Data Visualization

Data visualization

Quick Definition: Data visualization is the practice of visually representing data to aid in decision-making.

Key Takeaways:

What is data visualization in marketing?

Data visualization helps marketers visualize trends, patterns, and oddities in data via graphic representation.

Representing data graphically means that data comes across more viscerally. Ideally, that leads to better decision-making moving forward.

What is the data visualization process?

Specific technology, either used alone or as part of multiple solutions offered by a vendor, is used to visualize data.

Data visualization is a layer in a technology stack. You start with the data collection or warehousing layer. After that, you've got the data transformation layer, where you take the collected data and give it shape and meaning, and then put it in a usable format. The data analysis layer is where you're slicing and dicing and pivoting data tables, and then the visualization layer is what sits on top.

Every layer has a set of vendors that provides technology for it. And companies like Adobe, Oracle, and Salesforce provide multiple solutions up and down the entire tech stack. For example, Adobe offers Adobe Experience Platform as the data warehousing component, Adobe Analytics as the data interrogation component, and then Analysis Workspace, which is part of Adobe Analytics as the visualization component.

Typically, in the industry, you'll see a data lake combined with an analytics solution like Google BigQuery or Azure from Microsoft. At the top of the stack, businesses will use something like Power BI as the data visualization component. Within the stack, you have solutions like Domo and Tableau.

What is the purpose of data visualization?

The ultimate purpose of data visualization is to help marketers make business decisions using insights from large amounts of data.

Because of the large quantities of data necessary to make business decisions today and because it’s challenging to understand raw numbers, humans need visual representations to make sense of big data.

People consume raw data captured in tables. With data dumps or spreadsheets, it’s hard to find meaningful information or identify meaningful patterns. The difficulty of reading raw data is compounded by the fact that in enterprises today, you’re dealing with millions, billions, and potentially even trillions of rows of data. It would help if you had something that's able to go through the data, summarize it, and extract those patterns to create a visual summary that explains and clarifies the data. Using data visualization, decision-makers can understand what's going on and start to dive deeper into the information.

It's one thing to be presented with a table of data around advertising keywords and their associated impressions of view-throughs and click-throughs. When you can view that table of data in an information visualization like a heat map visual, in a matter of seconds you can identify which campaigns are driving better returns.

What is the business value of data visualization?

Data visualization serves as the basis for decision-making. It provides a single source of truth for an organization. If you have an approved, standardized set of data visualizations, or information you’re pulling from a data visualization tool, everyone can agree with those numbers. You can avoid political arguments around the best way to align budgets, for example.

Data visualization should introduce a level of precision and objectivity into the entire decision-making process across the organization, from mid-level campaign execution to top-level strategy.

What are the current shortcomings of data visualization?

One disadvantage of data visualization is that the current technologies available are not interactive. Data visualization tools like Tableau and Power BI are interactive in the sense that they have drop-down menus and the ability to customize colors, but they can’t continuously query. If you see something interesting in the dashboard and have more questions, you can’t go in directly and find out more information. You have to write a new SQL query and wait for it to process and populate the dashboard.

If a business user, like the CMO of a company, has a follow-up question about why campaigns are failing or which regions are showing campaign success, it could take up to several weeks to process. And only employees who know how to write SQL queries can do the work. The ability to provide self-service in data visualization and interactively query data in real-time is virtually nonexistent.

Data visualization, like any other data usage practice, is susceptible to manipulation. You'll have certain people or groups that may want to manipulate the data to tell a preconceived story. In the end, the visual is only as good as the data and the way that data was processed. If you have a preconceived notion, you can twist a data table or a visualization to match a biased analysis. It’s important to not take a visualization at face value but to continue to ask critical questions, even if the pattern appears to make sense.

How has data visualization changed over time?

