Validating Data Visualizations
Monday July 1, 2019 0 comments
By Sophie Auctor
Humans are predominantly visual. That’s why we say a picture is worth 1,000 words; we can often extrapolate more from a single image than from an entire essay trying to explain the same content. There’s another popular saying that further reveals the human tendency to trust only what their eyes observe: “You have to see it to believe it.”
Data visualizations turn stored data -- which, in its raw form, is complicated to interpret -- into a visual representation people can more easily understand. Rather than trying to make sense of a table or data set, people can look at a data visualization model, like a chart, to determine what the data’s saying.
But every data visualization is not necessarily created equally, and organizations looking to equip their employees with the best information for decision-making are increasingly seeking interactive models rather than the static ones of years past. There’s also the issue of trusting data visualizations: How can employees feel confident in the source data and the chart they’re seeing -- confident enough to factor it into decision-making that will affect company outcomes?
Producing Reliable & Understandable Data Visualizations
According to InfoWorld, organizations must ask these two questions: Are the data analytics showing results that make sense to subject matter experts? Is the data accurate and are the analytics valid? Basically, data visualizations must be understandable for end users and trustworthy in order for companies to reap their benefits.
Legacy business intelligence (BI) systems typically relied on data analysts to query data, generate reports and create visualization models. But there were a few shortcomings with this model, namely that analysts are well-versed in data but not necessarily well-versed in the information they’re working with. So, though a marketing manager or executive would be the one interpreting the visualizations and factoring them into decision-making, the data specialist was the one creating them in the first place.
Self-service analytics tools like ThoughtSpot remedy this disconnect by putting analytics directly into the hands of those actually using the insights. Now the person asking the question is the one receiving the answer and interpreting the accompanying data visualization. Users also have the ability to continually drill down into visualization models to keep exploring and asking questions as they go. This is a good thing -- provided this non-technical user can fully understand the chart in front of them. Interactivity helps in this sense because it allows users to continue to ask questions and drill down to see where information came from and what it means. Charts help turn dense information into interpretable insights for a wide variety of business users.
Then there’s the matter of producing valid insights in the form of data visualization models. Employees must be able to depend on the source data and the charts in front of them to make confident, accurate business decisions. Allowing users to go under the hood and understand exactly how their BI platform generated the visualization is key in this regard.
Tools/Services to Turn Visualizations into Value
Quality data visualizations depend on solid data sources, or else you’ll only end up with a visually appealing graph depicting potentially inaccurate information. Furthermore, in order to get everyone to be on the same page, your company’s BI solution must operate on a single version of the truth. As one Forbes contributor writes, “Having multiple versions of the truth can lead to confusion, paralysis and bad decision making.”
Only a data analytics platform of consolidating multiple disparate data sources into one version of the truth is capable of producing data visualizations that are truly valid for decision-making throughout the entire company. Business users need to be able to trust the data visualizations they’re seeing because they trust the underlying BI software and company data.
Data visualizations only provide the valuable benefit of speedy, simplified data interpretation when everyone’s speaking the same data language to start. It’s also increasingly important to harness the power of interactive charts so end users can keep drilling down to get answers and perspective.
Beyond this, it’s imperative to foster a data-literate culture in which all users feel comfortable working with data visualizations as they make business decisions, driving better outcomes and more value for the organization as a whole.