Web 2.0 Expo: Social Media Measurement Expert Tips
Measuring social media remains a black art. Experts offer advice, strategies, and favorite tools.
Social media measurement is hard to do, imprecise – and very necessary.
"New Metrics for New Media" was one of the opening day workshops Monday at Web 2.0 Expo, a UBM TechWeb/O'Reilly Media event in New York. The session featured Susan Etlinger, an analyst with Altimeter Group; Blake Robinson, who recently left a role with Rackspace startup program to work as an independent consultant developing a project called False Idols; and Margaret Francis, formerly of Lithium Technologies and recently a consultant to Exact Target on the development of its CoTweet acquisition, now working on an unnamed new venture.
While there is no single return on investment metric for social media, smart leaders seek out the ones that make sense for their business, Etlinger said. "It's really easy in social media to measure fans and followers, retweets, and stuff, but that's an activity--that's not necessarily a result." Instead, you want to get to a comparison like the cost to handle a customer inquiry online, which might be 10 cents, versus handling it over the phone at $6 to $10 per call when all infrastructure and labor costs are factored in. "Then you can come up with a social media metric like percentage of inquiries resolved outside the call center."
Frances noted that in the old media world the link between marketing and sales was very tenuous and could only be detected indirectly through means such as surveys. Digital media created a lot of excitement and expectations around the idea that direct links from marketing to sales could be established, she said. "However, the science of attribution is still pretty darn weak." The tracking cookie may show what link you clicked, "but it doesn't show all the other influences that caused you to click" such as TV advertisements or offline conversations with friends, she said.
[Social media is a powerful tool to connect with customers, but it can create big problems for your company if it's not done right. Learn more at 10 Social Networking Don'ts.]
Some old media metrics have social media equivalents--for example "share of voice" measurements (of the percentage of publications carrying a brand's story) now translate into "share of conversation" measurements (of how much social media users are talking about the brand). The two are different, but at least there is a parallel, Frances said. On the other hand, social media "engagement"--the extent to which people are motivated to interact with a brand or proactively spread its messages and links--does not really have an equivalent in old media, she said. The closest match would be "likelihood to recommend" findings from survey data.
Also, no one has quite figured out how such engagement metrics translate into business results, Frances said. "I haven't seen anyone say how the sum total of all these shares and Diggs translate into sales of Coke or Pepsi."
Even as you strive to make your measurement program the best it can be, "you can't take it too personally" if the meaning of the data isn't perfectly transparent, Frances said.
The panelists recommended some specific tools, a part of the discussion where Robinson often took the lead. Some of their picks:
-- Simply Measured, an affordable service that works with Excel, which the panelists used to generate many of their slides on social media trends.
-- Radian6 for social media monitoring and sentiment analysis.
-- Crimson Hexagon, a Radian6 alternative Robinson praised for the quality of its sentiment analysis and semantic understanding.
-- NetBase, which Etlinger recognized for going beyond showing positive and negative sentiment to scoring the intensity of that sentiment.
-- Tracx, an end-to-end social media management system that Robinson said includes "good workflow" for following up on the results of the analysis.
-- Carrot2, an open source tool you can use for your own data clustering analysis.
-- Chartbeat, a realtime Web and social analytics tool.
-- Peer Index, which Robinson called out as an alternative to Klout for ranking influencers, saying he finds the Klout algorithm too unstable to be relied upon.
Some of the results from these tools, particularly when it comes to sentiment analysis, have to be taken with a grain of salt, understanding that software algorithms don't always parse human language correctly. "Sentiment is definitely a dodgy metric, and you have to watch out for how it's being done," Frances said.
Because of the complexities and technologies involved, Etlinger said she sees many companies deciding to outsource the whole process of analyzing social media. "If you outsource, you really have to think it through," she said. For example, if you outsource to your advertising or marketing agency, you're likely to get back glowing reports of how their efforts are boosting your company's reputation online. An agency is also going to see social media analytics through the lens of marketing automation, she said. That means they may miss other signals that ought to be flowing to your customer service or product development teams.
Most of all, outsourcing runs counter to the realization that organizations need to become more analytical to thrive in the world of new media, Etlinger said. "You're slowing down that organizational learning" when you outsource, she said.
Learn the secrets to getting your employees to share and collaborate with one another in this Enterprise 2.0 webcast. Here's a hint--it's not about the technology, it's about your people! It happens Oct. 12. Sign up now. (Free with registration.)
