I’ve long been fascinated with the Net Promoter Score, a loyalty metric that holds companies and employees accountable for how they treat customers. It’s grown in popularity thanks to its simplicity and connection to profitability.
If you’ve ever been asked to rate how likely you are to recommend a company or offering, then you’ve experienced NPS. It’s calculated as the percentage of customers who are Promoters, rating the company 9 or 10 on a zero-to-ten point scale, minus the percentage who are Detractors, rating 6 or lower.
Historically, NPS has been derived from surveys that require active responses from customers. Satmetrix — a leading loyalty software and services company affiliated with NPS inventor Fred Reichheld — has introduced a new NPS offering that augments traditional NPS intelligence with unsolicited opinions from social media venues.
This is a logical move, as survey response levels steadily decrease and marketers realize the strategic and tactical value of unprompted consumer expression about their brands. Surveys and structured data collection will never go away, but gone are the days when marketers can ignore the overwhelming volume of customer opinions online.
Given my own passion and experience in building the pioneering company in social media measurement (Nielsen BuzzMetrics — aka NM Incite), I welcomed a recent opportunity to meet with Richard Owen, president and CEO of Satmetrix. We talked about Satmetrix’s new SparkScore offering, and the ramifications of social media’s integration with conventional customer loyalty metrics.
Following are Richard’s answers to my burning questions. This Q&A is dense, but well worth it if you work in marketing or customer experience, and are concerned about your own company’s profitability and customer loyalty.
MK: Social media monitoring — while still in infancy — has been around commercially for nearly a decade, with dozens of offerings. Why now?
RO: Although social media monitoring has been around for a long time, measurement in social media has been insular (tweets, likes, follows). There hasn’t been a way to measure online sentiment that connects to business results. This has long been a challenge and need for businesses actively involved in social media, i.e. connecting their social efforts to profitable growth.
Customers are expressing their loyalty to brands online, and companies are coming to the realization more and more that what happens in social media doesn’t stay in social media – it has an impact in the marketplace. Customer influence, both positive and negative, is amplified by social media in ways never before possible, and these recommendations and criticism online matter to sales and revenue. As social media becomes more and more relevant to companies, they’ve voiced their desire for a measurement solution that translates to other business measurements within their companies. This is where Net Promoter comes in. SparkScore measures a social Net Promoter score that shares the methodology of NPS, and NPS has become the most widely adopted methodology for measuring customer success.
MK: What was the decision-making process that “the Net Promoter Score company” underwent to commit itself to social media monitoring? Who were the champions? Who were the skeptics?
RO: It’s been something we were watching closely for some time, but it needed a combination of circumstances. We needed to find the right methodological basis, the right technology and the right people to make it a reality. And of course, our customers had to believe that the time was right to make this a relevant solution in the market.
Our first step was getting the research proof points. We looked at many approaches, but Dr. Minh and the team at Metavana, together with our own Laura Brooks, were the first to crack the code. Laura led the original research with Fred Reichheld that created the Net Promoter Score in the first instance, so she really understood the kind of results and methodology that would be needed in the social sphere. Dr. Minh and his team took our vast database of NPS industry data and created the algorithms and technology on which SparkScore was built. We couldn’t proceed unless we could prove out the techniques, as we knew that corporations would demand the same kind of intellectual rigor and proof points that created Net Promoter Score in the first place.
There was no shortage of skeptics when we started! It was no sure fire conclusion that the research would generate a result and it was a major commitment both in time and money. I have to say that once we got the concept proven, the reaction has been far more positive than we could have hoped for. People really understand that this could be a big part of the future of Net Promoter.
MK: Does SparkScore include any true integration into your conventional NPS methodologies? Or is it an add-on, complementary?
SparkScore is complementary to “classic” NPS. They are both measured on the same scale. Whereas NPS is extracted from structured surveys, SparkScore is pulled from unsolicited and unstructured data from the social web. Unlike content from surveys, social content is unanticipated, and it isn’t time-bound. With the addition of social data, we can see an integrated, 360 degree view of customer experience for the first time.
Furthermore, SparkScore correlates to NPS with a very small margin of error. Eventually, we’ll be embedding SparkScore into our Satmetrix products, so customers will be able to manage their SparkScore alongside their NPS. We recognize we’re at the early stages of innovation with SparkScore right now. We believe our customers will discover new and better ways of how to apply and integrate SparkScore into overall customer initiatives.
MK: Have you documented any meaningful correlations or predictive patterns between social media and the conventional NPS?
RO: We have documented some meaningful relationships, but it really depends on the industry and which sources we use (in social media data collection). When we have restricted the sources to more review oriented sites (for example in software) we have been able to find some close mapping, correlations above .85. In terms of the airline industry, this industry is more transactional in the types of feedback individuals give in social media (they only focus on a touchpoint or two rather than the whole experience), so the industry rankings are similar with minor variation, however the scores themselves are vastly different. In terms of sources like Twitter, etc. these are very much of the moment, without a lot of sentiment, nor a lot of staying power (meaning people move on to the next topic) and so anything predictive is very fleeting.