Ten to fifteen years ago, business intelligence was focused on SQL queries. The analyst would ask for specific information, and the query would return results. There were tools like Cognos that had basic visuals, but they could not customize and make those visuals meaningful. Over the last few years, a new industry has emerged to address this concern. Visualization is decoupled from the query language and solutions like Tableau can connect to the data warehouse and automatically visualize queries.

What is the future of data visualization?

Interactivity, self-service, and data access must be the next frontiers for data visualization. Currently, any user can log into a dashboard and view the data, but not everyone can interact with the data. The future of data visualization will allow users to update the processing in real-time, without wasting days of latency.

The big opportunity for improving data visualization is in rationalizing data and integration of the data itself. For example, with a retailer like Target, the customer has multiple ways to buy a product: they can buy it online, in-store, or over the phone with a service rep. And each one of those channels generates its dataset. If you use a data visualization tool, you won’t be able to do a comparative analysis of the channels unless the data is integrated. You can look at the revenue from each stream individually, but you can’t get a full understanding of how the business is performing.

Vendors need to figure out a better way to integrate data on the back end, to make it easier to consume and interact with data visualization on the front end. In theory, you can do it with data science and SQL, but that does not apply to the masses — it’s too highly customized. An organization's going to be paying huge amounts of money for data scientists to figure it out. And it's just not a practical solution. Integration and flexibility within new technologies will go a long way in democratizing data visualization.

How can companies avoid potential issues?

Data governance is key. The more you get things right on the data ingestion side with categorization, integrity, security, and so on, the fewer problems you’ll have as you go through each layer of the technology stack. Many organizations put off proper data governance because they get a large lift and decide to focus on managing it later, but that never works out. Problems that arise include the fact that people on the analysis side will be constantly doing custom work, the data could mean multiple things, and even data values could be radically different from one report to another report, if you're sampling data or doing something similar.

It's worth the upfront investment to standardize data and have rich data governance. It will take longer to implement, and when people get new technology, they usually want to use it right away. Spending a few months upfront to get things ready won’t be a popular decision, but the payoff is huge. You will experience massive time savings down the road as you collect new data because you will already have a standardization model to fit the data into.

How does an organization choose which type of data visualization to use?

In some cases, the data visualization you choose will depend on your vertical industry. A B2B company’s sales cycle will look radically different from a B2C company’s sales cycle, for example. If a B2B company focusing on in-person sales or new contracts wants to use an attribution visual, they might want to choose a W-shaped attribution model that assigns significant value to the first and last touchpoints before conversion, and also emphasizes the mid-funnel touchpoint.

In general, however, factors other than industry have more weight when it comes to deciding how to depict the data. One factor is you’ll use the information you’re trying to communicate. If you're looking for trends, a line graph will likely be a better option than a chart like a scatterplot, as the line will indicate the trend.

The choice of data visualization type also depends on the audience. Knowing your audience is a crucial part of deciding how to design your ultimate visualization. For example, an executive audience is typically concerned about the facts and prefers simple visuals with only a couple of points. A sales audience, on the other hand, will probably prefer a more creative representation. Understanding who your audience is and what they're looking to see is a critical element that you need to know before you begin pulling together a chart or graph.


Are specific types of data easier to visualize than others?

Structured data is easier to visualize than unstructured data because the data already fits into a defined model.

More established types of data will also be easier to visualize at first because more time has been spent identifying best practices. Data from traditional sources or channels will generally have existing models that have proved successful, while data from cutting-edge sources will require trial and error.

How do organizations visualize data?

There is an almost unlimited number of ways to present data visually.

Vendors are continually brainstorming new methods of visualizing data. The goal is to make it progressively easier for individuals to look at a visual and clearly understand the story being told by the data.

Options for data visualization range from charting data points on a simple bar chart or line chart to designing highly detailed infographics. Common visuals include heat maps, geography maps, pie charts, and scatter plots. Because the point is to make the data easy to read, organizations frequently customize the visuals to make the data appear as visually appealing as possible.

People also view