Facebook catching up on MySpace

Facebook is on a trend to closing some of the gap with Myspace's traffic. MySpace made the open and distributed widget economy flourish but Facebook has managed to successfully position itself as a platform; both inviting the community to add apps incremental value. If Search is the Internet OS, Facebook is the middleware!

Even though MySpace has over twice as many visitors as Facebook (Compete, QuantCast ), Facebook is growing a lot faster. With 1 in 8 users worldwide visiting MySpace.com in any given month, the social network remains an incredible success story of adoption in an industry in its infancy. This said, MySpace could have reached critical mass and now most of its growth potential, getting closer to saturation with its current value proposition? Already 55% of online teens have profiles online according to PEW. Adding incremental unique visitors beyond a certain point becomes expensive for a traffic harder to convert and of poorer quality to advertisers. Engagement metrics look about the same although still in favor of MySpace. Compete shows more pages per visit for Myspace at 55 versus 44 for FaceBook but the gap is closing.

Traffic and Buzz converge. Nielsen BuzzMetrics Blogpulse shows blog references to "myspace" still outnumbers "facebook" although the gap is closing. IceRocket also confirms the trend with 2,007 posts including the term "facebook" over the past 3 months versus 6,413 for "myspace".

Queries can be split many ways of which isolating i) navigation vs informational vs transactional queries and ii) convenience-based vs preference are only two behavioral dimensions. Enough navigational queries take place in search boxes to also extract some directional insights from Google Trends. Google's search volume confirms Facebook closing the gap on MySpace as well.

Now, not all user generated opinion is good. Playing around with OpinMind, MySpace still has a much more positive "Sentimeter" than Facebook. This said, I would not conclude that fast that MySpace drives more satisfaction, engagement for that much. Different demographics makes them apples and oranges. The Sentimeter displays the relative number of positive and negative opinions identified by Opinmind's automated search processes ... just another data point. It would be interesting to see OpinMind's trend.

And now for more fun, I passed both through the sucks-rocks grounder to put FaceBook on top again.

Sphere: Related Content

Digital influencers will grow to 34.4 million in 2011 - eMarketer

Many Computational Linguistic departments across the country are busy developing a broad range of Semantic-based algorithms these days. Browsers came from the academic world, so did a lot of search technology, why not sentiment extraction.

Advertisers traditionally follow consumers. eMarketer just released a study mentioning that already 66 million adults regularly share advice about products and services with others, and 27 million are exerting that influence online. As more people become comfortable with voicing their opinions on the Internet, the number of digital influencers will grow to 34.4 million in 2011. There is a shift underway from few-to-many to many-to-many publishing models resulting in an explosion of consumer-generated media. 44% of Internet users are content creators (PEW), and an increasing ratio of search engine's top results are now user-authored content. Brand owners are losing share of voice and control over their message. Internet sentiment analysis, buzz monitoring and online reputation management could very well emerge as the next significant search marketing era after SEO and SEM.

SentiMetrix is one of these emerging startups I have been keeping in touch with over the past months. SentiMetrix is offering innovative technology framework to measure sentiments or opinions expressed in the electronic media, worldwide. It combines data gathering with named entity extraction and text analytics, to track opinions expressed about the topics requested by SentiMetrix clients. SentiMetrix platform is based on the award-winning Oasys technology (http://oasys.umiacs.umd.edu/oasys/) developed at the University of Maryland Institute for Advanced Computer Studies by Dr. V.S. Subrahmanian and his team. By combining statistical methods with natural language processing techniques, the software closely mimics the way humans perceive opinions expressed in electronic texts, be it news articles, blog posts or customer reviews.

Sphere: Related Content

Why I left AOL Search

I started at AOL Search almost 3 years ago, hoping to be part of the turn around story, taking AOL to the Web and integrating search consequently. ... I was also thinking no one would single-handedly hold responsible if AOL missed a step.

I spent most of the past decade planning and building search experiences from developing the first SafeSearch application in 1994 with InterGO to crawling and parsing applications with Microsoft in 1996, then Altavista, Infospace and AOL.

