Visualization, especially live visualization, of social media data has never been so popular. I’ve never tried to attend a SXSW session so oversubscribed; there were over 400 people waiting on line to get into this session after the ballroom closed. This is testament to the number of startups working to help clarify the nature of conversation to through their tools, but the line I waited on was full of agency folks and brand representatives. Everyone wants to understand how to visualize the slice of social media that matters to them.
Analysts talked about results gained using their tools rather than functionality, and many are working to integrate web side platforms like Google Analytics in order to more closely track connections between social media and onsite behaviors. People said “ROI” less and were more focused on tracking overall effects, recognizing that sorting social media effects from other marketing is extremely difficult and it is better to focus on improving the KPI’s that matter to your goals.
Overall, it was interesting to note that none of the major social media analytics tools we use seemed to have a floor presence. Only the API based guys were trying to snag the rubes this year on the floor; everyone else was sharing best practices in sessions. Apparently, social media analytics sells itself.
The influencer analysis problem is getting a lot of startup attention. Aside from the general adulation of Klout, I had three different influencer analysis tools I’d never heard of come up to me after the session I attended to discuss potential solutions to the problems I raised in the Q&A around cross channel analysis. I’m looking forward to reviewing their solutions, but I believe that semi automated influencer analysis in non-owned channels is a very hard problem with many dimensions that will defy elegant solutions and will continue requiring significant manual effort. .
If we’re lucky, Twitter will do something about the API tool data problem. The Mashable article that we sparked is the first time this issue has been addressed in such a public forum to my knowledge. With luck, the issue of data quality and coherence will rise to the fore, and all of our stats will start making more sense in relationship to each other.