Social data is usually big data, because we’re talking about masses of data collected from social media. However it’s not big data that makes social analytics distinctive, rather it is the ability to derive required insights from such unstructured data. In short: it’s not just about the amount of data you have. It’s about understanding what that data is trying to tell you – and revealing non-obvious factors which influence that understanding.

Social data is generated by people

With the Internet we have moved from connecting systems to connecting people, and people have created data which we call social data. Unlike machine-to-machine data social data is not structured and systemised data with strict protocols, but rather is unstructured and contains the entire range of human language communication nuances.

Social data analytics is the processing of social data with an objective in mind. The objectives could be predictive – predicting box office success – or to gain insights around public sentiment about an issue. Because the range of objectives is so broad social data analytics often requires the connection, or mashup, of several analytics tools or systems.

What is social data analytics?

All social data analytics comprises three core functions:

1. Firstly, the collection of data generated from social networking sites (or through social applications);

2. Secondly, sophisticated analysis of that data, in many cases requiring real-time (or near real-time) data analytics, and including measurements which understand and appropriately weigh factors such as influence, reach, and relevancy, and an understanding of the context of the data being analysed, including time horizon considerations; and lastly,

3. Reporting of the findings in ways which best inform the objectives, and which may have to be used programmatically to drive further business actions.

In short, social data analytics involves the analysis of social media in order to understand and surface insights which are embedded within the data. These insights include insights not only about data and potential outcomes, but about people themselves. That’s a dramatic new power derived from social data – personality insights.

Now, although we’ve said that social data is data generated by people an essential component of social data is the associated meta-data generated by the social media systems e.g. Facebook, from where the data has been extracted. The analysis systems also have to make best use of this meta-data (for example time/data data and location data, or from Facebook the Social Graph which links you to your friends and interests).

The key characteristics of social data analysis

The fundamental characteristic of insightful social data analysis is that it has to take into account more than just content, and more than just transaction analytics i.e. “Likes Follows and Number of Comments”. It is not just what the Community Manager sees in the typical back-end analytics of Facebook. It has to combine content, context and sentiment.

The latter element – sentiment – has proven to be the most elusive component to nail down. This is why IBM with IBM Watson and their text analytics background have a powerful edge in social data analytics – because Watson is good at understanding the meaning of what is being said in social and elsewhere. You could say that what distinguishes social data analytics from sentiment analysis is the depth of analysis and that’s where Watson makes its mark.

Other characteristics include:

• Attention to time horizons – sometimes windows of opportunity are significantly limited in the field of social networking. What’s relevant one day (or even one hour) may not be the next. Hence being able to quickly execute and analyse real-time data streams can be very important, and obviously in the case of a social media crisis this is a critical feature. Current systems tend to be simply content identification systems requiring a high degree of human interaction and interpretation. Future systems will provide for insights to drive intelligent programmatic responses.

• Influencer analysis – understanding the potential impact of specific individuals can be key in understanding how messages might be resonating. It’s not just about quantity, it’s also very much about quality. To date, influencer analysis as a component of social data analysis has been relatively unsophisticated and often not analysed in the context of the topic but simply by a raw influencer metric. This is about to change and the next generation of social data analytics will be much more accurate in isolating the real influencers about topics and events, using such techniques as semantic enrichment.

• Network analysis – social data is also interesting in that it migrates, grows (or dies) based on how it resonates with people in a network and in their shared networks. And how people in one social media site might react to it can be very different to how people in another social media site react to the same content.  Understanding more about these network effects might give insights into how to create “more viral” content. But in fact the real insights from network analysis is to know how the network works as a “living organic system” in relation to the desired objectives, and in relation to all the preceding characteristics.

We are in the early days of social data analytics, and like other technologies in their early days the hype can overtake the reality. But with the right tools, realistic expectations, an appreciation of the limitations and of the need for experimentation, we are going to see social data play an ever-increasing role in how business is conducted.

In fact most companies understand that social media may have important roles to play in how they conduct business… the challenge is that most are unaware of how to go about creating a systematic approach to social data analytics. It’s not a job for the social media intern.