For the last seven years, I have worked closely with graduate students and more than 90 organizations on over 100 different projects. When we first started, the projects were relatively straightforward – What is the value of a follower? Can you help us understand the ROI associated with our social media strategy?

Lately, however, these projects have become much more nuanced, combining social data with other data types to develop multi-varied projects that tell detailed stories about how information gets exchanged from one community to another, insights on perception and consumer behavior, and how influence gets brokered across social channels and permeates other forms of media.

We have seen distinct phases associated with social media analysis:

Phase 1: Simple counting - How many followers do I have? What is the change in followers from one period to another? What does it mean?

Phase 2: Combining measures into indicators and association to different types of constituents – Share of Voice (How does my brand compare in terms of overall social mentions to my peers?), PTAT – People Talking about That – How do we correlate a social thematic string to a particular online community?

Phase 3: Taking these measures and evaluating themes across different dimensions – Time series analysis, channel differentiation, content type, etc.

Phase 4: Adding in multivariate analysis and moving from descriptive to predictive analysis. Regression analysis looking at relationships to determine correlation between a numbers of variables.

Phase 5: Integration of social data with other collected and stored data. In marketing/sales, this might mean combining social data with CRM systems. In healthcare, it might involve combining patient curated data on condition assessment with Electronic Medical Record (EMR) information to get a holistic picture of the patient’s health and wellness. This is possible and highly beneficial but several barriers exist that make this difficult to achieve.

Phase 6: Prescriptive analytics and Social Marketing Automation: Now that we have years of social data to analyze, mechanisms to normalize it with other data forms, and have applied advanced statistical relationships to model the data, we can assess benefit and automate repetitive processes.

Our earliest projects involved simple means of measurement – Share of Voice before, during or after a campaign, Increase in Buzz (mentions) both positive and negative sentiment associated with a timeline, thematic strings that can be identified, collected, measured and used to engage with different constituents. These would largely fall into descriptive analytics and might involve some regression analysis to find the relationship between one set of inputs and an observed impact.

Recent projects have been using social data to identify internal communication patterns that can be mapped to a social network analysis visualization to show how communication spreads from one community to another. This might consider account popularity vs. influence, bridging point between disparate communities and information distortion.

Other projects look for insight on how to engage different consumer communities at different parts of the customer journey. Does social have a greater impact on the awareness or consideration phase? If so, how do you design social campaigns appropriately (phrasing, links, graphics) to ensure optimal level of effectiveness?

Another project might use pattern recognition algorithms to identify event detection and event horizon using social data as a means to assess a company’s reaction to secondary impacts associated with a natural disaster.

These projects don’t view unstructured social data as an oddity that needs to be assessed, categorized and rationalized associated with its potential value, but rather as an integral part of data driven decision making that can be integrated into many different functions (sales, marketing, product development) necessary to build a durable competitive advantage.

The challenge that remains is providing the fabric to integrate social data into an organization’s overall data strategy. For example, a pharmaceutical company might value social data to provide insights during the clinical development phase but is unequipped to use it as means to determine sales enablement.

Social data as a means of descriptive analytics is well known using social data in conjunction with other data to provide prescriptive analytics (Phase 6) is still in the nascent stage. It has been demonstrated, but the availability to normalize it and act upon it is still in the early stages.

Organizations today are asking interesting questions that will take a great deal of analysis and integration with other data.

For example, is there a correlation between a user’s social behavior and their credit risk? Is there a relationship between consumer values and their propensity to watch and discuss a specific show? What percentage of a specific community would watch the program when it airs or binge watch and how does one community influences another?

These are challenging questions, but they can only be answered if a company is willing to design adequate experiments to test some of these hypothesis, and that brings up a whole different set of challenges.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access

Ari Lightman

Ari Lightman

Ari Lightman is a professor of digital media and marketing at the Heinz College of Information Systems and Public Policy at Carnegie Mellon University.