 38 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 39. Mike Moran Google was embarrassed by negative links Making customers irate yields links Google said sentiment analysis wouldn’t work But many believe they use it now 39 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 40. Mike Moran Google Panda listens to human raters Human beings rate a small subset of search results: Nice design? Speedy response? Quality content? Would you return? Sites that people like get bumped higher in ranking The sites they don’t like are shoved down 40 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 41. Mike Moran Machine learning scales the human ratings Even Google can’t afford human ratings for every page for every search So, it looks for patterns—common features If your site looks like the low-rated sites, your site gets ranked lower 41 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 42. Mike Moran What is the practical effect of Panda? Sites that ranked highly with the old algorithm have been affected If your site was great for search engines, but not for actual people, time to up your game Who seemed to get hit? “Content farms” and screen-scrapers Older content Sites loaded with ads Vertical search sites—but not Google sites! 42 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 43. Mike Moran Next: Listen to non-text objects Audio Video Images 43 43 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 44. Mike Moran Next: Cross national languages Those who need to know can’t speak every language Machine translation crosses the gap Automation will be augmented with human beings at first 44 44 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 45. Mike Moran Next: Predictive modeling Sales Brand Tracking Conversion/Analytics Traffic Conversation Mining 45 45 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC
- 46. Mike Moran Read all about it “Buy this book, read it, and then The read it again.” search --Chris Sherman, Search Engine Watch marketing Updated at each printing best seller Miami Herald: A Top Biz Book of 2007 “Great book.” --Robert Scoble, Scoblizer blog Web: mikemoran.com “Act now and read Twitter: @mikemoran it.” --Bryan Eisenberg, Author of #1 best seller Waiting for Your Cat to Bark? Blog: biznology.com 46 © 2012 Mike©MoranMike Moran © 2012 Mike 2008 Group LLC