Advanced integrations of social media analytics
8 likes3,059 views
The document discusses advanced integrations of social media analytics and emphasizes the importance of objective-based strategies and understanding where conversations occur within social media platforms. It outlines various social media metrics utilized to measure engagement, influence, and sentiment, along with examples of analytics tools for tracking performance. Additionally, the document provides insights on optimizing content engagement, analyzing user interactions, and measuring the impact of social media activities on brand perception and customer behavior.
Advanced integrations of social media analytics
- 1. March 15 2011 Advanced Integrations of Social Media Analytics #SOCIALMEASURE
- 2. Advanced Integrations of Social Media Analytics 1. Objectives Based Strategy 2. Understand where conversations occur 3. Social Media Metrics 4. Demos 5. Social Media Reporting 6. Shaky Metrics 7. A Number Is Not a Decision 8. Integrated Program Design
- 3. Presenters Blake Robinson, Rackspace @blake Susan Etlinger, Altimeter @setlinger Megan Costello, Crimson Hexagon @megcos Margaret Francis, MarketLive @margaretfrancis Chris Lightner, Edelman @clightner #socialmeasure
- 4. 1. Objectives Based Strategy 4
- 5. Objectives differ by level Corporate Risk management Business unit Consistency Leadership development & culture across brands Brand Social strategist & COE Community Engagement Channel focus manager & metrics education Value metrics ROI metrics Source: Jeremiah Owyang, Altimeter Group
- 6. Use appropriate metrics at each level Corporate Business metrics: revenue, CSAT, reputation. LOB/Geo Social media analytics: Insights, share Stakeholders of voice, resonance, WOM. Social Engagement metrics: Strategist/Community fans, followers, clicks. Manager Source: Jeremiah Owyang, Altimeter Group 6
- 7. The new lifetime value calculation • Percent that refer + Value of purchases • Size of their networks -Cost of acquisition • Percent of referred + Value of new customers from people who purchase referrals • Value of purchases + Value of insights • Percent that provide + Value of support support + Value of ideas ____________________ • Frequency and value of = Customer lifetime value the support Source: Charlene Li, Altimeter Group
- 8. Objective Setting Exercise Assess current state: Envision future state: 1) Categorize your use case: 1) What are you trying to social media marketing, achieve with social media? PR, brand marketing, or 2) What do you want to executive, support measure? Why? 2) What are you currently 3) What would the ideal doing with social media? social media analytics Why? program be able to tell 3) What are you currently you? measuring? Why?
- 9. 2. Understand where social media conversations occur 9
- 10. Tsunami
- 11. Ford Focus
- 12. Oreo
- 13. Where do people talk about Coke Zero? Used with permission of Netbase 13
- 14. Where should I invest my marketing $? Used with permission of Netbase 14
- 15. 3. Social Media Metrics 15
- 16. “Social” Metrics • # Video Views • # Facebook fan pages • % YouTube Favorites • # Facebook fans • % YouTube channel subscribers • Facebook Fan Rate • % YouTube Video Plays • Facebook Likes • # of YouTube Channel Comments • Email Open Rate • # of YouTube video reviews • Email Click Rate • # slideshare views • Email Forward rate • # delicious bookmarks • Email opt-out rate • # diggs • # of Email subscribers • # of Blog Mentions • # of SMS subscribers • # of Forum mentions • Clicks/ CTR • # of Facebook mentions • Impressions • # of Twitter mentions • Traffic • # @mentions • Reach • # of reviews mentions • Registration • # of comments • Opt-in • Positive : Negative Mentions for any channel • Page views • # Twitter followers • Visit/session length (% site visits > 60 secs) • Twitter Follower-Rate • Click path analysis • # reviews • Eye tracking studies • # stars in reviews • #/% Downloads • Thread size • Changes in SERP results/ rankings • Unique contributors • Earned/ Owned traffic • Unique commenters • Inbound links
- 17. Paid/ Earned/ Owned 17
