3 Challenges for Reporting with Data

Bringing data to the conversation in a relatable, digestible fashion has always seemed like a challenge to me. Off the top of my head, there are 3 main data challenges I personally face each day, and some quick suggestions of how to deal with them.

1. Information Collection

It seems like any conversation about some factoid or percentage ultimately highlights another statistic that would have been nice to have.  In fortunate scenarios, that “desired statistic” is not necessary for the conversation to proceed.  But regardless of need, putting a comprehensive data collection process into place moving forward can be tough.  Sometimes a creative analyst (thought by some to be an oxymoron) can find an indirect measure to stand in for a metric not previously collected.  But think about how often you’ve wanted to assess effectiveness of a process but discovered you did not have all of the pertinent information – very frustrating!

Fortunately, data analyses can be an iterative process, which almost always mature over time. Having re-usable processes in place that can be tweaked to include or exclude variables over time will cut down on the time the end-user needs to invest, leaving more time to ensure that those new data-collection procedures are valid, and achieving what you had hoped.

Rapid Insight has put together a series of 20 videos that cover different topics around predictive modeling. As I was writing this, I was thinking of some of those videos and some of the great suggestions included in them that relate to how to address these data challenges.  Check out this short video on Data Considerations for some more related tips and ideas.

2. Reliability/Mixed Messages

I’ve been following the Sweet Briar College closing controversy pretty carefully.  It’s a relevant story for plenty of reasons. The governing board decided to close the college due to financial struggles and the general lack of any viable fixes.  One of the key metrics in the controversy is the size of their endowment.   You would think that the size of the endowment would be well known by now…  yet in two articles, published on the same day, both by credible sources, that number varied between $86 and $94 million dollars. When even a straightforward financial figure is too slippery to pin down, it’s hard to blame anyone who doesn’t accept statistics as reported.

It may sound alarmist, but I don’t think I need to persuade you that all the skepticism out there is undermining our effectiveness as purveyors of fact. One thing that I’ve found really helps folks buy into the process, and trust the results, is using an ETL (data cleanup tool) that is easy to step into, from a non-technical perspective. Take, for example, this data processing job from Veera Construct:


Most people will be able to understand the nature of this job simply by looking at it, perhaps without explanation.

3. Indifference

I think this is the biggest issue facing data analysts. You can lead the consumer to data but you can’t make them care about it.  Without a clear, concise explanation – better yet, a suggested course of action – to accompany the numbers, people are likely to turn away from the discussion altogether. At the very least, there is a message behind the data, and how these facts get reported can influence whether or not viewers miss the point of it all.

A really great way to address this point is to deliver results in a format your audience can understand. That might include a conversation with the target audience, or some trial and error, but in the end, investing in the process of using data to inform decisions will be worth it.

Consider the following examples of how visualizations can get people to care:

This graph illustrates (poorly) the expected employee turnover for the next month. Nothing about it really communicates your standing with regards to the company's goals.

The graph above illustrates the expected employee turnover for the next month. It doesn’t give viewers an idea of what kind of shape the company is in, though.


This graph contains the exact same information, but now, viewers can immediately understand that they are nearing a level of turnover they have identified as critical at their organization.

This tri-colored dial chart contains the exact same information, but now, viewers can immediately understand that they are nearing a level of turnover they have identified as critical at their organization. Even if they aren’t familiar with the numbers themselves, the use of colors can send the desired message.


In another context:

There's a lot of great information here, but nobody is going to want to read through it to figure out what they need.

There’s a lot of great information in the table above, but nobody is going to want to read through it to figure out what the data is telling them. It would be far better to help viewers get the insights faster, by narrowing the view.


Despite also being a data table, this helps viewers see the important takeaways much faster.

Despite also being a data table, this “Top 10” style approach helps viewers get the critical takeaways much faster.


For more on tips on best practices when reporting and sharing results, here is another video from our predictive modeling education series.

So what do you find are your challenges in using data to tell a story?

Decentralize analytics.
Harness the power of many.

Create and share reports and datasets across the enterprise, and put analytical power in the hands of everyone. Veera creates a truly data-driven culture. Try it for yourself today.



Decentralize analytics. Harness the power of many.