Katie Delahaye Paine, chief executive of consultancy KDPaine & Partners, was speaking to FoodNavigator-USA after analyzing 301,497 social media posts on GMOs, vaccinations and high fructose corn syrup (HFCS) from October 2010 to October 2011 using Radian6 online monitoring technology.
This found that more than one-third of comments were generated from ‘pay-per-click’ sites, ‘content farms’, ‘robot responders’, faux social media accounts or firms with a financial interest in the debate such as marketers of health products and alternative therapies that were using discussion platforms to market their products, she added.
“Clients are often so overwhelmed by the volume of data on social media sites that they are increasingly trusting third parties to analyze it for them; but they need to be clear what methodologies these companies are using and how transparent they are being.”
1% of posters were responsible for 15% of all posts on HFCS and GMOs
The research also revealed that a large proportion of the remaining posts (that were generated by real human beings) were created by a small number of posters with strongly negative views creating “a disproportionate amount of conversation”, said Paine.
"More than a third of all posts on GMOs and HFCS were created by 10% of all posters, while 1% of posters were responsible for 15% of all posts," she said.
As anyone that has ever commented on a travel website will appreciate, consumers that had a good or even an excellent stay at a hotel are considerably less likely to comment about it online than the small number who have had a bad experience, she said.
While this does not invalidate the negative comments, it should always be borne in mind when analyzing the ‘buzz’ on any controversial topic, she observed.
“In fact, mid-volume posters—who tend to be more authentic and engaged—trended more positive than other groups.”
What’s really trending on Twitter, and how is it being manipulated?
When a topic is trending on Twitter or other social media sites, it also provides an opportunity for people with no stake in the debate to jump in and plug viagra, porn or legal services, driving up the volume of posts and creating the impression that more people are engaged in the debate than is in fact the case, she said.
“Porn spammers will see what topics are trending [eg. pink slime] and then embed that hashtag in their tweets, for example.
“However, the media often gauges the level of interest a topic has by the amount of discussion that topic generates.
“Unfortunately, less than ethical marketers and pay-for-post operators have beaten them to it artificially raising the volume on certain issues like the ones we tested.”
So what’s the takeaway?
“I’m not saying that sentiment analysis isn’t useful”, says Paine, “and it can certainly alert you to potential problems, but it is no substitute for good solid consumer research, where you take a sample of people and actually ask them what they think about something.”
Click here for more information on the study, which was commissioned by the Corn Refiners Association.