- 18. Assembling Metrics into KPIs Measurement Attention Engagement Authority Influence Sentiment • Unique Visits • Total Interactions • Total “Likes” • On-Message Chris’ Slide • Impressions • External Links to • Fan • Subscribed “Likes” • Positive/Negative/ Facebook • Page Views Content & Photos/Videos • Audience Profile as Neutral • Media Discussion • Post Quality Reflection of Target • Change Over Time Consumption • Retweets • Extended Network/ • On-Message • Inbound Links • Direct Messages Influence of Followers • Positive/Negative/ Twitter • Total Followers • External • @Replies • Follower Profile as Neutral Coverage • Shared Content Reflection of Target • Change Over Time • Likes/Dislikes • Inbound Links • On-Message • Subscribers • Views • Positive/Negative/ YouTube • Comments • External • Demographic Insights • Subscribers • Audience Profile Neutral • FavoritesdReplies Coverage • Change Over Time • Inbound Links • On-Message • # of Comments Blogger • Subscribers • Audience Profile as • Positive/Negative/ • Share of Voice Relativeto Engagement • External Reflection of Target Neutral Audience Size Coverage • Change Over Time • # of Check-ins • Positive/Negative/ Foursquare/ • # of Unique • # of Repeat broadcast to • Audience Profile as Neutral Commentary Geolocation Visitor Check-ins Check-Ins Twitter or Reflection of Target w/Check-ins Facebook • # Page Views • Traffic to Site (Overall) • Profile of Visitors as • Most Popular Search Search • Inbound links • Unique Visitors • # Page Views/User Reflection of Target Terms • Time Spent on Site • Total Ongoing Engaged • Total Trackbacks • Overall Sentiment • Total Media Subscribers to Overall • Total Interactions & Coverage of • Total Shift in Consumption Content/Community Activities Sentiment • Community Crossover
- 19. EXAMPLE: CREATING A MEASUREMENT PROGRAM TRACKING PERFORMANCE AGAINST OBJECTIVES OUTCOMES AND PERFORMANCE ENGAGEMENT IMPACT How a communications program or campaign How a communications program or campaign impacts changes the behavior of the target audience or attitudes and opinions within the target audience. We impacts real world events. measure not only who the program is reaching but what DAILY MONITORING is said and when it occurs. This is where we measure Changed Behavior, Brand The measureable results of the success of our engagement. FOCUS Perception and Awareness core Digital tactics. Engagement, Reach and Perception Analysis Tools, FOCUS TOOLS Insights Surveys, Web Analytics Daily Content and - Consideration, Preference FOCUS Social Monitoring Tools, Social Community Management TOOLS and Purchase Intent CRM, Web Analytics - Increase in Share of Voice Social Monitoring Tools, -Engagement Insights and - Information Request and TOOLS Analysis EXAMPLES Lead Gen Social CRM tools - Community Growth - Campaign Objectives and - Share of Voice v Competition KPI Tracker - Actionable Insights EXAMPLES - Recommendations - Brand Perception Analysis - Trending Topics and - Favorable media coverage - Message Penetration Conversation Tone EXAMPLES -Favorable social media -Blog and Influencer Monthly Reports discussion Mentions - All up Summary of - Sentiment and Tone - User Generated Content engagement performance Weekly Reports and conversation insights Daily Reports - Summary and Analysis of - Long term strategy - Brand and Industry Buzz Weekly Conversation development based on - Brand Protection and -Engagement insights, success DELIVERABLE DELIVERABLE insights Crisis Alerts and recommendations - Performance analysis DELIVERABLE -Community Management - Community and Audience against key business and Engagement Growth objectives Opportunities -KPI Tracker - Web Analytics and - Influential Voice Campaign Report Identification 19 -KPI Tracker
- 20. EXAMPLE: REPORTING AND SOCIALIZATION MEASURING ACTIVITIES AND RESULTS AGAINST PROGRAM OBJECTIVES AND BENCHMARKS Daily Monitoring Weekly Engagement Monthly Performance Impact Analysis 20
- 21. 3. Demos 21
- 22. Shareaholic - http://www.shareaholic.com/siteinfo/usatoday.com • View the sharing stats of any sites shared through the Shareaholic browser extension • Pages are accessible at http://www.shareaholic.com/siteinfo/sitename.com
- 23. Bit.ly • Bit.ly, and other URL truncation services, provide a layer of analytics that can extend your tracking capabilities to any destination that allows clickable links. • The analytics it provides behind each link breakdown the number of clicks on your link and other truncated links to the same site, as well as referral and geographical data.
