Visualizing BigQuery Data

"The purpose of visualization is insight, not pictures."

―Ben Shneiderman

Access to data ensures that we have a part of the story. We use this data to learn about some intricacies with respect to the performance of a website, marketing campaign, CRM program, or whatever we have data warehoused for. With that said, data itself can only answer some of the questions that most analysts will have. Most of the time, an analyst will need additional help to find the answers they are looking for (or at least to efficiently find those answers). Every analyst dealing with data should strongly consider adding data visualization to their arsenal of tools.

Why is data visualization important?

Julie is a data analyst working for a sporting goods e-commerce site. She has responsibilities that include reporting on the site's performance. The site includes multiple site categories, such as menswear, womenswear, as well as sporting equipment and shoes. Each category requires separate reporting as well as analysis on an ad hoc basis. Julie loves her work as she has been afforded the opportunity to monitor and analyze datasets of extreme size and complexity. This, however, is also one of her biggest challenges as she has a hard time understanding data of such size and complexity. This is one of the most important challenges data visualization tackles.

By adding data visualization to her repertoire, Julie can turn her unwieldy and static text data into several charts that both drive action and convey the underlying data in a much more...

The danger of summary statistics

When performing analyses, most analysts will use what is called summary statistics. Summary statistics is defined as descriptive statistics that are used to summarize a larger set of observations. These types of statistics are usually used because the larger set of observations is too large to analyze efficiently. For instance, an SEO analyst might use the average click-through as a measure of performance for their SEO efforts on two separate days of analysis.

The analyst finds that the average click-through rate for both days was identical at 8.4. However, here is what is revealed when visualizing the data:

The analyst was able to see how the data was extremely different when visualized rather than just summarized.

Making data visualization work for you

Here are a few tips for making actionable and efficient visualizations:

• Consider your audience:
  • Keep the decisions made by each of your audience in mind. A paid search analyst will want to look at the differences in ROI for different paid search campaigns, while a social media manager will want to know which posts drove the most visits. If you are unaware of the type of data desired by the end user, set some time to discuss the goals of the project.
• Choose chart types wisely:
  • Be mindful of scales for line and bar charts.
  • Pie charts and 3D charts should be avoided. In most situations, pie charts can be replaced with bar charts.
• Show only what is important:
  • As initially stated, visualization's focus is to drive action. Do not confuse the viewer with overkill, either by too many charts or too many elements within a single chart...

Three tools for visualizing BigQuery data

In this section, we will discuss the various tools for visualizing data and their features in detail.

Simple yet basic – Google Data Studio

• Cost: Free
• Difficulty: Easy
• Flexibility: Low

Google Data Studio is Google's main tool for visualizing data. Data Studio can be used to pull data directly out of most of Google's suite of marketing tools, including Google Analytics, Google AdWords, and Google Search Console. Data Studio also supports connectors for database tools such as PostgreSQL and, of course, BigQuery. Google Data Studio can be accessed at datastudio.google.com.

...

Summary

Visualization is a tool used to help the viewer understand data that simple text or numbers in a table can't convey. Also, visual data is much more quickly processed than text (15x faster per an MIT Neuroscience study). It is hard to argue that visualization is not a technique that should be used by each and every data analyst. BigQuery, in conjunction with visualization tools such as Google Data Studio, Tableau, or the R programming language, can be used by analysts to help convey meanings in data that might otherwise be overlooked, undermined, or misunderstood.