Among the topics covered at this week’s Sentiment Analysis’ Symposium was an exploration of just how much the negative or positive expression of sentiment about a company or a product really matters – and in what context it does. (Another one, which The Semantic Web Blog covered yesterday here, looked at the expected transition from sentiment to emotions analytics.)
Augie Ray, director of social media at Prudential Financial, and formerly a social media leader at USAA and Forrester, recounted some of the bigger blow-ups online in recent years: The passenger whose guitar was broken by United Airlines and made a Youtube video that went viral; NBC’s 2012 London Olympics coverage that was criticized for dissing a tribute to the victims of terrorist bombings, among other things; and Bank of America’s being castigated for its announced plan to institute debit card fees.
“We live and die by the concept that negative sentiment matters,” he said.
Yet, it turns out that while these events had people expressing outrage over social networks, that anger didn’t add up to corporate damage. United outperformed competitors by 150 percent six months after the video was posted; the 2012 summer Olympics were the most watched TV event in history; and Bank of America, though it reversed its debit card fee decision, continued to gain customers while its stock price soared by double digits, Ray noted.
His point, he said, isn’t that negative sentiment doesn’t matter at all, but that marketers shouldn’t be living in fear of those spikes. When it comes to monitoring social media for impressions of positive or negative text, they should instead be more concerned to “pay attention to the constant daily grind we give to consumers.” After all, “there’s a lot of slacktivism out there. People pour into Twitter and complain, but there are few chances they’ll change products or services.”
It’s About More Than Likes
At the same time, they shouldn’t be too caught up in how many Likes their brand’s presence on Facebook is affiliated with. The goal, he said, should not be to create stirrings of positive sentiment by affiliating your brand spokesperson with images of puppy dogs that people will click on, but to earn attention, consideration, awareness, and purchase intent. The Progressive Insurance’s Flo character, he said, may have tens of thousands of people talking about her on Facebook and 5 million Facebook fans, but she gets them by posting images of puppy dogs and Easter candy.
“Success metrics aren’t merely to be liked but to sell something, to make people want to be your advocates,” said Ray. “Marketers do long studies of brand attributes and likeablity is one attribute in a long list that also includes trust, appreciation, mission,… But on Facebook we settle for one attribute, likeability, and forget the other things. When you settle for likeability, at the end of the day, when it comes time for [for a customer to make a choice about a product like insurance], he goes with who he trusts, not the funniest act on Facebook.”
PayPal can serve as an illustration of how to be more effective at putting sentiment analytics to work in a smart way that speaks to helping a business better help its customers with their daily grind. As described at the Symposium by Han-Sheong Lai, director of customers advocacy and text analytics at PayPal, the company applied very basic sentiment and text analytics to identify specific issues with respect to customers and merchants.
The company was able to look through customer verbatim online comments to filter for negative impressions of its password recovery capabilities. “We haven’t paid enough attention to it even though password problems have been among the top five in our call centers for years,” he said. After sussing out negative sentiment online and making an improvement to some specific password complaints, it again applied basic text and sentiment analytics to check into whether customers took a positive or negative spin on the results.
“From a verbatim perspective we verified and got proof from the customer voice that the situation [around one particular password problem] had subsided,” he said. “We try to combine text analytics and Six Sigma principles to fix the customer experience.”