- 24. Export.ly for Facebook • Facebook analytics tools focus on the composition and size of the advertising audience, not the interactions they have with the content itself • It can be very difficult even to see the user generated content on a Facebook page via the web interface • In order to see and analyze social media content on a Facebook page, you can use Export.ly
- 25. Facebook Insights • Facebook Insights provides a top down perspective on who comprises your audience and what they do on your Facebook Page. 25
- 26. Facebook Insights for Websites • You can include a tag on your site which will allow you to measure the impact that Facebook Likes/ Share have on you site traffic. This is a brand new product from Facebook (last week) 26
- 27. RowFeeder for Twitter • Less than 18 months ago Rackspace received <500 Twitter mentions/month • Today Rackspace receives >15K Twitter mentions/ month • To deal with this volume we’ve begun aggregating this data using Rowfeeder
- 28. Chartbeat: Simple Server Side Analytics Via Tagging • Chartbeat provides an excellent way to visualize the correlations between content publishing and real time site traffic • Simple for marketers to use: put a (really light) tag on your site and Chartbeat’s server side analytics will analyze the data for you • For this demo, we are using a WordPress plugin on Scobleizer.com
- 29. PostRank for Publisher Insights • PostRank enables you to correlate site activities and social media reactions from non-Facebook channels
- 30. Lithium SMM (Scout Labs) for Social Listening • Lithium SMM (Scout Labs) will let you set up a crawl for any eligible Facebook and give you a 30-45 day back crawl for wall post data • Easy, intuitive interface • Automated sentiment and quotes extraction 30
- 31. The VI Analyze tab allows you to analyze trending conversation across multiple media types, time periods, across competitors, and across different geographic regions
- 32. The VI Dashboard allows you to configure customize reports that you can then pin to your custom brand/product dashboard
- 33. The VI Monitor tab allows you to run ad hoc reports that display results as a list, by volume by media type, by sentiment, authors and sites, and also within specific geographies or time periods
- 34. Crimson Hexagon for Social Analytics iPhone on Verizon: Key Themes Crimson Hexagon’s proprietary algorithm separates relevant from irrelevant content and measures user-defined themes in online conversation. 34
- 35. 5. Social Media Reporting 35
- 36. March 10, 2011 PRESENTED BY EDELMAN
- 37. INSIGHTS INSIGHTS
- 38. INSIGHTS
- 39. FACEBOOK CONTENT POPULARITY INSIGHTS COMMENTS AND LIKES ON CONTENT POSTED BY BRAND – SEGMENTED BY TYPE ENGAGEMENT BY ADMIN POST TYPE (AVERAGES) Likes/Post Comments/Post 1400 1200 1000 800 600 400 200 0 Link Text Video INSIGHTS AND RECOMMENDATIONS: The ―I choose to drive a this car because …‖ post drove a huge spike in engagement with the Facebook community , generating over 1,800 comments and 1,200 likes. The videos are generating a large volume of likes, but giving users a reason to comment could also be advantageous and generate attention from outside the community, growing the fan base. We also recommend posting more photos to the page as those tend to generate high engagement rates. There were no photos posted in this time period. Source: Export.ly
- 40. CONTENT POPULARITY BY AUTOMOBILE LINE INSIGHTS BREAKDOWN OF COMMENTS AND LIKES ON CONTENT POSTED BY BRAND Average # of Likes Model 8 Average # of Comments Model 7 Model 6 Model 5 Model 4 Model 3 Model 2 Model 1 0 500 1000 1500 2000 2500 INSIGHTS AND RECOMMENDATIONS: There was very high engagement with the test run video posted in February. The Model 4 which were driven by the television commercial support which generated a high volume of ―Likes‖ but didn’t see the same spike in comments. We want to focus on keeping the likes up, but bringing the ratio of Comments to Likes closer to that seen with the Model 3 which experienced a ratio of 3.4 to 1, Likes to Comments. Open ended posts such as this one from February asking fans why they choose to drive a particular car will generate high response and also bring the ratio closer together Source: Export.ly
