Approaching Robust Data: Tips and Tricks From the Front Line

Approaching Robust Data: Tips and Tricks From the Front Line

The task at hand was clear: collect, organize, and interpret data for the upcoming Retail Mobile Benchmark Report, which examines the disparities between mobile and desktop performances for average US websites. A broad topic like this could easily span 20 or 30 industries.

As this was my first time approaching this data (based on analyses of more than 290-billion visits to more than 16,000 mobile sites and over 85-billion app launches), I’ll admit that I was a little more than overwhelmed. But, knowing that it would take significant effort on my part to familiarize myself with the data well enough to report something interesting, I jumped right in. What I learned along the way will, undoubtedly, save me time as well as a lot of headaches in the future. Following, you will discover some of my takeaways — the lessons I learned, key insights I gained, and perspectives that shaped my experiences all along the way.

1. Manage Expectations From the Beginning.
Almost immediately, I discovered that the key to approaching data of this size was to understand, in advance, exactly what it was that I wanted to deliver. Access to this much information can surely seem imposing, but without a plan, it’s almost impossible not to become lost in the process. This meant that I had to look closely at the kind of data I had access to and think about how I wanted to present it. Taking time in the beginning to devise a strategy helped me plan, so not only could I deliver what my audience was seeking, but also manage expectations for the results.

2. Plan for Both the Data and Its Analysis.
A. Pull Data in Advance.
One of the first challenges I faced was trying to decide how I wanted to slice the data. Should I break it down by states or by industries? Did I want visits to retail websites separated by state or by mobile-device type? The most efficient method was to plot all the different options and pull all the data up front — which meant you wouldn’t find yourself halfway through a project, wishing you had sliced the data by industry rather than state.

B. Know What Your Audience Cares About and Align Data Accordingly.
With a robust dataset, it’s easy to simply dive right in and create tons of charts. However, if you don’t know what your audience cares most about, you could very likely be wasting your time. An internal, pre-analysis review helped us make sure expectations were aligned before we began tackling main ideas.

With regard to this report, for instance, I knew a shift was taking place. People are moving away from using desktops (the traditional browsing method) and toward using mobile instead. While interesting, the crux of the issue isn’t the move itself, but rather, how much the move is costing companies. After identifying what my audience cared most about, I focused on the implications associated with people leaving desktops and moving toward smartphones, and then I was able to align the data accordingly.

3. Build the Case, Visualize the Results, and Make It Interesting to Your Audience.
The ultimate goal for any data analyst is to communicate insights clearly and effectively — to build the story behind the data. Internal support is critical to achieving this, and asking senior-level analysts for technical assistance is a very smart move. Having an expert on hand with whom you can collaborate and bounce questions off of is invaluable. In many cases, a senior manager can resolve an issue in mere minutes that could otherwise take several days to resolve. It was extremely helpful to have an expert nearby, especially in the beginning phases when I was slicing data.

A. Step Back and Look at the Bigger Picture.
Often, data contains contradictions or unexpected surprises. In this case, for example, I knew smartphone usage was climbing, as tablet use was dwindling. I also knew that there were major behavioral differences between smartphone shoppers and desktop shoppers. But, what was happening overall? What were the net effects? Taking a step back to view the full context surrounding the data helped me to compare a specific data piece against other views of it to solidify the narrative.

B. Balance Tech With Storytelling.
Any great data analyst is driven to capture audiences’ attentions using compelling headlines. But, it’s also imperative that you guarantee technical accuracy so you can defend the results to your peers. This is why the time you spend vetting data and making sure it’s clean is so crucial. Nevertheless, what you ultimately create must also make sense. A strong manager is an excellent resource for helping to balance tech-heavy aspects while shaping the story behind the data.

C. Welcome Feedback.
To ensure accuracy, data analysts are often required to vet data against any questions senior analysts have. Doing this well means spending some time studying the industry, understanding trends, and making sure that whatever you find is defendable. Identifying the reasons behind various findings, uncovering issues, and adjusting for them via feedback are critical steps in the process.

Ultimately, the Best Tips Begin With You.
The biggest lesson I learned in the course of this project also turned out to be the simplest: remain organized. From the very beginning to the absolute end of the process, carefully following steps, avoiding shortcuts, and doing things right the first time proved essential. If you can manage those things correctly, you’ll have set yourself up for success.

Recommended Articles