I was realistic. After IAC acquired Ask, I knew there was no room for AOL Search to rank among the top 3 players in the traditional search category. Personal technology categories, from desktop operating systems to browsers and productivity apps have traditionally had enough room for two leaders and a third rotating player. I also knew the search ecosystem was heading for fragmentation. Crawling, indexing, computational linguistic and managing consumer media experiences definitely don’t require the same core competencies. Given the rise of social networking, broadband penetration, the explosion of consumer generated content and the emergence of awesome technologies re-inventing user interaction models. I also knew that AOL on the Web still had a unique opportunity to reclaim its historical online community fame and core competency.

I thought AOL had a unique opportunity to repositions itself as a Discovery Engine, taking the lead on becoming the leader of an emerging new social search category. FullView was the base for exactly that. Social search is This is just what my team focused on after integrating Search capabilities and optimizing revenue across the Network including the AOL portals, Email apps, MapQuest, Netscape and more to reach about 110 million UVs.

In May, the new AOL management team - probably rightfully when considering corporate strategic goals - decided to copy and paste the Google experience to gain some level of business parity. I always knew that was a possibility while evangelizing social search as a differentiation factor and stuck around to the end. Let Semantic Search technology, Social Search, and Sentiment analysis strive.

Sphere: Related Content

Umbria’s CEO Howard Kaushansky on Online Market Intelligence, from Web2Expo

Very interesting presentation from Umbria's CEO Howard Kaushansky - Blogs, Chat Rooms, Wikis, Oh My! The Yellow Brick Road to Online Market Intelligence - at Web 2.0 in San Francisco earlier this week. Surprisingly packed room for a Monday morning with many in the audience traveling from East coast time. Below are my – raw, incomplete, and slightly editorialized – notes from Howard’s presentation. Umbria is a marketing intelligence company that mines the blogosphere and other public forums for real-time insights into companies, products, people, and issues. Including some interesting work for CNN, looking at the buzz, the overall opinion ... Alberto Gonzales, for example, from March 14-20, 2007.

My notes …

Word of Mouth media revolve around Blogs, Opinion sites, Message boards, Forums, Chat rooms; somewhat a parallel world to old traditional sources including Panel data, Industry pubs, Surveys, Custom research, Syndicated and Business pubs.

• 42% of broadband users post content – PEW Broadband usage 2006
• 71 million blogs, doubling every 6 months – Technorati
• Blogs becoming part of mainstream media
• 40% of Internet users in the US read blogs; 10% regularly post
• 100,000 blog posts a day, one per second
• Since September 2004, the Blogosphere has increased over 15 fold – Technorati
• 75% of people don’t believe marketing anymore
• 92% trust word of mouth for product decisions

Howard went over some interesting examples and sentiment extraction challenges, including the now legendary Chevy Tahoe ad, “coke” versus “coke”, Topics and sub-topics, Trending of sentiment, Demographic segmentation, Filtering spam blogs. And why we should care about Internet sentiment analysis:

• Is marketing having an impact?
• With what types of customers?
• What themes are resonating?
• How does the competition stack up?
• Where are they strong/weak?
• What features do people like?
• What irritates people?

Monitoring online buzz is fun and entertaining. For brand managers and others who care, sentiment analysis is also the basis to reputation management. Howard went over some Word-of-mouth strategies to be considered:

Encouraging communications

• Tools to make telling a friend easier
• Creating forums and feedback tools
• Working with social networks

Giving people something to talk about

• Information that can be shared or forwarded
• Stunts, advertising that encourages conversation
• Building WOM-worthy elements into products

Creating communities and connecting people

• Creating user groups and fan clubs
• Supporting independent groups that form around your product
• Enabling grassroots organizations such as a local meeting and other real world participation

Working with influential communities

• Providing recognition and tools to active advocates
• Recruiting new advocates, teaching them about the benefits of your products, and encouraging them to talk about them

Creating evangelist or advocate programs

• Tracking online and offline conversations by supporters, detractors, and neutrals
• Listening and responding to both positive and negative conversations

Researching and listening to customer feedback

• Encouraging two-way conversation with interested parties
• Creating blogs and other tools to share information

Engaging in transparent conversation

• Co-creation and information sharing

Sphere: Related Content

Internet buzz and The Oscars

The oscars and nominated directors, actresses, actors and everybody else generate quiet a bit of buzz online. What about each of The Oscars' nominated Best Pictures? Can Internet Sentiment Analysis predict the winners? Although just emerging, the explosion of consumer generated media makes online buzz monitoring an increasingly viable tool to take the pulse of attitudes in social media towards pretty much anything, brands, products, ideas, people and more. Internet sentiment analysis draws on information retrieval, data mining, machine learning, statistics, and computational linguistics to transform unstructured online dialog (blogs, chats, boards, ...) into marketing and social media insights. Search is the Internet OS!

