Tips on Social Media Analysis

The process of social media analysis involves essentially three steps: data identification, data analysis, and finally information interpr...

The process of social media analysis involves essentially three steps: data identification, data analysis, and finally information interpretation. In explaining each of these steps, we provide important insights and techniques that can be used to maximize the value derived at every point during the process. The approach we take is to first define a question to be answered (such as “What is the public’s perception of our company in the light of a natural disaster?”). In attempting to analyze these questions, we suggest that analysts think like detectives, always asking the important questions “Who? What? Where? When? Why? and How?” These questions help in determining the proper data sources to evaluate, which can greatly affect the type of analysis that can be performed.

Data Identification

Any social media investigation is only as good as the data in which you are searching. The first part of this book explores proper data identification—or where to look in this vast social media space. In searching for answers, keep in mind that we will be searching through massive amounts of unstructured data, all in an attempt to make sense out of what we find in the process. Once we uncover some interesting artifacts, we will be transforming them into (hopefully useful) information. In the long run, the ultimate social media business value objective is to derive real business insight from this data, turning the information we’ve gleaned from these sources into actionable knowledge.

To ensure that what we are collecting is the proper data or it explores the correct conversations, we look into questions such as these:

• Whose opinions or thoughts are we interested in?

• Where are the conversations about the topic in question happening?

• Do we need to look at the question back in time or just current discussions?

Data Analysis 

We explore the data analysis techniques that can be utilized in answering questions within the data collection. Again, putting on our detective hats, we return to our “honest serving men” as described previously by Rudyard Kipling and explore a variety of topics. How we want to look at this newly uncovered information is important. A data model is used to represent the unstructured data we collect and is an important (and complex) part of answering our questions. These data models are living and breathing entities that need to change over time or when newly discovered insights need to be incorporated into the model. These relatively long-running models tend to be complex and difficult to finalize, and as a result, many people may want to take a less-detailed view of the information. Many choose a real-time view of the data, where watching metrics or trends in real time (or near real time) provides a valuable, yet low-cost, set of insights. As an alternative between long-running analysis and a real-time view lies a structured search model that allows for the searching of common words or phrases within a dataset in an attempt to reveal some insightful information. Each type of analysis has its pros and cons.

In an attempt to understand what people are saying, we begin to explore some of the interpretations of the data, looking at simple metrics such as:

• In a collection that contains Twitter data related to a new product or service, what is the top hashtag?

• Are those hashtags positive or negative in their sentiment?

• What is the volume of conversation about the product or service? (Are people talking about it?)

Other techniques used to discern what people are talking about include the use of word clouds or the collection of top word groups or phrases. These visualizations can help analysts understand the types of conversations that are being held about the company or service in question. More advanced analysis may include the use of a relationship matrix that attempts to understand the interrelationship between concepts or terms (for example, how is the public’s view of customer service correlated with perceived cleanliness of a store?).

Marketing teams will be sponsoring advertising campaigns or coming out with press releases at strategic points during a new product release or during a particular point in time—all in an effort to attract new customers while exploiting the loyalty of their existing customer base. But is their message reaching the intended audience? The question of where people are talking becomes important in evaluating the outlets that people use when discussing a topic. If the company is advertising mainly in trade journals but there is a large amount of conversation happening in Twitter, would the message be better spread via microblogging? (Or perhaps the use of microblogging can augment the marketing message?) Along those same lines, if we stand on a box in the center of a square and preach our message, do we want to do it in the middle of the night when the square is empty, or at lunchtime when the square is bustling with traffic. The same is true in the social media space: when we choose to disseminate information may be just as important as where.

Information Interpretation

Once we have all of this data reduced into information nuggets, making sense of the information becomes paramount . In some cases, the goal is not only to identify who is doing the talking in our analysis but, more importantly, who is influencing the conversation or who is influential in their thoughts and opinions. It’s important to remember what SunTzu once said: “Keep your friends close, and your enemies closer.” The identification of the “movers and shakers” can be important in social media; these are the individuals we want to follow or attempt to get closer to in order to have them use their influence for us as opposed to against us. In other cases, what people are saying about a particular issue or topic is the object of the research.

• Are people excited about the newly designed web experience that our company just released, or are they talking about the difficulty in finding information within our website?

• How critical are the outsourcing decisions that we just made to the brand perception of our company or product?

• What were the key issues or topics that people cited when they were expressing negative sentiment?

In our experience, we have also encountered cases in which the where is the most important finding. For a newly launched marketing campaign, is the conversation happening more in company-sponsored venues, or is it also happening in neutral venues? Analysis and insights around when are also important. For example, is the sentiment for your company becoming negative around the same time as the sentiment for a key competitor (perhaps indicating a downturn in your market)? More importantly, has sentiment for your company or brand gone negative while the competition has gone positive?