The age-old problem with data analysis is making the best use out of the information obtained by carefully parsing it for conclusions. It’s not an easy task, so there’s a reason data storytelling has become such a popular and lucrative career. Separating the wheat from the chaff is a fine art honed by extensive experience.
SEE: Big data management tips (free PDF) (TechRepublic)
I spoke with Keelin McDonell, general manager of business intelligence and integrations at Narrative Science, an artificial intelligence (AI)-powered software startup that turns data into stories; and Jolene Wiggins, CMO of Gravy Analytics, a data analysis organization. Bill Hewitt, CEO of Aternity, a digital experience management solutions provider, added some thoughts to the conversation, too.
Scott Matteson: What are the challenges companies face in their ability to quickly act on data?
Keelin McDonell: Companies are dealing with more data than ever before. The size of our data universe doubles every two years. We’re now sifting through so much data that it’s almost become meaningless because companies today don’t have the context they need to understand what their data is telling them.
Companies need to act on data faster than ever before. Data depreciates fast, and everyone in the business—from data analysts to sales, marketing, and customer success teams—needs to be able to receive, understand, and act on insights from data in real time. This lets them get ahead of their competitors and stay nimble in a landscape that’s always changing. Spending too much time puzzling over charts, graphs, and other data visualizations is time that could be spent making the next major business decision to get ahead.
There are a number of business intelligence tools that have tried to tackle the data problem (the market is worth about $30 billion and grows 15 percent each year). But many of these tools are designed for people with data analytics backgrounds, so they’re not easy to use for people in other departments who rely on data to make major business decisions every day.
SEE: 4 tips for using data visualization in a board presentation (TechRepublic)
According to Gartner, at a typical company these data analytics and business intelligence tools only have about 25 percent penetration, suggesting that three-quarters of employees find them too difficult, too time-consuming to use, or don’t have the skills to use them at all.
Jolene Wiggins: The biggest hurdle between data collection and analysis that keeps companies from acting on data in a timely manner lies in the organization’s data structure. To get the most holistic view, companies need to pull data from several internal and external sources, which can be a very time-consuming and tedious process made more complex by different data formats and management systems.
Scott Matteson: How can those challenges be addressed?
Keelin McDonell: We think the easiest way to help companies act faster on their data is by presenting it through stories and language. That means providing plain-English stories about what the data is telling you, as opposed to a bunch of scatter plots and pie charts.
There are a variety of benefits to doing it this way.
- Meet and exceed goals faster. Because you can devote more resources to where they will have the biggest impact, and ambitious goals become more realistic.
- Democratize data for the entire company. By presenting data in the form of a story, literally anyone in the business can understand what the data is telling them without having to pore over complex charts and graphs. What’s more, research shows people remember information better when it’s in the form of a story.
- Make decisions faster. When you know exactly what the data is telling you, you can confidently make major business decisions without having to worry if you’ve read a chart or dashboard wrong. Meetings can be spent discussing what really matters as opposed to asking the room to read your pie chart.
- Get everyone on the same page. Charts and graphs are open to interpretation by the person reading them. By presenting data as a story, you reduce the chance that two different departments are arriving at two different conclusions from the same data visualization.
- Improve resource allocation. Spend less time reading data, and more time on tasks that move the needle, like drafting a new marketing email or putting more money behind a social media post.
Jolene Wiggins: Overcoming roadblocks to acting quickly on data depends on data integration: Combining data originating from different sources into a single location with unified processes. Also, companies need to fundamentally find a way to make data part of the culture of the organization at every level. This culture is much easier to cultivate when the collection and structure of data is unified across organizations, making it quick and easy for the right people to access regardless of team. When companies have the right systems in place to provide access to data, and the right resources to analyze and pull learnings from that data, then data can become central to operations and decision-making.
Scott Matteson: What types of responsibilities do data analysts possess?
