Have you ever wondered how far that one person from Guam got while they were watching your newest video on their iPhone last Monday?
After about a year of spreadsheets, information architecture matrices, and a couple of mocks featuring one of my favorite dog videos, we set out to answer important questions like this one with the launch of a seriously powerful new anayltics tool. If you happen to have some videos of your own on Vimeo, TAKE IT FOR A SPIN.
In with the new
Vimeo is all about offering the best-in-class tools for anyone looking to host their videos. Unfortunately, some tools get left in the shed for a while. The advanced stats tool was in need of some love, so we decided to give it a complete overhaul. Not only did I completely redesign the UI and UX for stats, but we also rebuilt the entire system for collecting and reporting data, which was no simple task.
One of the stats tool's biggest shortcomings was the fact that we just weren't using a lot of the data we were collecting. There was so much more we could be surfacing to users, like in-depth geo data for their plays, and viewer engagement on their videos. So the biggest challenge we faced was designing and building a tool that organized all of this data in a way that was easier to access and digest.
Laying the foundation
I did a lot of the research and competitive analysis for the new stats platform. Looking through other analytics tools, I knew I wanted to avoid making stats too complicated. We basically wanted to give users a powertool, but didn't want them to feel like they needed a license to operate it. Our target users differ from your run-of-the-mill number crunchers in that they are mostly creative profressionals interested in figuring out how their videos are performing, but don't have experience parsing a ton of complex data.
After organizing all the different data points we collect and seeing where all the vectors overlapped, I put together a spec and direction for an interface that allowed users to essentially build query strings (a pretty daunting method of retrieving specific stats that's popular on other platforms) using friendlier, more organized methods, like switching between views and adding filters.
The key was designing something that was flexible and more future proof than our previous tool. With this new framework, we're able to easily add new data points as we start collecting them without the interface running out of room or becoming more complicated to use.
The old dashboard for stats was cluttered and unfocused. There wasn't a lot of care given to empty states, which resulted in a lot of sections that were just big holes with no data or messaging. To address these problems, we did a round of user surveys to identify how our users ranked the importance of different stats. The clear winner was play count, so we decided to set up the dashboard accordingly.
Users can see an at-a-glance summary of their recent activity across plays, finishes, likes, and comments, then get highlights for their play data. The stats dashboard is meant to be a jumping-off point into the more in-depth reporting pages. So aside from being able to navigate using the menu at the top of every stats page, users can also click through from any module on the dashboard to the corresponding report.
The reports are where users can really get the most out of stats. Improvements included revamping our data visualizations using HIGH CHARTS, adding an interactive map, introducing filters to customize and save out specific views, streamlining the amount of stats the user has to choose from, and building the ability to select, expand, and narrow in on the data in the table.
Users can now view their stats over time in the date report, pinpoint what's happening world-wide in the region report, see how they're performing across different devices and platforms in the device report, look at what sites people are watching their videos from in the source URL report, and see how people are engaging on their individual clips in the video report.
City-level geo data
The region report now has an interactive map that we put together using OPEN STREET MAP. Users can zoom in and see breakdowns for their stats all the way down to the city/town level.
For the video report, we built real-time graph overlays that show users what portions of their videos perform the best, and how long the average user watches.
One thing we hadn't done a great job of in the past was marketing the Vimeo stats tool. The old page featured an outdated video and had a big CTA to upgrade, but it wasn't a very good resource for educating and upselling potential users. So I put together a marketing page that highlighted the new features we added to the platform with a breakdown of benefits for each subscription type.
Vimeo stats was a great opportunity for me to take a hands-on product role and really lead the architecture of a big project. I definitely learned a lot about how data infrastructures are organized, and how to better work across a bunch of teams. Stats was also a lesson in the importance of iteration. We set up this massive undertaking as one big release, which really bit us in the donut charts. So after this, I'm a big supporter of more agile approaches with phased releases. There really is way too much going on here to get the job done in a few screenshots, so if you're a Vimeo user CHECK IT OUT for yourself.