The first session of Thursday’s “Getting Social” track, chaired by Raj Bhargava of Satmetrix, was “Measuring your Social Net Promoter: Methods and Techniques.”
During his morning keynote, Satmetrix CEO Richard Owen announced Satmetrix SparkScore In this session, Dr. Laura Brooks of Satmetrix and Erick Watson of Metavana covered how social media activity can be measured in the context of a Net Promoter program using SparkScore.
Dr. Brooks emphasized that survey data (derived from existing Net Promoter program) and social media data may have differences – anchored data (i.e. the data has a Net Promoter score attached at collection vs. unanchored data); structured vs. unstructured; easily tied to CRM data (knowing who the respondent is) vs. not; and so on. However, despite these differences, survey and social actually yield complementary, not conflicting, information – especially if viewed through a common lens. Satmetrix SparkScore is the first product to put social media activity in a Net Promoter context.
Erick Watson of Metavana – creator of the sentiment engine that powers SparkScore – provided an overview of how the product works. The engine is focused or “trained” by industry (or “domain” in Metavana’s lingo), first collecting pertinent data through “focused web crawling” from a variety of sources – popular social web sites such as Facebook and LinkedIn, Twitter, and so on. But it can also draw from review sites such as Amazon.com, Yelp, and TripAdvisor, as well as public and private customer communities (peer support and discussion venues).
The data is then analyzed and classified, attaching sentiment and polarity (a continuum of scores from very negative to very positive sentiment) to each data point. This sentiment engine is the most accurate available, and its output is coupled with Net Promoter methodology, to calculate a SparkScore, or social Net Promoter score. SparkScore can be calculated by industry, data source, and even individual expression. The engine has been “tuned” so that a SparkScore can be related to survey-based NPS calculations, but may not yield the same score. Satmetrix and Metavana’s research has developed reliable algorithms across a number of different industries, including airlines, computer software, laptops, printers, hotels and motels, and coffee and tea brands.
Dr. Brooks continued the session by presenting more detailed findings from one particular industry: airlines. SparkScore is dynamic (it was calculated daily during the research phase), so it is able to capture the impact of discrete events and occurrences that might be missed in solicited surveys. For example, Dr. Brooks demonstrated how SparkScore reflected immediate impact in social media following Alec Baldwin’s run-in with American Airlines over not turning off the “Words With Friends” game he was playing on a mobile device (he was kicked off the flight). There was an immediate score impact on Twitter (Baldwin’s tweet following the incident was widely retweeted), and a subsequent impact on Facebook following Baldwin’s appearance on Saturday Night Live the weekend following the incident. SparkScore captured this effect in social media for both events (the incident itself and the Saturday Night Live appearance).
Dr. Brooks concluded by stressing that SparkScore is not a replacement for survey-based NPS, but rather a powerful complement that brings additional richness and power to a Net Promoter program. SparkScore is dynamic, so it can reflect the immediate impact of events (product introductions, publicity, mergers, and so on). SparkScore can provide detailed insight by source (for example, are Facebook comments different than Amazon reviews?) and theme (service, transactional experience) that provide context and depth to similar information solicited through more traditional channels.
In short, SparkScore – the expression of Net Promoter behavior in social media – adds a powerful new tool to the Net Promoter toolbox, one no company should really be without.