When I started out with my colleagues at Intelliseek's Applied Research Center, Social Media Analysis was in its infancy. Intelliseek had in some sense discovered it via exploring meta search applications and the value of indexing the deep web. As Intelliseek further explored and established the market we also made a clear decision to follow a data driven path. Content is the oxygen of the space, and we approached it with an all or nothing attitude. With an approach that grew from this data rich platform, the natural path to a scalable solution involves the application of a number of automated elements not just at the data acquisition stages but importantly at the analytical layer.
On the business side, the first age of social media analysis was very much an educational one. Helping customers understand firstly what social media was and secondly why it was of high value had to be a key focus. As we worked towards these goals, the space started to fill with a number of different types of competitors including those which followed a similar path to ours and those which took the consulting approach. Regardless of the path taken, educating customers about the space remained the most important aspect of our combined efforts.
We have now moved out of the education phase and are entering the second age of social media analysis. The key focus of this period will be methodology: how are results produced, how does an approach guarantee success and how can they be understood and relied upon?
Towards the end of the first age, customers could appreciate nuggets of information that they received - they were likely to be precise. As we get in to the second age, a transparent methodology is going to be key. This is an issue that applies not just to enterprise facing systems but also to consumer facing services. For example, while we may enjoy the results of TechMeme and Digg, and while they often surface interesting material, we have no idea what the bias in the system is. Systems like that put the user in the position of reading a newspaper with no reference as to the leanings or agenda of that publication.
On the enterprise side of the fence, methodology is going to be a key component of establishing both value and trust. The competitive landscape is already starting to indicate the importance of this issue with the emergence of services which propose not to deliver the broad range of services of the established leaders, but rather focus on specific elements of service such as influence measurement or sentiment analysis.
But what is a methodology? Superficially, a methodology is simply a series of steps that get something done. However, in order to establish a methodology, one needs to go deeper. I believe that the following pillars are where all the interest lie:
- Models: a model of the world, to some degree, is required to capture how events out there translate in to events in the data. People buy stuff, the use it and write about it - that is the type of model that enterprises are interested in. Secondly, data models are needed to clearly describe the data that is being used within the system. Data models are also vital to guiding the scope of content acquisition systems.
- Theory: how do we explain what influence is? How does a viral video spread? How can we characterise the topology of social networks in a meaningful way? What is sentiment and how do observable features (expression) relate to the true distribution of opinions? These are all issue of theory without which one cannot answer the questions that enterprises have about monitoring social media.
- Objectives: what exactly is the methodology intended to deliver? A comparison of buzz between X and Y? A measurement of sentiment over time? By capturing the objectives of the operation, the methodology can be developed via an understanding of the models and theory of social media. In addition, understanding the business objectives that the analytics will feed is vital.
Recent M&A activity has certainly helped validate the social media analytics space. These are, in some sense, the external, business indicators of the second age. Under the hood, the issues outlined above are going to be the real issues to watch.