The following charts present blog coverage, good, neutral, bad, authoritative and not, planted and spontaneous conversations. The first two charts come from BuzzMetrics' Blogpulse.


The data is as good as what you're looking for, so for transparency purpose, here are the exact queries I used: The Queen >> the queen movies, Little Miss Sunshine >> little miss sunshine movies, The Departed >> the departed movies, Babel >> babel movies, Letters from Iwo Jima >> letters from iwo jima movies. I added the "movies" filter to reduce the noise around generic titles like The Queen or Babel.



The next two charts come from IceRocket. Over the past 3 months, IceRocket reports that The Queen received about 14,578 posts, an average of 162 posts a day. Then comes Little Miss Sunshine with 9,008 posts, 100 a day. The Departed follows with about 8,021 posts, 89 a day and Babel received about 5,545 posts, 62 a day. Letters from Iwo Jima follows with 3,159 posts, 35 a day.



Beside the general public's attitude, Internet sentiment analysis also informs on how effective the studios have been at executing online marketing and viral word-of-mouth campaigns. With marketing budgets getting bigger than production budgets, the studios are no stranger to online marketing.



The next charts come from Technorati, querying their index for "any blog" in "all languages" for the past 360 days, using the same queries.

Posts that contain The Queen per day for the last 30 days.



Posts that contain Little Miss Sunshine per day for the last 30 days.



Posts that contain The Departed per day for the last 30 days.



Posts that contain Babel per day for the last 30 days.



Posts that contain Letters from Iwo Jima per day for the last 30 days.



Mining user generated media is one way to look at it. The number of searches for each of these movies is another way to look at it. The following chart comes from Google Trends, representing the search volume for each movie title. If you start digging into the coverage, you'll see that there is a lot of noise, not all of these searches actually reflect the buzz specific to these movies.



This is really all for fun and entertainment - like The Oscars - because i) subtle changes in the query formulation can dramatically change the counts, ii) there is a lot of "linguistic-related noise" out there that has nothing to do with the movies, iii) this is just coverage, who knows, maybe it's all negative coverage ... not :-) The serious analysis comes in when you start looking at the tone, positive, neutral, negative, comment categories, authority level, momentum, conversational threads, correlation to marketing spend and a whole lot more. And who knows, maybe the wisdom of blog crowds will be right tonight.

Sphere: Related Content

Internet buzz about Sproose social search engine

Sproose has been getting a bit more press coverage in the past few days. Sproose is now reporting over one million voted Web sites. Nice. Not everybody needs to be submitting content, voting and tagging for collaborative filtering and directory building to work. Sproose added video Search earlier this month from Blinkx and News from Moreover last month. Sproose taxes itself as an interactive search engines, providing peer-moderated, ranking, prioritizing and community networking for consumer use. Search and Vote, basically.

So, what's the buzz about Sproose out there?

Nielsen BuzzMetrics' Blogpulse shows a coverage spike at the very end of 2006 when the engine was first introduced. And so does Technorati. Reporting consistency is good. And finally, so does IceRocket, showing about 1.28 posts a day in average over the past 3 months, for a total of 115 citations.




Sproose is one of the most comprehensive social search engines out there. Sproose' s got Search for text, video, news, and popular tags, tagging, discussion, cloud navigation, bookmarking, vote and remove this site, collaborative ranking, messaging and profiles. I wish I could find out how to submit content such as bookmarks, Web pages, pictures and so forth. From anecdotal testing, relevance is just ok, though. Monetization seems to come from Kontera's ContentLink double-underlined terms ... not a big fan of it, although Sproose is obviously no philanthropic organization.

Go Sproose!

Sphere: Related Content