- 41. FAN WALL POSTS VS. ADMIN WALL POSTS INSIGHTS SIDE BY SIDE COMPARISON BY HOUR AND DAY OF WEEK DAYS OF WEEK TIME OF DAY Fan Posts 90 Admin Posts 78 76 80 74 76 80 69 70 66 70 62 60 55 60 51 46 50 47 45 50 40 40 27 26 30 30 22 20 20 10 6 10 1 4 6 3 3 2 10 0 0 2 0 1 0 0 0 0 12am - 3am - 6am - 9am - 12pm - 3pm - 6pm - 9pm - 3am 6am 9am 12pm 3pm 6pm 9pm 12am INSIGHTS AND RECOMMENDATIONS: • The Facebook fans are posting on the wall every day of the • Make sure to engage earlier and continue engaging by giving the week, including Friday through Sunday which makes up 35% of community content after work hours. the content posted. The brand engagement is heavily weighted on week day posts, with 85% of posts occurring Monday through • On the next slide we’ll take an even deeper dive, looking Thursday. Extend engagement to all days of the week. at fan engagement by hour of day, specific to each day of the week. • The fans also engage at all hours of the day, but heavy engagement starts around 7am PST and continues late into the • Analysis includes 410 posts to the wall by fans, and 19 posts by evening. The brand is posting from 7am to 2pm and the brand stopping, but the fans continue to post with 20% of posts to the wall occurring between 5pm-10pm PST. Source: Export.ly
- 42. FAN POSTS BY HOUR AND DAY OF WEEK INSIGHTS DAY OF WEEK BREAKDOWN BY HOUR INSIGHTS AND RECOMMENDATIONS: SUNDAY: Heavy engagement in the evenings, with 62% of Tues - Thurs: Engagement occurs during peak work posts coming after 3pm PST hours, again starting at 7am PST but slowing at 4pm PST Monday: Fans post bright and early when they get into the Friday: Posts spike mid-day, but continue to occur in the office. Engagement spikes at 7am PST, and slows evening at a higher rate than Monday through Thursday consistently throughout the day Saturday: Continual posts throughout the day, with peaks in the morning and late evenings Source: Export.ly
- 43. COMMENTS ON WALL POSTS INSIGHTS HOURLY AND DAY OF WEEK ANALYSIS DAYS OF WEEK TIME OF DAY Links Text Photos Video 1200 1061 1800 1708 1000 1600 1400 800 710 1200 1000 600 800 400 312 600 366 214 223 175 400 286 226 239 200 158 140 157 130 128 126 208 206 128 132 200 96 35 36 6 11 0 34 2 42 13 21 3 37 10 56 51 57 35 22 7 13 1 22 2 3 23 17 21 11 7 14 5 9 3 7 7 7 7 4 0 0 12am - 3am - 6am - 9am - 12pm - 3pm - 6pm - 9pm - 3am 6am 9am 12pm 3pm 6pm 9pm 12am INSIGHTS AND RECOMMENDATIONS: This analysis includes comments in response to the posts on the fan wall from both the Page Admin and the fans. Although The large spike seen in the 12-3pm hour, and on Wednesday was 98.8% of comments were in response to posts by the Admin. caused by the response to the post asking fans why they drive a specific car. The day of week and time of day analysis was largely driven by when the brand is posting on the page. For example, updates that We recommend posting similar types of content at different times of contained a video were always posted on a Monday or the day and different days of the week so that we can better Tuesday, thus 90% of comments in response to video posts came understand when these posts get the highest engagement. This will on Mondays or Tuesdays. help inform the engagement strategy going forward Source: Export.ly
- 44. INSIGHTS
- 45. @AUTOMAKER 6 MONTH FOLLOWER GROWTH INSIGHTS @AutoMaker Twitter Followers 13,477 11,720 10,295 9,317 8,545 15% 7,843 7,035 growth 14% 10.5% growth 9.0% 9.0% growth 11.5% growth growth growth September October November December January February March INSIGHTS AND RECOMMENDATIONS: The @AutoMaker Twitter profile has grown the follower base by 92% in the past 6 months. Looking back to September, there has been a average month over month growth rate of nearly 12%. A spike in growth has recently occurred during the months of January and February, where fan growth rose at 14 & 15 percent respectively. Follower growth in January and February was up 82 percent (combined 3,182 new followers) when compared to growth in November and December, which had only 1,750 new followers. Source: TwitterCounter
- 46. @AUTOMAKER TWITTER CHANNEL OUTBOUND TWEET VOLUME INSIGHTS Tweets @AutoMaker Tweet Volume Replies 200 Retweets 180 160 140 120 126 119 100 96 103 80 60 58 60 7 12 40 26 20 29 47 39 18 18 28 0 6 3 2 September October November December January February INSIGHTS AND RECOMMENDATIONS: Consistent with follower growth, the @AutoMaker profile has seen an increase in outbound Twitter activity in the past 3 months (includes all regular tweets, @replies and retweets). During the months of December – February, outbound Twitter volume increased 68 percent, and more importantly, engagement with other Twitter profiles increased 109% which contributed to the large fan growth during that time period. @Replies volume dipped a little, so we recommend continuing to @Reply with other profiles which will help increase the influence score of the Twitter handle as calculated by tools such as Twitalyzer, Klout and TweetLevel, along with building the follower base. In addition, continue to retweet positive conversations about Brand - for example when people post that they ―Love the brand, or ―considering a purchase‖ or even just entertaining tweets such as this one This has been effective in building the follower base Source: TweetStats
