The Problem With Big Data: I Can Tell You Everything, But You Will Learn Nothing
Thanks to a media blitz in everything from The Financial Times and The Wall Street Journal to Wired and Forbes, pretty much everybody has heard about big data by now and developed an opinion on the subject.
Ever since the expression’s meteoric rise about two years ago, the business world has been clamoring for more data. It doesn’t matter if you’re the marketing manager or CEO; it’s all-aboard the big data ship in order to become more “data driven.” However, as mentioned in an interesting article by TechCrunch, big data is not the new, revolutionary idea people make it out to be. In fact, big data originated more than 10 years ago in the halls of the University of Pennsylvania when Francis Diebold wrote about the term in relation to financial modeling. Since then, big data has joined the buzzword hall of fame with terms such as cloud computing and Web 2.0.
For digital analysts, it’s a refreshing shift to see corporations making decisions based on Web data. We now live in a world where marketers look at data from a variety of channels — such as Web, social and search — to analyze what people are doing in order to improve their digital experience. There has been a trend, however, of businesses moving from using data to answer targeted, insightful questions to asking for everything. And when I say everything, I mean every metric, every insight, on every page. If you could measure it, they wanted it reported.
The problem with this approach has nothing to do with being able to collect the enormous amount of data a large website creates; we can do that. This is a problematic approach because it’s inefficient. You can have all that data, but do you need all that data to answer the questions you’re asking? For example, if a large organization requested a reporting system that analyzed every page ranging in size from 1 million visits a month to 10 million, I could do that to some degree. However, all extraneous digging of data will not provide valuable insights in a timely manner. In fact, this kind of analysis will not make you more efficient; to the contrary, it will make you slower than you were before. Before wasting your time and mine, I would ask, “What are you trying to learn?”
Instead of analyzing everything all at once, be specific. Define a goal, behavior or campaign that you want to measure, and then come up with a plan of action on how you will capture and evaluate that data. Before you start pulling numbers and populating spreadsheets, keep these five points in mind:
1. Come up with specific questions that you can measure successfully: Again, these questions cannot be extremely broad. For example, asking your marketing team to track the path of every visitor to your homepage is a bad idea. Instead, ask questions like, “Are our social media campaigns impacting online conversions?” or “What was the impact of our PPC campaigns on sales?” If you are still struggling for questions, try asking yourself where you spend the most marketing dollars online. Are you measuring that channel’s success? If not, that’s a good place to start.
2. Make sure your web tracking is spotless: Implementing your website tracking code is one of the most important and underrated steps when creating an analytics plan. If there are implementation issues on your website, then this will critically endanger the validity of your data. And if your data is unsound, you may be making erroneous decisions, which could cost you a lot of marketing dollars down the road. Therefore, whatever analytics tool you are using, it is imperative that your implementation is simple and easily replicable. Here’s a test: Give your implementation how-to guide to an intern. If they can replicate it for new web pages, you’re good to go. And if you don’t have a how-to guide, make one.
3. Answer your question, then go deeper: Once you get started, you must still resist the urge to dig into every insight. Try to answer that specific question you asked in the beginning, AND THEN investigate more. Let’s go back to the PPC example I brought up earlier. After some analysis, you may see that visitors who clicked on ads from Google are making bigger purchases than those coming from Bing. In this case, you may want to dig deeper and see how traffic from Bing ads is behaving on your website or you may want to dig deeper into the specific keywords.
4. Don’t get silo’d in: For large companies, it is easy for one marketing team not to know what the others are working on. As a result, sit down and have a cup of coffee with your fellow marketers. You just might find out that the spike in Facebook traffic was the result of an intense campaign that your co-worker is responsible for.
5. Have the appropriate number of analysts on your team: Once your analytics team starts getting the hang of it, you may start giving them more work to do. Before you pile it on, make sure you have the right amount of analysts in your department to take on the load. Otherwise, valuable insights might be slipping through the cracks just because people are overwhelmed.
Right now, we live in a world where almost every web interaction can be measured. For marketers, the problem is not how we access this data but how we approach it and provide valuable insights. While the thought of big data can be crippling, you need to sit down and thoroughly think about what questions are important to answer to further your brand. Once you can grasp that concept, you’ll finally see how commonplace big data really is.
After two years of consulting with startups and small businesses, Jason Bunk joined Digitaria in 2012 as a web analyst. Jason helps brands understand consumers' digital behavior, using tools like Adobe Omniture Suite and Google Analytics to measure and provide actionable insights to improve campaigns and digital properties. Jason started his career in marketing and business development after earning his Bachelor of Science in engineering from Binghamton University.