Keelin McDonell: Data analysts collect information about the company’s current and potential customers and use it to draw meaningful conclusions about their behavior.
Basically, data analysts are responsible for telling you how customers are reacting to the ways your company interacts with them and why.
Data analysts are also increasingly responsible for helping their organizations understand the context of the data they’re collecting. For example, it’s more important for the sales team to know that sales have rebounded rather than risen 4%. They’ll forget the “4%” figure (and it’ll be different in a week, anyway).
SEE: Why managing data science projects is not the same as IT projects (TechRepublic)
Day-to-day, data analysts produce reports that detail trends with customer behavior and potential improvement areas for the company. They’re also responsible for identifying patterns and trends in the data, and then working with multiple departments inside and outside the company to exploit those trends. They also work with IT teams to set up systems to collect customer and company data.
Data analysts are arguably one of the most important roles at any company, because they provide access to the information that the company needs to make the major business decisions necessary to survive and beat the competition.
Jolene Wiggins: Data analysts, of course, need to have technical know-how and solid math skills, but they also have to be able to interpret the numbers to concisely tell a story. This means visualizing the data, which requires a bit of design and creativity, and explaining the data simply enough that the storyline is clear.
Scott Matteson: What types of skill sets are beneficial for data analysts?
The data analyst role used to be very technical in nature. But as data becomes even more vital to a company’s ability to remain competitive, data analysts are critical for helping the company make better-informed decisions. That means you need to influence people. Predictive models, line charts, and numbers don’t do that; stories do.
SEE: How to become a data scientist without getting a Ph.D. (TechRepublic)
Jolene Wiggins: Critical thinking is possibly the most important skill set data analysts should have. Data doesn’t always provide straightforward, cause-and-effect answers. Analysts need to be willing to look for alternate explanations and view findings with enough suspicion to dig a little deeper. Their main job is to interpret what data means, which requires them to walk a fine line between simply taking the numbers at face value and overreaching with assumptions.
Scott Matteson: What might a typical day for a data analyst look like?
Keelin McDonell: A huge amount of a data analyst’s day is spent reporting, or communicating to various stakeholders what the data means.
A lot of this reporting is ad hoc: Executives make one-off requests for information on how sales have performed in a specific time frame, for example. It’s the analyst’s job to provide those executives with those insights, which is important, but can also mean they are running from one menial reporting task to another all day.
A lot of that reporting can be automated, freeing analysts for more strategic, high-level, value-added work.
Jolene Wiggins: An analyst’s workload could look very different from day-to-day, with a mix of research for internal purposes and client-oriented analyses. They may be working on several different projects at the same time, requiring them to switch gears frequently. This diversity in projects can help spark creativity and overcome roadblocks—you never know when you’re going to learn something that can be applied to another unrelated project.
Scott Matteson: How can data analysts present the most meaningful information?
Keelin McDonell: To deal with the overabundance of data that companies collect, data analysts spend hours putting spreadsheets and bar charts together, then share that with sales, marketing, product ,and RevOps teams via dashboards. But the problem with this approach is that you’re leaving a lot open to interpretation, which can take up loads of time when trying to make a major business decision.
SEE: Why big data tracking and monitoring is essential to security and optimization (TechRepublic)
When thinking about what information is most meaningful, consider what information your colleagues need to take the next step.
Here are a couple examples:
A salesperson might see that sales have increased four percent, which is inherently good news. But the missing context is that sales have rebounded. Had the salesperson known this, they would have been able to better understand where to allocate resources.
The marketing team needs to know how well their channels are performing and the quality of business they’re bringing in, so they know where to increase and decrease the money they’re spending.
Marketing and sales teams need to know that their revenue is up. But it’s even more important for them to know that the number of licenses sold decreased, but the average deal size increased.
Jolene Wiggins: Analysts need to be able to concisely communicate meaningful data, so visualization, of course, is a great vehicle for doing that. But simply generating a chart or a graph isn’t necessarily the most effective storytelling tool. After thinking about what the data actually says, analysts need to present it in a format that allows their audience to follow the logic and grasp the story being told.