- 47. FEBRUARY TWITTER REPORT TWEETS, RETWEETS, @REPLIES BY DAY INSIGHTS February Tweet Timeline Retweets Replies Tweets 14 12 10 7 7 8 6 6 5 7 6 0 6 5 0 4 4 4 5 4 2 4 4 5 0 4 2 6 4 0 0 0 5 3 3 3 2 0 0 0 3 3 3 3 3 3 3 3 2 0 0 2 0 2 0 2 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 INSIGHTS AND RECOMMENDATIONS: During the month of February, the @AutoMaker Twitter handle had a total of 96 tweets with an average 3.4 tweets per day. Source: TweetStats
- 48. FEBRUARY TWITTER REPORT OUTBOUND ENGAGEMENT VS. ALL CONVERSATION ANALYSIS INSIGHTS Outbound Hourly Tweets Total Distribution of 30 @AutoMaker Conversation 25 24 23 20 30% 25% 17 25% 21% 19% 20% 15 15% 10% 7% 9% 10% 6% 10 4% 5% 6 6 0% 5 4 9AM to Noon 3PM to 6PM 6PM to 9PM Noon to 3PM 3AM to 6AM 6AM to 9AM 12AM to 3AM 9PM to 12AM 5 3 2 2 2 1 1 0 0 0 0 0 0 0 0 0 0 0 0 INSIGHTS AND RECOMMENDATIONS: The most concentrated efforts for outbound Twitter volume occurred between the hours of 3:00 P.M. and 6:00 P.M. PST, with 67% of the outbound tweets occurring at this time. And when looking at the overall conversation volume on Twitter mentioning @automaker (chart on right), you see that 65 percent of the conversation is happening between 9 A.M – 6 P.M. It is recommended that the @automaker Twitter profile engage with users earlier in the day, as conversation seems to be peak by 3 P.M., which is the time the @automaker profile just be begins to engage. Source: RowFeeder and TweetStats
- 49. FEBRUARY TWITTER REPORT – POSTS AND IMPRESSIONS (SAMPLE) INSIGHTS Twitter Posts & Potential Impressions Total Posts Potential Impressions 50 47 90,000 45 80,000 77,729 40 70,000 35 33 32 63,413 60,000 30 53,726 26 50,000 25 47,577 40,000 20 14 30,000 15 25,712 10 10 20,000 5 2,368 10,000 - - 3/2/11 3/3/11 3/4/11 3/5/11 3/6/11 3/7/11 INSIGHTS : Between March 2, 2011 & March 7, 2011, there was a total of 162 total tweets with a reach of 270,525 impressions. Note the large spike in impressions on 3/7/11. This was caused by the tweet by a specific influential blog tweeting about the line of autos Source: RowFeeder
- 50. MOST ACTIVE & FOLLOWED ACCOUNTS MENTIONING @AUTOMAKER INSIGHTS Most Active Users Most Followed Users 5 List 69,514 List 4 14,863 removed removed 4 13,752 for for 3 12,504 privacy privacy 3 10,974 3 7,544 2 6,537 2 6,448 2 5,547 2 4,760 2 4,467 2 3,173 2 3,031 2 2,836 2 1,634 2 1,347 2 1,323 1 1,222 1 1,155 1 1,115 1 1,084 1 1,055 1 986 1 576 1 565 Updates/Person 0 2 4 6 Number of Followers - 20,000 40,000 60,000 80,000 Source: RowFeeder
- 51. TOP CONVERSATION DRIVERS ON TWITTER (SAMPLE TIME PERIOD 3/2/11 – 3/7/11 ) INSIGHTS INSIGHTS AND RECOMMENDATIONS: Total Posts & Conversation Drivers Day of Week Between March 2 – March 7, 2011, a total admired companies car jetta #vw campaign 2011 of 162 mentions. The chart at right provides a breakdown of common terms and phrases used in these tweets 10 8 6 4 2 - Time of Day 3/2/11 3/3/11 3/4/11 3/5/11 3/6/11 3/7/11 Admired Top Keywords Car Model X #auto Campaign 2011 Companies Mentions 10 20 15 12 11 12 % of Total 6% 12% 9% 7% 7% 7% Volume Source: RowFeeder
- 52. FEBRUARY EDITORIAL ANALYSIS TOP MENTIONED KEYWORDS THAT INCLUDED THE @AUTOMAKER PROFILE INSIGHTS Keyword @Automaker Usage Mentions Impression Reach Top Influencers @stevewhiteBrand Keyword 1 39 169 123,684 (1,985 followers) @kelseySchnell Keyword 2 10 27 2,883 (175 followers) @kellybluebook Keyword 3 5 55 35,026 (6,434 followers) @alovey Keyword 4 1 10 41,802 (168 followers) @m80parents Keyword 5 4 18 9,888 (3,670 followers) @grantimahara Keyword 6 1 429 875,592 (238,846 followers) @idiotsguides Keyword 7 1 88 20,746 (1,433 followers) INSIGHTS : During the month of February, the above hashtags & keywords were the most used in all outbound tweets: • #AutoMaker– had the highest usage at 39 times, yet only received 169 mentions with just 123,684 impressions • #AutoModel– was just used once, but had the highest impression reach at 876,000 • This was due to the nature of the content and popularity of the Television Commercial that supported and a number of influential people responding or retweeting the content • The misspelled hashtag, – only had 88 mentions and an impression reach of just 21,0000 Source: TweetStats and Sysomos
