Ian Thomas
With all the many challenges of capturing, storing, transforming, and delivering big data to business users, it’s easy to overlook one vital aspect: how to get those users to incorporate data- driven insights into their daily decision making. This chapter will focus on several important techniques for achieving this, including organizational design and staffing, the cre- ation of robust processes around data publishing and the evolution of data assets, and delivering technology solutions that address the needs of a diverse user base.
CONTENTS
Context ... 154 The Triad: Skills, Trust, and Access ...155 People ...156 Functions and Scope—The Goldilocks Principle ...157 If You Want to Get There, I Wouldn’t Start from Here ...160 Not Too Hot, Not Too Cold ...161 Process ...162 Regular as Clockwork ...163 Only Half the Story Can Be Worse than No Story at All ...164 Adding Value without Adding Problems ...165 Rewriting History ...166 Technology ...167 All Users Are Not Alike ...167 Summary ...170
CONTEXT
My own entrée into the world of big data began in 2000, when I joined a tiny software start- up in the UK that specialized in web analytics tools. The web analytics industry was very different from how it is today. Back then, dozens of technology vendors strove not just to convince potential cus- tomers to buy their solution, but to convince organizations that web analytics was worth investing in at all. Of course, in those days, big data meant megabytes or possibly gigabytes of data per day, rather than the terabytes and beyond of today’s world. But in many ways, the web analyt- ics industry at the turn of the millennium was not so different from the big data industry of today.
In 2000, the dialog with web analytics customers (and within the indus- try itself) was almost exclusively focused on technology. The available tools rapidly gained new features, and these features were paraded in front of potential buyers: funnel reports, 3D visualizations of traffic patterns, heat maps. The total number of out- of- the- box reports provided by each product became a competitive differentiator, until it reached such a ridiculous extent (“We have over 300 prebuilt reports!”) that it became an object of parody.
For their part, the industry’s customers were on an incredibly steep learning curve. Many of them had only recently made a decision properly to invest in the web, and it was still very much seen as an IT function in many organizations. For many, the only key performance indicator (KPI) attached to the website was a simple yes/ no answer to the question “Do we have a website?”
Since IT people are used to evaluating tools based on features, many purchasing decisions were made, and later regretted, based on whichever tool had the flashiest demo. Our own little company wasn’t immune to this phenomenon—I remember several features that we shipped specifi- cally because they looked good on the screen.
What many of these early adopters (and the vendors that supplied them) discovered, however, is that getting value out of an investment in web ana- lytics was not a simple case of installing some software, setting up some logging, and then waiting for the insights to come rolling in. Most orga- nizations simply didn’t have any staff with either the skills or the time to spend analyzing website traffic. As the decade progressed, organizations like the Web Analytics Association (now the Digital Analytics Association) would spring up and champion the role of the web analyst—but then,
hard- pressed IT staff or marketing managers were expected to take on web analytics in addition to their normal duties. The vendors themselves had yet to develop the full- service implementation and analytics capabili- ties that they have today, and no third- party analytics services compa- nies existed.
Not surprisingly, this combination of an overemphasis on technology, unsophisticated buyers, and a shortage of implementation and analysis skills meant that the early days of web analytics were bumpy for many. At the heart of the problem was the fact that customers simply weren’t getting a return on the investment they were making. Once implemented, many tools (including our own in several cases) gathered metaphorical dust as they were largely ignored or forgotten about by users.
Of course, the web analytics industry grew up. Substantial vendor con- solidation created a much- easier- to- navigate field of players for custom- ers to choose from, who themselves became more sophisticated. Most importantly, a vibrant services industry grew up around the discipline. Nowadays the discussion at industry events like the eMetrics Summit is hardly at all about which technology to choose—it is about best practices and advanced analytical techniques like predictive modeling.
Today’s big data industry is not as immature as the web analytics indus- try of 12 years ago, but it does share some of the same challenges. Many discussions of big data today tend to focus primarily or exclusively on the technology. This is partly because the technology landscape is currently changing very rapidly. It’s also caused by the wide variety of vendors who are in the market with solutions that are not easy to compare with each other. As a result, making technology choices for big data is very difficult, and so a lot of energy is expended on these discussions.
However, as with the early days of web analytics, this focus on technol- ogy crowds out discussions about the real purpose of implementing big data systems, which is to enable people across organizations to rely on data every day as they make many large and small decisions about how to do their job and run their business.