MK: What will the applications be for SparkScore? Will it appeal to NPS’s traditional stakeholders in customer service and customer experience? Will it appeal to new stakeholders?
We think SparkScore will appeal to all companies measuring customer experience and loyalty, in addition to the thousands of companies that are using Net Promoter. We are working on both the initial and long term applications and we keep coming up with new ideas. We think it will feed into many other firms’ technologies as a core metric.
From our perspective, we are looking at SparkScore as a benchmarking tool where brands can compare their true loyalty in the social web with other industry players. Think of it as a “stock market” for meaningful customer loyalty data. Our first release of SparkScore will be a free website that will enable major brands across top industries to access their SparkScores and compare their results to their industry’s benchmark and best and worst scores on a daily bases.
Following this first release, we will integrate SparkScore into our offerings. Internally our customers want to use it to complement their existing NPS measurements so they can understand the difference in behavior between customers they survey and those who post in social media. Everyone is interested in identifying detractors and promoters so they can do service recovery and advocacy management. Finally, we see this technology being turned on internal data so companies can generate real time NPS data from the conversations with customers they are already having. This means a massive reduction of dependencies on surveys and sample bias problems.
MK: Social media monitoring solutions run the gamut from pure chatter-mining technology to highly customized human-led research and insights. Where does SparkScore stand? And how does it differentiate from the wide array of options?
RO: SparkScore is automated. Dr. Minh and Metavana developed a highly accurate, advanced algorithm that we haven’t seen in other monitoring tools currently on the market. At a high level, we looked at declared promoters and detractors who had offered large amounts of unstructured data, and coupled that data with their “classic,” structured data. Then we used the data to calibrate and iterate through a variety of algorithms that Dr. Minh had developed. Through these algorithms, the team reached a high-level accuracy for large numbers. The challenge then became achieving good accuracy at more granular levels. Once we were capable of taking a lot of data and able to predict a company’s Net Promoter Score across 1,000 customer comments, the team moved focused on the detailed level of measuring the SparkScore of one individual customer comment.
As you can imagine, it turns out that there are challenges in doing that. Very negative customers are very easy to spot. Very positive customers are very easy to spot. People who are passives in their promoter language are very challenging, because those individuals don’t go on the social web and post passive comments. I don’t go online and write a blog to explain how ambivalent they about this product. Passives are disguised in their commentary by positive or negative sentiment words. However, it turns out that they are disguised in a predictable and uniform fashion and statistically, they are behaving consistently in their bias. So the algorithm adjusts for that.
The other difference is that the social web differs from Net Promoter Score in a fundamental way, in that Net Promoter Score asks whether you would recommend, not whether you do recommend. Those who already do recommend are “activated,” and are a subset of your potential promoter or detractor pool. In the social media environment by definition on the whole, everyone is already activated. We’ve done a lot of research over the years to understand the differences between activated and non-activated promoters. That plays into our overall theory.
MK: Lots of companies already use social media monitoring solutions and the more sophisticated brands are unlikely to quickly abandon custom implementations. Why didn’t you simply create a platform in which to import data from existing social-mining solutions?
RO: The existing solutions don’t generate metrics that meet our customers need for an NPS score. To the degree that they provide useful complementary data, we will import that data, along with feeds of other applicable social data—whether it be from Facebook, Twitter, Amazon, Yelp, TripAdvisor or any other applicable social site. The algorithmic processing required for SparkScore is unavailable on the market outside our platform, and the proven methodological linkage does not exist elsewhere.
MK: How do you anticipate social media monitoring will change NPS in the long run?
RO: We don’t look at this as a change. We’ve always seen it as a marriage. NPS asks for structured feedback using a survey. Social media gives us a deluge of feedback, but none of it is structured. Social NPS brings these two together, marrying the methodology and approach of Net Promoter with social media-acquired sentiment classification and rigorous analytics technology to create a reliable, yet breakthrough approach to measuring social media activity in a meaningful business context. As long as individuals are expressing their opinions about companies, no matter what the source, it’s our job to find a way for NPS be there, measuring those comments.
MK: What does Fred Reiccheld think of bringing social media monitoring into the world of NPS? “Bring it on?” Or “Keep it clean?” Or “Interesting, let’s experiment?”
RO: For the answer to that, I defer to Fred himself! But, I would add he’s been a great supporter of our efforts over the years and every idea that helps companies get success out of NPS.
MK: Anything else?
RO: I think overall, I just want to stress the power of social media right now. Facebook has 845 million monthly active users, and that’s just one piece of the puzzle. Customers are online, but for years, everyone has obsessed about the best ways to measure influence and impact in a business context. SparkScore is the first social media measurement solution to address this.
###
(Photo credit: Vilseskogen)