Take a locomotive, for example. Every component on that machine is connected via sensors sending thousands of signals every second to inform the Uptake platform about how it’s performing. Those sensors then allow dispatchers to detect where the locomotive is on its mission, how fast it’s proceeding and whether it’s operating at full efficiency or showing signs of potentially dangerous and costly wear and tear.
All together, these millions of data points can tell a larger narrative about a rail network, enabling operators to identify ways to increase efficiency in the system, whether through savings in fuel, time or costs.
Let’s narrow in on a specific example of this in the field that we see every day: presenting machine fluid interpretation to our partners’ analysts.
If Uptake marked a fluid sample as anomalous without presenting it in a way that’s going to take the analyst through the story of how we arrived at that conclusion, the analyst could be lost and potentially default to disbelief or disagreement. If we provide the full story with the appropriate visualizations and context to support the outcome, we can reach stronger consensus and improve our analysis and predictive recommendations in the future.
Creating a narrative in a way that makes sense to our audience—analysts and other enterprise stakeholders—is critical to instilling confidence in our data analysis. If we don’t tell a compelling story, the analysts who consume our analysis can’t combine our study with their own intuition successfully; they won’t have good faith in the insights we provide, and they won’t come back for more.
Presenting data in a thoughtful way is important, but what are some data-storytelling best practices to follow? Drawing off of our experience, let’s take a look at four principles to ensure effective storytelling through reams of data.
Keep your story simple, but not too simple.
Providing minimal context around your approach, analysis and conclusions will put the onus on your user to fill in any holes and make assumptions. This is problematic because not only are you not giving the complete picture, your users might be making the wrong extrapolation. On the other end, too much information can overwhelm your user and make it difficult for them to parse. Again, there’s a delicate balance when it comes to story simplicity.
Create a conclusion for every analysis.
You can show countless numbers and stats to your user, but if you’re not drawing a conclusion, you’re not doing your job. Drawing a conclusion and explaining it to your user is going to bring everything full circle. If your user is taken on a journey and brought to meaningful conclusion that they can put to use, they will come back for more analysis.
Use the right visuals to support your analysis.
If you’re telling a story, you need to customize your visuals to be extremely useful in the context of your story. Depending on the data, you might need a particular visual to showcase it. Data scientists who are working with the data closely might lose sight of the fact that outsiders aren’t as familiar with the analysis and choose a visualization that doesn’t provide the right context. A rule of thumb is to provide a visual with enough explanation that you don’t have to be there to talk someone through the analysis to help them understand it.
Show your reader that your analysis is a sensible one.
Communicate your good methodology without going overboard on a full-blown scientific methods section. Users need an understanding of how the data was obtained and how you carried out your analysis. This is necessary because the methodology you choose affects the findings and your interpretation of them. An unreliable method produces unreliable results and puts the significance of your interpretations of the findings into question.
Remember that great storytelling involves practice and refinement over time. As you ask the right questions and understand your users’ problems, you’ll improve. With good storytelling practices in place, you’ll not only get better feedback and generate buy-in from your user, you’ll improve the quality of your analysis throughout your working relationship. Your user will see your value and continue to trust you as you solve their most meaningful and critical problems.