- 54. SHARE OF VOICE INSIGHTS TOTAL MENTIONS OF AUTO MAKER WITHIN SOCIAL MEDIA (30 DAY SAMPLE) INSIGHTS AND RECOMMENDATIONS: Day of Week Between February 4 – March 6, there were 826,468 total mentions of Auto Maker. Over 75% of mentions occurred on Twitter or on Blogs Time of Day *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 Source: Radian6
- 55. SHARE OF VOICE INSIGHTS TOTAL MENTIONS OF AUTOMAKER WITHIN SOCIAL MEDIA (30 DAY SAMPLE) Day of Week Conversation peaks on Twitter with 37,389 total mentions Mentions on Twitter top out at 11,554 with the unveiling of a new model Time of Day *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 Source: Radian6
- 57. SHARE OF VOICE INSIGHTS – MODEL CONVERSATION TOTAL MENTIONS OF AUTOMAKER & ECO FRIENDLY WITHIN SOCIAL MEDIA INSIGHTS : Day of Week Between February 4 – March 6, there were 58,387 total mentions of specific models, about 7 percent of the overall Automaker conversation A much larger percent of mentions are occurring on Blogs as it relates to Diesel, with 64% of posts happening on blogs Time of Day *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 Source: Radian6
- 58. MODELS SHARE OF VOICE INSIGHTS 30 DAY POPULARITY CHART (FEBRUARY 4 – MARCH 6, 2011) Day of Week 124,581 blogs posts, with major content topics being: •Commercial buzz •2011 Chicago Auto Show Time of Day *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 Source: Radian6
- 60. COMPETITORS 30 DAY SHARE OF VOICE (FEBRUARY 4 – MARCH 6, 2011) Day of Week Brand Y 20% AutoMaker 47% Brand X Brand W 14% Time of Day DIESEL SOV - AUTOMAKER *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - & COMPETITORS 3/7 Top Keywords Automaker Brand X Brand Y Brand Z Mentions 58,387 17,380 23,581 24,679 % of Total Volume 47% 14% 19% 20% Source: Radian6
- 61. AUTOMAKER - SEARCH UNIQUE VISITORS (JANUARY 2009 – JANUARY 2011) Day of Week Time of Day INSIGHTS : *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 The Automaker site traffic over the past two years has had an interesting history, between January 2009 – August 2009, unique visitors average nearly 811, 975 unique visits a month. But over the next year, the site traffic dropped dramatically , averaging only 662,780 unique visits a month. The site bottomed out completely in September 2010 when it recorded only 535,000 unique visits, lowest in two years. After this point, the site seemed to gain better traction and improved greatly, with its highest peak in two years with 912,000 unique visits recorded in January of 2011. Source: Compete.com
- 62. TOP REFERRAL SITES JANUARY 2011 TRAFFIC SUMMARY Percentage of Overall Source Referring Volume Traffic Direct Traffic 103,608 8.69% Google 314,306 26.35% Yahoo 166,070 13.92% Facebook 49,272 4.13% MSN 28,770 2.41% Bing 24,943 2.09% Ask.com 24,688 2.07% Autotrader 15,370 1.29% AOL 10,274 0.86% YouTube 9,052 0.76% TimeeBayDay of 6,396 0.54% INSIGHTS : Webcrawler 6,376 0.53% *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 Edmunds 5,852 0.49% The traffic summary for January 2011 had nearly 1.2 million site visits, with Honda.com 5,767 0.48% nearly 8 percent of the traffic coming from direct sources to Brand.com, and Cars.com 4,636 0.39% the remaining 92 percent coming from Toyota.com 4,459 0.37% referral traffic sources Audiusa.com 4,445 0.37% CNN.com 4,342 0.36% Source: Compete.com
- 63. TOP REFERRALS PAID VS. NATURAL SEARCH TRAFFIC (JANUARY 2009 – JANUARY 2011) Day of Week Time of Day *Sample data pull of all Twitter conversation mentioning @Automaker from 3/2 - 3/7 INSIGHTS : In January 2011, there were a total of 338,672 Search Referrals (Google, Yahoo!, Bing) and 35,970 Paid Search Referrals Source: Compete.com
- 64. TOP REFERRAL KEYWORDS NATURAL VS. PAID SEARCH KEYWORDS (JANUARY 2009 – JANUARY 2011) Top Organic Keywords Search Referral Share Top Paid Keywords Search Referral Share keyword 1 2.04% keyword 1 16.57% keyword 2 1.8% keyword 2 10.23% keyword 3 1.75% keyword 3 4.08% keyword 4 1.75% keyword 4 2.04% keyword 5 .88% keyword 5 1.80% keyword 6 .83% keyword 6 1.75% keyword 7 .71% keyword 7 1.58% keyword 8 .46% keyword 8 1.23% keyword 9 .43% keyword 9 0.92% keyword 10 .43% keyword 10 0.88% keyword 11 .37% keyword 11 0.83% keyword 12 .29% Time of Day keyword 12 0.71% keyword 13 .29% keyword 13 0.70% keyword 14 .25% 0.65% *Sample data14 of all Twitter conversation mentioning @Automaker from 3/2 - keyword pull 3/7 keyword 15 .22 keyword 15 0.60% keyword 16 .19% keyword 16 0.50% keyword 17 .18% keyword 17 0.46% keyword 18 .17% keyword 18 0.43% keyword 19 .17% keyword 19 0.43% Keyword 20 .16% keyword 20 0.40% Source: Compete.com
- 65. THANK YOU Copyright ©2011 Daniel J. Edelman, Inc. All rights reserved. All information contained herein is confidential and proprietary to Daniel J. Edelman, Inc. (―Edelman‖).