Scott Matteson: What is the purpose of a data-based story?
Keelin McDonell: A data-based story gives you crucial information about what’s happening at a business and why, in plain English. This helps anyone in the company better understand what the data is telling them and helps turn that into action. It helps data analysts become better data storytellers, too.
SEE: How to create your first Tableau Software data visualization chart (TechRepublic)
A good data story is just a story, the same as any other story you might read. It starts with a hook, is easily digestible and (perhaps most important) memorable. Ultimately, a data story helps everyone get a clearer picture of how the business is doing.
We believe in this approach so much that my colleagues wrote a book about it. It’s called Let Your People Be People. In the book, Nate Nichols and Anna Walsh explain how anyone in the business can tell a compelling story and the tremendous impact that a good one can have on your organization.
Jolene Wiggins: Telling stories based on data is becoming increasingly important in our current social landscape because verifiable data marks the difference between opinion and fact. Although there is still some reasonable room for error, data-based stories help remove human bias by quantifying past events to help predict future outcomes. The goal of most writing is to convince the reader—great data helps to strengthen the writer’s arguments and makes whatever is being said that much more powerful.
Scott Matteson: What should it contain?
Keelin McDonell: Telling your data-based story within the right context is one thing. But actually having it make sense to the person reading it is another.
Here’s what you need to know:
- Summarize your story first: Provide a clear summary of your findings by setting the scene. For example, instead of titling a report “Sales Performance Week of April 3,” summarize what the performance was like: “The First Week of April Was Top Sales Performers’ Most Successful Week of Q2.”
- Tell your story with words and visualizations: Describe why or how that week was successful with words that explain how that outcome relates to the goal for the initiative or project. Even if the result is normal, that could also mean that the result is above or below your company’s goals.
- Supplement your story with visualizations: Don’t assume that everyone will know what insights you want them to pull out of visuals. Supplement them with words so your audience has a clear understanding of what the charts, graphs, and data are telling them (and what you want them to know).
- Include only the details your audience needs: You should provide actionable recommendations for teams to implement; not unimportant, fluffy details. Even if you don’t have anything new to report, contextualize your findings. For example, “Even though we launched a new product last month, we have no new leads.”
Jolene Wiggins: Data-based stories should refrain from speculation and opinion—after all, correctly analyzed data is meant to represent verifiable facts. Add color to a story with specific anecdotal examples, but remember that one example is not necessarily representative of the whole. The best way to tell a data-based story is to keep it simple—don’t let the story get lost in the numbers. When storytellers overwhelm readers with numbers, the message doesn’t land. Remember: Use the numbers that are most relevant and necessary to paint an accurate picture.
SEE: How graph databases help analyze complex relationships (TechRepublic)
Bill Hewitt: Data storytelling is an integral function for executives as they demonstrate performance, drive business alignment, and communicate outcomes across their organizations and to executive boards. While there is a finite amount of ‘gut instinct’ that goes into managing a successful enterprise, qualitative data uncovers important insights that could influence how an organization allocates its budget, positions products within the marketplace, and even evaluates personnel.
This is especially true when discussing the value—or lack thereof—of complex IT projects and implementations. Many tech leaders struggle to communicate with executives because they’re focused more on output than business outcomes and insights.
For instance, it’s not enough to say that Office 365 was rolled out to employees. Instead, IT teams should be able to provide granular insights that demonstrate traction and ROI of their investment, including the percentage of employees using the new applications, whether productivity has increased, and whether the platform has had an adverse performance effect on the entire end-user infrastructure.
Additionally, it is critical to present the “so what” in the context of your business relative to best practices, industry benchmarks and the current business climate. Ultimately, context plus instructive data points help business leaders connect the IT team’s efforts with overall business priorities and performance.