Deciphering the Data
Social media evokes a variety of opinions from marketers, but many are still seeking definitive answers on how to best use it in their marketing toolkit. Most companies agree that engaging on social media is an important component of a well-rounded marketing strategy, yet many are unsure how to best communicate via the various social channels.
According to a recent report by Burson-Marsteller, Fortune Global 100 companies are mentioned online more than 10 million times each month. This level of activity generates terabytes of data every day from which companies need to derive information, assurance, or insight from online buzz. However, most organizations are not equipped to intelligently sift through millions of tweets, hundreds of Facebook posts, and an unlimited stream of forum conversations present in this big data. For the marketer driven to derive real-time action, the task may appear nearly impossible from a strategic standpoint. The use of social media dashboards and technology designed to analyze data has grown significantly in the last few years—but can new technology be trusted to glean the right details from your data?
One of the most recurring concerns about social dashboards is their ability to analyze sentiment. Sentiment is a key directional metric of data derived from the millions of tweets and posts online. Sentiment measurements can be valuable in determining consumer advocacy for your brand versus that of your competitors. By using social dashboards in combination with geography tools that identify where people are most engaged, marketers are able to create a dimensional understanding of consumer sentiment. While marketers may strive to generate the highest share of voice, this may only be advantageous if the share of voice primarily expresses positive sentiment.
Determining sentiment has been a great challenge for text mining, data analytics, and the development and application of social media intelligence tools. Many times there are questions about how accurately software can translate and attribute sentiment to a string of words. Language and sentiment have significant variation, from shorthand expressions to new lingo and even sarcasm—sometimes the context matters more than the words themselves. Before choosing a vendor to help analyze data, companies should consider the following:
Degrees of Sentiment
Almost all text contains some sentiment, but the strength and tone of it varies greatly depending on the diction used. There is no standard for quantitative labels, so in most cases, the argument isn't whether or not sentiment is expressed; it's a matter of gradation and how it's measured. Will sentiment measures be black-and-white—positive versus negative? Or will there be degrees—slightly positive, slightly negative, mostly positive, mostly negative? The details matter when data is concerned.
Knowing the geographic breakdown of the positive versus negative sentiment can help larger companies identify their problem areas. For instance, if negative feedback is received from one particular city, solutions can be implemented to fix the issues in that locale. Find a vendor that can accurately drill down to the location of the social opinion—for instance, knowing that New York City is a problem area is much better than knowing there is an issue somewhere on the East Coast. Find out exactly how deep the dashboard can dive.
How a company plans to use the data collected can also significantly affect what type of vendor should be used. A small business looking for a simple solution will have fewer requirements than a large corporation requiring simultaneous dashboards. What type of reporting will the solution provide? Can you easily generate reports at will, or is there a delay in finding actionable data? If the company needs to consistently pull reports for board members or track down data to adjust a campaign in real time, having easy-to-generate numbers is a must.
Even at their best, automated sentiment scores still can't match human perception for inferred meaning. However, a platform that allows the user to earmark a sentiment score for review or adjust settings to report such scores differently can go a long way in assuring it produces a more accurate data picture. Look for a tool that offers flexibility to tune the scoring system until a setting is achieved that works best for your overall needs.
The real key to sentiment analysis starts with outlining what the measurements will be used for and what type of data outputs are desired, and being realistic about what can be achieved with sentiment analysis. With social media, consumer-generated, opinion-based data continues to grow, giving businesses the opportunity to derive authentic feedback straight from the mouth of the consumer. This ability is second to none for achieving solicited (and unsolicited) feedback from a variety of sources in a short period of time. This gives businesses the tools necessary to change marketing tactics and programs that have better consumer impact. Plugging in to social data is one of the key tools for successfully analyzing a marketing campaign in the viral world we live in today. By finding a strong technology and data partner to complement this direction, companies are more informed and ready to take insightful action like never before.
Richard Pasewark is the CEO of Visible Technologies.
Satmetrix Debuts Social Net Promoter
Metric draws on sentiment analysis to tie customer conversations to brand customer experience strategy
IBM Launches New Class of Analytics Software
The new predictive analytics software incorporates social media activity into every business decision.
What Companies Miss on Their Social Engagement Journey
Don't overlook your advocates and influencers.
UnboundID Releases Updated Identity Services Platform
Solution provides foundation for better tracking of customer identity data.