- 66. 6. Shaky Metrics 66
- 67. What is Sentiment? • From PR: A judgment about whether an author or document expresses a positive opinion about a person, place or thing. Sometimes called tone or opinion. • Usually expressed as positive, negative, or neutral • Sometimes expressed with polarity (strongly positive, strongly negative)
- 68. Challenges in Sentiment Implication vs. Information • ―Today Morgan Stanley upgraded GM from a sell to a hold.‖ • ―An earthquake of magnitude 8.9 has struck Japan off the island of Honshu.‖ Interpretation & Irony • ―The average creationist exhibits all the intellectual rigor as a modern day Neanderthal.‖ • ―Microsoft is about as cool as Austin in August.‖ Interpretation/Context • MUST WATCH: http://bit.ly/9nOU7 ROTFL #charliesheen #winning • ―I totally love Chris. I would kill myself kill myself if he left me for another man.‖ • ―Just read the book.‖ • ―I love Hulu. I just wish they didn’t have so many ads.‖
- 69. More Challenges in Sentiment Entity Clarification • ―I love Coke Zero, even though I hate Coca-Cola.‖ Relevance • ―sitting on the couch with Bruce watching burlesque on netflix‖ • ―This is amazing coke.‖ • ―ZICOZICOZICO!!!!!!!!!!‖ • ―I <3 Beeeeeiiiiiibbbbeeeerrrr!!!!!!!!!!!!!!!!!!!!!‖ • ―OMG you gotta watch this http://awe.sm/O7gb6T Slang • ―Totally digging Netflix streaming‖ • ―I <3 Beeeeeiiiiiibbbbeeeerrrr!!!!!!!!!!!!!!!!!!!!!‖
- 70. Sentiment: Understand Error Tolerance vs. Level of Effort Rate of Human Agreement on Sentiment: 80-95% • <80%: Mechanical Turk • 85%: College educated, same set of guidelines • 90%: PR companies using same set of guidelines • 95%: KD Payne & her crew Factors in Machine Learned Agreement with Humans • Cleanliness of content (POS, grammar, punctuation) • Completeness of vocabulary (Taboo) • Ability to look beyond the text words (punctuation, URLs) • Sophistication of the formula used (documents vs. statements) • Interpretation vs. information (Index Falling)
- 71. Sentiment Can Vary By Channel 71
- 72. Sentiment Analysis: Not Just Positive/ Negative/ Neutral
- 73. Partial Twitter is Good for Trend 73
- 74. But You Need To Know What You Don’t Have Twitter clients that use the search/track APIs don’t necessarily give you the full stream • Depends on popularity of term and frequency of polling • Rate limits from Twitter per vendor/ Distribution of polling across applications • Sometimes, Twitter clients don’t even have full @ data (though DMs mostly seem to arrive) • Search criteria are limited in length (140 characters) and functionality (embedded logic) or other issues (Ke$ha; language classification)
- 75. Twitter Partnership with Gnip Twitter has made full firehose data available by keyword search through GNIP. Approximate costs: • Volume or coverage based • URL expansion • $2K/ month for a box • $0.1/ 1000 tweet ($1K/ 10K Tweets) • API from which to pull data • Contact @robjohnson for more information
- 76. 7. A Number is Not a Decision 76
- 77. Social Intelligence… …the management and analysis of customer data from social sources, used to activate and recalibrate marketing and business programs. Source: Forrester Report ―Defining Social Intelligence‖ 77
- 78. Metrics Cannot Make Decisions What strategic actions can result from knowing… • Added 200 Facebook fans this month • Positive sentiment decreased by 5% this quarter • Twice as many Twitter mentions this week 78
- 79. 79
- 80. Go Beyond Metrics to Social Intelligence How are consumers reacting to the Verizon iPhone? 300,000 mentions of iPhone & Verizon 50% Off-Topic 27% - Phone Feedback 19% - Pro Verizon 14% - Pro AT&T 50% On-Topic % relevant conversation 1/11/11 – 1/17/11 80
- 81. Case Study: Major Insurer – Social Media Analysis of Advertising Effectiveness Overview: Insurer has spent millions running national ad campaigns and needed to understand effectiveness Key Findings / Insights of messages and drivers of consumer purchase intent. Both Insurer’s and Competitor’s campaigns generate healthy “buzz”, but Using Crimson Hexagon, Insurer opinions indicate that positive sentiment analyzed the nature of online doesn’t reflect intent to purchase conversations about its own and a insurance based on campaign key competitor’s ad campaigns: • The Insurer’s ad campaign is better What topics dominate the online received but does not generate conversation related to Insurer? conversation related to buying or Is the ad campaign driving intent to switching insurers purchase? • The Competitor’s ad campaign, while How does competitor stack up in more divisive, spurs a conversation similar analysis? trend related to intent to purchase What’s missing from current campaign messaging? 81
