What about Analytics in Social Media monitoring?

I was speaking to a client the other day.  This company was very excited about tracking its brand using one of the many listening posts out on the market.  As I sat listening to him, I couldn’t help but think that a) it was nice that his company could get its feet wet in social media monitoring using a tool like this and b) that they might be getting a false sense of security because the reality is that these social media tracking tools provide a fairly rudimentary analysis about brand/product mentions, sentiment, and influencers.  For those of you not familiar with listening posts here’s a quick primer.

Listening Post Primer

Listening posts monitor the “chatter” that is occurring on the Internet in blogs, message boards, tweets, etc.  They basically:

  • Aggregate content from across many,  many Internet sources.
  • Track the number of mentions of a topic (brand or some other term) over time and source of mention.
  • Provide users with positive or negative sentiment associated with topic (often you can’t change this, if it is incorrect).
  • Provide some sort of Influencer information.
  • Possibly provide a word cloud that lets you know what other words are associated with your topic.
  • Provide you with the ability to look at the content associated with your topic.

They typically charge by the topic.  Since these listening posts mostly use a search paradigm (with ways to aggregate words into a search topic) they don’t really allow  you to “discover” any information or insight that you may not have been aware of unless you happen to stumble across it while reading posts or put a lot of time into manually mining this information.  Some services allow the user to draw on historical data.  There are more than 100 listening posts on the market.

I certainly don’t want to minimize what these providers are offering.  Organizations that are just starting out analyzing social media will certainly derive huge benefit from these services.  Many are also quite easy to use and the price point is reasonable. My point is that there is more that can be done to derive more useful insight from social media.  More advanced systems typically make use of text analytics software.   Text analytics utilizes techniques that originate in computational linguistics, statistics, and other computer science disciplines to actually analyze the unstructured text.

Adding Text Analytics to the Mix

Although still in the early phases, social media monitoring is moving to social media analysis and understanding as text analytics vendors apply their technology to this problem.  The space is heating up as evidenced by these three recent announcements:

  • Attensity buys Biz 360. The other week, Attensity announced its intention to purchase Biz360, a leading listening post. In April, 2009, Attensity combined with two European companies that focus on semantic business applications to form Attensity Group (was formerly Attensity Corporation).  Attensity has sophisticated technology which makes use of “exhaustive extraction” techniques (as well as nine other techniques) to analyze unstructured data. Its flagship technology automatically extracts facts from parsed text (who did what to whom, when, where, under what conditions) and organizes this information.  With the addition of Biz360 and its earlier acquisitions, the Biz360 listening post will feed all Attensity products.  Additionally, the  Biz360 SaaS platform will be expanded to include deeper semantic capabilities for analysis, sentiment, response and knowledge management utilizing Attensity IP.  This service will be called Attensity 360.  The service will provide listening and deep analysis capabilities.  On top of this, extracted knowledge will be automatically routed to the group in the enterprise that needs the information.  For example, legal insights  about people, places, events, topics, and sentiment will be automatically routed to legal, customer service insights to customer service, and so on. These groups can then act on the information.  Attensity refers to this as the “open enterprise.” The idea is an end-to-end listen-analyze-respond-act process for enterprises to act on the insight they can get from the solution.
  • SAS announces its social media analytics software. SAS purchased text analytics vendor Teragram last year.  In April, SAS announced SAS® Social Media Analytics which, “Analyzes online conversations to drive insight, improve customer interaction, and drive performance.”  The product provides deep unstructured data analysis capabilities around both internal and external sources of information (it has partnerships with external content aggregators, if needed) for brand, media, PR, and customer related information.  SAS has then coupled with this the ability to perform advanced analytics such as predictive forecasting and correlation on this unstructured data.  For example, the SAS product enables companies to forecast number of mentions, given a history of mentions, or to understand whether sentiment during a certain time period was more negative, say than a previous time period.  It also enables users to analyze sentiment at a granular level and to change sentiment (and learn from this), if it is not correct.  It can deal with sentiment in 13 languages and supports 30 languages.
  • Newer social media analysis services such as NetBase are announced. NetBase is currently in limited release of its first consumer insight discovery product called ConsumerBase.  It has eight  patents pending around its deep parsing  and semantic modeling technology.  It combines deep analytics with a content aggregation service and a reporting capability.  The product provides analysis around likes/dislikes, emotions, reasons why, and behaviors.  For example, whereas a listening post might interpret the sentence, “Listerine kills germs because it hurts” as either a negative or neutral statement, the NetBase technology uses a semantic data model to understand not only that this is a positive statement, but also the reason it is positive.

