Analytics and Measurement in Practice at SxSW
For me, Sunday at SxSW could best have been described as analytics day. They decided to have all the research and measurement die-hards make the trek out to the AT&T Executive Conference center. The surroundings were plusher, the seats much more comfortable and the data was fascinating.
There were two interesting panels on measurement and analytics and the most exciting part is none of the speakers were from vendors. Shawn Brown from MIT’s Sloan Management Review and Chris Traganos at Harvard spoke about how they measured Web success and worked to increase traffic; while Elizabeth Winkler from the University of Texas – Austin, discussed how she and her colleagues developed a model that showed how Twitter chatter accurately predicted movie revenues.
Harvard was doing some very interesting things and relies primarily on five social media tools:
By comparison, the tools MIT highlighted included:
- Google analytics
- Bit.ly (Think about using Bt.ly pro so you can have custom links you can control if Bit.ly goes out of business in five years)
- Google alerts
These are tools most of our readers know, but it shows that tools like these, used intelligently and backed by great content can deliver very good results. A few practical tips on how they grew traffic included advice such as:
- Lots of tags are your friend: every story they push has 30 tags that are fed into blog aggregators
- If you aren’t using Facebook’s OpenGraph and you create content, you need to start doing so. Instead of the share button, use the like button. Despite the power of the tool, you should check URL Linter to see what Facebook is getting and not getting.
- Chartbeat is a perfect complement to Google Analytics. I haven’t used it on my sites, but I plan to check it out. It gives you real time analytics in a way Google does not.
- Probably the biggest “huh” in the session was the tidbit that the smaller the URL, the less dense the QR code is, so it works faster. Keep using those URL shorteners, and make sure they point to a site that is optimized for mobile.
These are great practical tips for growing and analyzing site traffic.
This was a great primer for the next session that looked at how Twitter chatter could be used to tell how good a movie will do on a weekend. Elizabeth Winker and her team used a set of 20 servers to gather and aggregate tweets on movies and store than in a database for further analysis.
They first eliminated the false positives (which was very tricky for the movie “The Hangover”) and broke the results down by positive, negative and neutral using an automated system they built themselves. By then analyzing how many people said they planned to go to the movies and the reaction and buzz afterwards, they showed how positive Twitter chatter mapped nicely to box office sales for the 60 movies they analyzed.
One area Winker touched on briefly, but I think is worthy of further consideration is the power of Twitter and geolocation. They could map the Tweets on a certain movie to the subset of those that have enabled geolocation on their phones and break out the analysis and prediction on a regional level. This could help allocate ad and marketing spends to the areas where it is most needed. One Winkler didn’t explore that I would love to see is if there is greater correlation based on the chatter that is made immediately after a movie (which a movie company should be able to do by mapping the tweets to theatre locations). This also has implications for CPG and consumer technology companies.
The panels weren’t earth shattering, but they gave good practical advice and a glimpse into the future of quick, high-volume analytics.
Latest posts by Mark McClennan (see all)
- What Can We Learn From the First Social Media Fail of 2015? – January 5, 2015
- When B2B Tech Memes Go Bad – July 14, 2014
- The Wearable Wheel: Visualizing Wearable Tech – April 15, 2014