- 82. Insurer Ad Effectiveness Analysis Data reveals “Insurer’s‖ ad campaign is liked but didn’t generate conversation about buying or switching insurers ―Competitor‖ ad campaign, is more divisive, but spurs a conversation trend related to purchasing 82
- 83. Case Study: Medical Device for Diabetes & Insulin Overview: Customer planned to Key Findings & Recommendations introduce a new Insulin delivery device. Majority of conversation intertwines emotions with How can the online conversation inform experiential factors and improve the product marketing and • Impacts of diabetes and insulin usage are both launch messaging? medical and lifestyle related Insulin itself is often the issue Using Crimson Hexagon, the • Over one-quarter of the discussion shows customer analyzed the nature of online concerns about having diabetes which requires using insulin or about the efficacy of insulin itself conversation about diabetes and Addressing convenience factors presents a big insulin, specifically: opportunity • 24% of all conversation (and over 40% of How do patients discuss the frustrations voiced) focus on either the lifestyle complexities of treatment and complexities or problems with delivery devices control? (primarily needles) Opportunity for brands to get more engaged and What core messages will resonate become trusted-partner most with these consumers? • Consumers are looking for advice, recommendations, shared-support within the What is the significance for new online community product launches? 83
- 84. Medical Device Consumers’ Concerns Presented Opportunity to Provide Lifestyle, Community & Support Lifestyle and personal experiential factors jump out as leading categories of opinion. Comment on the actual devices is low. 84
- 85. 8. Integrated Program Design 85
- 86. Why Care about Metrics? To understand if the activities you undertake help you to achieve a goal or purpose.
- 87. A Tale of Two Facebooks 87
- 88. Finding 1: Improved Visit Loyalty Paid Search Natural Search Referring Sites Twitter E-mail Facebook 0 0.5 1 1.5 2 2.5 3 3.5 4 Visit Loyalty — how may repeat visits are generated in the following 30 days? Source: Rob Brosnan, Seton Hall University
- 89. Finding 2: Improved Conversion Efficiency Facebook Marketing Programs Natural Search Referring Sites Direct Load 0 0.5 1 1.5 2 2.5 3 Conversion Efficiency — how visits are converting visits? Source: Rob Brosnan, Seton Hall University
- 90. Finding 3: Facebook Delivers Local Visitors Source: Rob Brosnan, Seton Hall University
- 91. Thank you. We’ll see you at the closing parties! @blake @clightner @setlinger @margaretfrancis @megcos
Editor's Notes
- #88: Review: 3 objectivesBasic metrics & how to get themStart for designing your own measurement programNow what?What are you going to do, and how are you going to do it? Relationship you want with your communityBrand consistency/ equity buildingInnovationOrganic activity Other marketing programs Sell it in to the orgProgram design/ execution + metrics designSMM: hootsuite, cotweet, Facebook Insights, Sprinklr, Involvr, Vitrue, Buddy, WordPress, Llithium, Jive, Drupal, Webtrends,SOCIALIZE!!! PROMOTE!!!
- #89: Initially, I was skeptical that social media could deliver more than a something trendy. After all, what's the profit margin of a tweet. Yet, using click-through analysis from Coremetrics, we uncovered at three key findings about some social media sources. Our first finding was that Facebook and Twitter delivered visitors who made more than twice as many repeat visits during a 30 day period as visitors from search. With as long a purchase cycle as we have, repeat visits are critical.
- #90: The college enrollment process is essentially a long string of micro-conversions spread over several years. During the first half of 2010, Facebook provided for the most efficient source of conversions of any channel. And here we are specifically looking to Facebook; while Twitter provides a good source of inbound linking, it is (or at least was) not a strong performer with respect to conversion performance.
- #91: With more than 3,600 colleges and universities in America, the college market is highly regionalized. Channels that deliver local visitors are absolutely critical to future revenues. Using this heat map we were able to see that the location and engagement levels of Facebook visitors most closely matched our historical enrollment profile. Plotting these visitors by distance, we could see that Facebook was far better than any other channel at delivering local visitors.