Each of these products and services are slightly different.  For example, Attensity’s approach is to listen, analyze, relate (it to the business), and act (route, respond, reuse) which it calls its LARA methodology.   The SAS solution is part of its broader three Is strategy: Insight- Interaction- Improve.  NetBase is looking to provide an end to end service that helps companies to understand the reason around emotions, behaviors, likes and dislikes.   And, these are not the only game in town. Other social media analysis services announced in the last year (or earlier) include those from other text analytics vendors such as IBM, Clarabridge, and Lexalytics. And, to be fair, some of the listening posts are beginning to put this capability into their services.

This market is still in its early adoption phase, as companies try to put plans together around social media, including utilizing it for their own marketing purposes as well as analyzing it for reasons including and beyond marketing. It will be extremely important for users to determine what their needs and price points are and plan accordingly.

10 thoughts on “What about Analytics in Social Media monitoring?”

  1. It is certainly a burgeoning space and one that I’m excited to be working in with NetBase. There is so much conversational information being created at the moment the ability to understand language at scale is a huge competitive advantage for any organization that has a vocal customer base. As the cluetrain said in 1999 “markets are conversations”, I hope our tools can help companies actually start to understand what is being said, what is motivating it, what the emotion is behind it, and then start leading the conversation in their marketplace.

    Anyway, thanks for the great writeup, if anyone wants to keep up with us please check out our blog http://netbasse.com/blog/ and on twitter at http://twitter.com/net_base

  2. Agreed that a company’s needs play a major part in what kind of social media monitoring/analytics services they should use. At Sysomos, we’ve taken the approach that one size doesn’t fit all. We offer two services: Heartbeat for people that want monitoring and measurement, and MAP, an analytics and reporting service with a rich feature set. In many cases, companies will start with Heartbeat to get their feet wet before migrating to MAP.

    cheers, Mark

    Mark Evans
    Director of Communications
    Sysomos Inc.

  3. Thanks Fern for providing this detailed description. As you mentioned the market is in its early stages and conversations as such are very helpful to people trying to make a decision on utilization of such tools.

    I do agree with you about the need for a more detailed analytic engine due to nature of social media conversations being raw and unstructured. Not to mention all the spam and ads thrown in the mix. Here in Collective Intellect Inc. our primary objective is how to extract relevant meaning out of text through creation of an advanced semantic engine helping many of our customers understand truly what social media is humming about their topics of interest. We have been at it for the past five years making us one of the oldest companies in this rapidly growing and dynamic space.

    Collective Intelelct is able to help its customers not only by tracking the activity level and sentiment on their topics for up to past 12 months, but also through advanced metrics such as intentions and dimensions. Examples of intentions include intentions to view and to purchase; and for dimensions concepts such as quality, customer service, problem, etc. Thrown in the mix are demographics and psychographics providing the ability to do segmentation of their audience at various levels.

    Both your blog post and my comment are examples of social media conversations touching on various topics, thus the need for a semantic analysis at snippet level. We have experienced very accurate categorization and meaning extraction once such a micro analysis is performed by breaking down a post into a series of snippets with each snippet consisting of the text surrounding their topic of interest.

    I hope the information above allows your readers to consider another alternative to mentioned solutions and how we have decided to extract meaning out of this vast social media space. For more information please don’t hesitate to refer yourself to our site or blog at:


    Sincerely Yours
    Mehrshad Setayesh

  4. I’m hesitant to offer what may be considered a product pitch, so let me first clarify that I’m specifically responding to “…they might be getting a false sense of security because the reality is that these social media tracking tools provide a fairly rudimentary analysis about brand/product mentions, sentiment, and influencers.”
    I can not speak for other tools. I know that NextStage’s Sentiment Analysis tool (http://nssa.nextstagevolution.com) has what users consider unprecedented accuracy (based on tweets, posts and 3rd party testing. A version of this tool is documented as accurately predicting political outcomes in the US and Canada as much as three months in advance) and a very sophisticated level of analysis (see http://www.bizmediascience.com/2009/06/sentiment_analysis_anyone_part.html to learn about this particular tool’s evolution. In a nutshell, we asked users what they wanted a “sentiment analysis” tool to do rather than telling them “this is what sentiment analysis tools do”). Our tools have long been in use (we can document tool use beginning in 2001-3, long before sentiment analysis was called “sentiment analysis”) and recently announced the Sentiment Analysis Voices tool (based on Syncapse VP Chris Berry’s suggestions). The Voices tool takes anywhere from 100 to 10,000 URLs and can analyze them en toto, individually or by the “voices” of individuals contributing to the thread (what I believe you’re referring to as a “listening post”).
    Last item (and what I’m willing to recognize as a pitch), NextStage Sentiment Analysis’ pricing is low enough that individuals are using it to analyze their own online writing prior to publication.
    Joseph Carrabis, CRO&Founder, NextStage Evolution

  5. The main issue is whether customers of these services keep them in mind as mere tools, rather than seeing them as solutions to what is typically an organizational problem as much as a technical one, not to mention the fact that language use is not amenable to predictive analytics with nearly the confidence necessary for decision making.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s