Following yesterday’s conference, the group reassembled this morning for a second day in a bootcamp format. In today’s event, the material is being presented in the form of case studies and workshops in order to get a better understanding of monitoring tactics. To compare tools, two vendors demonstrated their monitoring of the buzz around the Monitoring Social Media conference.
Analyzing the Buzz Around the Conference
Synthesio presented their monitoring. To track keywords, they demonstrated a tag cloud, from which he could drill down to tweets and comments that used those tags. Tweets can be ranked by influence. Another tool is ordering of Twitter influencers, which uses the number of on-topic tweets along with standard Twitter influence metrics like number of followers and updates. “Other Influencers” ranks influence of on-topic blog posts and comments. (In this case, the freedmarketer held the first and second place of influence, for posts and comments, respectively.)
Brandwatch presented next. Their monitor is based on queries. For example, they queried for the #MSM10 hashtag. They showed a list of sites that included links and referring messages to the query. (Once again, the freedmarketer fared well, beating out everyone but the big guys like Twitter, Facebook, and Flickr.) They demonstrated a word cloud, as well as the ability to add custom tags or “recurring phrases”. To analyze retweets, the query can be changed to include RT. They also demonstrated a visualization that connected posters/tweeters and the topics that they discussed (in the form of “recurring phrases”).
While sentiment analysis is helpful for keywords in one or two languages, more complex monitoring requires human analysis. The growth in use of automated analysis may be due in part to volume. Human monitoring breaks down when there are hundreds (or more) keywords being tracked. While different tools have different features, there is a benefit to using an existing package instead of creating an ad hoc system. In creating your own system, you run the risk of getting garbage out of your system (if it isn’t properly calibrated.) Marshall Sponder mentioned an e-book he had written on building your own monitoring system, which should be a necessary read if you’re setting up a custom tool.
It’s dangerous to boil influence down to just one number or factor. In regard to an audience question, there is a value in retweeting, for example, there is a difference between influence and visibility. In regard to demographics, listening is limited in the ability to discern, for example, the profession of a speaker. It’s no different than walking into a room and trying to figure out who people are just by overhearing conversations. A better way to figure out the professions of monitored individuals may come from self-reported Twitter profile taglines and links. Ultimately, the tool you use is only as good as the people driving it. Even the best dashboard will give you the wrong output if your inputs are wrong.
The challenge in choosing a monitoring tool is that you want flexibility, but you don’t want complexity. You need to be able to translate data into actionable suggestions. After all, your boss or client isn’t going to want data. He or she is going to want stories. There are a lot of free tools and free trials that people can take advantage of to learn, practice, and show off what they can do with monitoring.