A new focus on user-friendly data analysis

Developers and investors look to close the gap between data analysis and user experience.

BackType2.pngThe ability to easily extract meaning from unwieldy datasets has become something of a Holy Grail in data analytics. Technologies like Hadoop make it possible to parse big datasets, but the process isn’t to the point where an average business user can run reports and conduct analysis.

Roger Ehrenberg, managing partner at IA Ventures, touched on the importance of user interface in a recent interview. He noted that in some ways, we might be developing in the wrong direction:

There’s a new-found appreciation for an even greater focus on UI and UX. The experience that a consumer has with a product or application — it’s almost as if you need to start there and work backwards as opposed to [saying], “Hey, I’ve got a cool technology or application. Let’s see if this thing works,” and then hacking together a UI. Oftentimes, the UI is a secondary consideration and the core technology is the primary. But in many ways you almost want to go the reverse.

The gap between technology and user experience is not lost on developers — or investors. BackType, a social analytics company that developed ElephantDB to export data from Hadoop, just brought in $1 million in investment funding. The company’s platform serves as an interface for users to measure social media impact.

The day before BackType announced new funding, HootSuite launched a social analytics dashboard that lets users track social brand performance across platforms like Twitter, Facebook and Google. Adobe has also joined the fray with Adobe SocialAnalytics, a service scheduled for later this year that expands on Adobe’s SiteCatalyst product and other Adobe Online Marketing Suite tools.

One additional signal to watch: Social data, which is the current focus of most of these companies and dashboards, may ultimately serve as an entry point for different and deeper types of data analysis. Once users get accustomed to asking big questions against big data, they’ll likely expand their queries beyond the social realm.


tags: , ,

Get the O’Reilly Data Newsletter

Stay informed. Receive weekly insight from industry insiders.

  • Jenn,
    Thanks for a great article. IMHO too many vendors in the space have ignored the UX to concentrate on eking out the last bit of performance from their backend. Speed of calculation and scalability is important, but it is just the foundation for what is needed in a next generation analytics system.

    If the analytics system is not designed so that the avg. business user can easily self-serve – including creating, sharing and comprehending analytics within the system, the speed of the backend engine is irrelevant! In those systems, the gating factor will always be the time it takes the user to explain what they need so that IT or a consultant can perform the magic incantations and produce the desired result. The user then will usually end up in Excel to further massage the data into exactly what they need.

    Complex analytics will always require external experts, but the majority of tools in the big data space reflect their heritage. They were designed by programmers for programmers and are often beyond the capabilities of all but the most sophisticated IT departments and statisticians.

    Analytics systems on top of big data can change the world in so many positive ways but only if we lower the barriers to entry for their use. Any system that requires a PhD in stats, an understanding of Map Reduce, or denies users a point and click interface is not going to reach enough people to create the efficient data driven organizations we have all been pontificating about.


    CEO, PatternBuilders

  • 3 decades ago I was initiated into the world of COBOL, which I quickly left. The prime directive then was “design the reports, then define the files for the reports”. This was before the RDBMS was a commercial reality, but after Dr. Codd’s paper was published.

    The problem with the approach is that it devises logically impaired flatfile datastores. Such datastores are efficient only for the narrow, and I do mean *narrow*, reporting purpose. A UI screen, no matter GUI or CHUI, is just a report. We’ve learned a lot more since then. Well, some of us have.

    A logically structured datastore will have an obvious UI, in terms of content. What widgets to use is another matter, and not relevant to deciding the structure of the datastore; make it as pretty or ugly as you see fit. That is the hard lesson learned by COBOL coders back then, and will be learned all over again by today’s java/C#/whatever coder.

  • — Any system that requires a PhD in stats, an understanding of Map Reduce, or denies users a point and click interface is not going to reach enough people to create the efficient data driven organizations we have all been pontificating about.

    I suppose one might assert that any old chum should be able to do brain surgery without being a neurosurgeon. Remember, a little knowledge is a dangerous thing. Much of the evil in this world is perpetrated by know-nothings who assert that their “I’m just like you”-ness is justification for decision making.

    Should we empower half-cocked partisans with easy to use weapons? Just what we need, Tea Baggers (bankrolled by the likes of the Kochs) with a tool to produce slick, but false, data presentations. Figures don’t lie, but liars figure.

    Some things should be difficult to do; the incompetent and stupid are less likely to muck of the world for the rest of us.