Building Blocks of Social Media Analysis


Following the announcement of the Lexalytics merger, I wrote that the availability of off-the-shelf text analytics and custom software development suggested some questions about technology strategy for social media analysis vendors. The main question is an old one in software: Build or buy? The follow-up flows from a value chain analysis, and it sets up a series of decision points for the build/buy decision. It turns out that a social media analysis platform is built from a few, basic building blocks, and a decision to build a custom system doesn't imply a decision to build every part.

Three basic building blocks
The technology behind monitoring and analyzing social media content fits into three basic categories: content sourcing, analytics, and the software application that provides end-user features, such as dashboards, reports, and alerts. Many service providers add value beyond what is baked into their software, of course, but these are the basic technology building blocks. What makes them interesting is that each generates a potential product that can be offered as the "buy" option to other companies.

  • Content sourcing
    The first step in analyzing social media content, sourcing includes source discovery, content aggregation, filtering and metadata tagging. The result is a feed of relevant content data, with duplicates, spam and other noise removed. Feeds of traditional media sources have been available for years; syndicated content sourcing is available for social media types now. (See a list of social media aggregators)

  • Analytics
    The next step is analysis of the content, also known as text analytics. The system extracts key words, concepts and sentiment from individual messages; depending on the vendor, it may also evaluate messages and sources for influence, demographics, location, audience size, and more. This is the hard part, which inspires the question of whether computer analysis is good enough.

    Perhaps assuming that text analytics must be homegrown, many vendors skip the analytics entirely, using human analysts or leaving metrics out of their product. But text analytics has its own build or buy decision. The technology is available from software companies in the business intelligence market, and at least two companies focusing on social media, Lexalytics and Leximancer. (See a list of text analytics options)

  • Application
    Content sourcing and analytics constitute checklist features in a social media analysis platform: "Sentiment? Check!" They represent a lot of work, but it's hard to tell whose works better. The application software, on the other hand, is what clients see, and it provides many opportunities for evaluation. Is the user interface dated and clunky or sleek and current? How is content presented? What can users do with it? How easy is it to explore the data or create reports?

    The once clear line between dashboards and social media analysis platforms for workgroups has blurred over the past year. Outside sofware development services will tend to increase the capabilities of dashboards from less traditionally technical vendors.

Room for differention
Syndicated and licensed components don't lead directly to commmoditization, but the threat is there for vendors who fall behind in capabilities. There's still room for differentiation, especially in application features, not to mention the research and consulting services many vendors offer in addition to their systems. But separating the build or buy decision into at least three discrete decisions does have effects:

  • Lower barriers to entry. New entrants—startups or established players in adjacent markets—can launch social media analysis tools without building everything from the ground up.

  • Higher baseline expectations. Dashboards from companies without strong technology traditions will soon include advanced technologies. Basic monitoring dashboards are becoming commodities (witness the $5/month entry pricing).

  • Fewer reinvented wheels. Vendors can compete on their strengths in content sourcing, analytics or application development. External sources remove the need to develop their own capabilities in areas outside of their core strengths.
This should be good news for customers: the tools continue to improve, and vendors will compete on what really adds value. For vendors, the question is back to basics: which parts of your system are really special, and which might be better outsourced? It's a question to reconsider from time to time as the market for building blocks matures.

More posts in the "Build or Buy?" series:


Clean, concise and spot on

That's one of the cleaner and simpler descriptions of the various pieces of media analysis. It's nice to see somebody brining things down to the basics. This is a great first read for anybody looking for one of these solutions, so that they can read between the lines when talking to vendors.

Thanks, Jeffrey. I've had a lot of practice summarizing this topic over the last couple of years. :-)

Excellent summary.

One thing that strikes me so is that companies engage in sophisticated "monitoring", "sentiment analysis" effort at the time where 100's to 1000's of their employees blog, build social networks and engage in conversations that often are work related/expertise related.

Social effort is by nature distributed and the "sentiment" cannot be captured by just assessing whether post mentionning brand names and products "sounded" positive or negative. If an HP engineer is bringing value in conversations I am participating in, it's a plus for HP, no matter whether there is any comment (positive or negative) that can be connected to the HP brand.

Also this is dynamic and more complex that "counting mentions".

Let's take an example:

Imagine you stumble upon a post that describes the very bad experience J Jarvis had with Dell a few years ago. (yes it's that long)
The very fact that Dell was able to please J Jarvis at the end of the conversation should zero the negative sentiment of this post . Or should it not?

Depending on whether you assume the reader will be able to find the "happy end" on J Jarvis site or not, you may zero it or not, and even put it positive ? ( after all being able to recover from this horror story is better than doing nothing).

Frankly I think the emphasis on monitoring is an excuse not to spend the time in building a real strategy around social media marketing.

To me, monitoring should come at the end ( and most of the case what is really important to monitor is impact and results.

Thanks, Dominic. My inclination is that there's a lot of value in online discussions, and monitoring is just one important activity. Company questions about which strategies to try should include the word "and" a lot. Monitoring and measurement. Customer service and market research. PR and Marketing, HR and Finance. We're just getting started.

Glad to see this coverage Nathan. Thanks.

We're seeing that is a "building block" that fits in a number of ways... to not only monitor, but to gain insights that allow a company to act/interact with social media and other online sources.

And in support of your general point, on it's own analysis lacks the value that a broader customer strategy needs, so we're letting firms with software or services that support the broader process of managing customer experience or social media management license the platform so that the analysis fits as a building block in their solutions.

You must be able to not only know what people are saying, but WHY. Just counting text concepts doesn't cut it. We call this pathway analysis, and it seems to resonate very well with our customers.

If you or one of our readers would like to see the system work, just let me know and we'll run an analysis and share it...


Comments are now closed for this entry.

About Nathan Gilliatt

  • ng.jpg
  • Voracious learner and explorer. Analyst tracking technologies and markets in intelligence, analytics and social media. Advisor to buyers, sellers and investors. Writing my next book.
  • Principal, Social Target
  • Profile
  • Highlights from the archive


Monthly Archives