Q&A: How Data Visualization Boosts BI's Value
The visual representation of business intelligence is making BI more accessible to a broader audience, says industry analyst Lyndsay Wise, who discusses where data visualization is headed and how to get the best value from it.
- By Linda L. Briggs
- April 2, 2013
"Even though there is plenty that users can accomplish now using data visualization, the reality is that we are just at the tip of the iceberg in terms of how people will be using this technology in the future," says industry analyst Lyndsay Wise. In this interview, she shares her views on where data visualization is headed, and how to get the best value from it now.
Wise is president and founder of WiseAnalytics, an independent analyst firm and consultancy specializing in business intelligence for small and midsize organizations. She also conducts research, consults, writes, and speaks about implementing BI and improving its value within organizations. Wise is author of the new book, Using Open Source Platforms for Business Intelligence -- Avoid Pitfalls and Maximize ROI, a guide to implementing open source BI solutions that maximize ROI.
BITW: What are some trends in data visualization right now?
Lyndsay Wise: There are three key trends I keep seeing: self-service interaction, collaboration, and predictive analytics.
Let's start with self-service interaction. What's happening there?
Software vendors are attempting to move more toward "self-service" in order to provide better ease of use and autonomy for users. The level of self-service -- or what that actually translates to -- will differ depending upon the targeted user groups. Overall, however, there is an ongoing attempt to make data visualization tools easier to maintain and customize without having to rely exclusively on IT departments for continual changes or customizations.
What about the second trend, collaboration?
As with self-service, I see collaboration becoming more commonplace. In addition to sharing views, more solutions from vendors are adding features such as annotations and other methods of making notes and commenting on information. That information can then be shared within the organization. Because of the proliferation of social media, and because data visualization is focused on easily digestible information, solutions are starting to incorporate similar functionality to enable broader interaction among users.
The third trend you mentioned, predictive analytics, seems to be all over the place in BI these days.
Yes. I include predictive analytics as a key trend based on the fact that many organizations have high maturity levels in their overall BI use. This means that they've exhausted the value proposition associated with historical analysis of the data. Now, they want to be able to plan and forecast into the future; due to advancements in technology, they are now able to do so without having to be a statistician or financial guru.
A number of analysts, including you, have mentioned the current importance of data visualization. Why is that? Data visualization has actually been around for a while. What's happening with it now?
When organizations approach me for consulting, more often than not they ask about dashboards. As we talk, they realize that they may or may not need a data warehouse or broader data management infrastructure, but data visualization seems to be the new entry point within BI. I think a lot of this has to do with the visualizations themselves as well as the hype associated with the visual representation of BI. That visual representation is making BI more accessible to a broader audience, thereby increasing the perceived value of business intelligence.
In addition, vendors such as Tableau and QlikView have done a great job at marketing, meaning that more people are aware of dashboards and see that they are easier to use than traditional OLAP cubes, parameterized reports, and the like -- and are, therefore, more attracted to their use.
Where are we headed with data visualization? Are we just in the beginning phases of what can be done with the technology?
I think that technology is now finally keeping pace with what organizations require. What this means is that analytical databases, and the data warehousing world in general, can meet diverse data access and analytical needs. On the one hand, this helps with issues such as data access and latency. On the other hand, this means that data visualization is no longer limited by access to historical data only.
In addition, newer interactive technologies such as GIS, 3D, and others will allow the data visualization market to become more interactive. I've seen vendor labs exploring additional ways of interacting with visual data, such as eyeglasses that identify the information a person is viewing, then automatically display complementary data. Even though there is plenty that users can accomplish now using data visualization, the reality is that we are just at the tip of the iceberg in terms of how people will be using this technology in the future.
What methods are available beyond dashboards for the display of visual data to users?
As I mentioned, many organizations are looking at dashboards as their primary information access point in relation to analytics. Now that so many types of visualizations exist, people are also looking at embedding visualizations within reports, as well as making reports more interactive with spark lines and other types of charts embedded within the report itself.
Data visualization makes insights easier for users on the front end, but what about the back end? How difficult is it to connect to live data, set up dashboards or other methods, test your creations, and so forth?
That's a great question, especially because many business users tend to download a tool from the Internet, get some quick dashboards up and running using data from Microsoft Excel, then think that a broader implementation will be just as simple. That's rarely the case.
Connecting to live data can be easy, providing connectors exist and the information required is formatted correctly already. However, many tools are limited in relation to consolidating data or ensuring that a level of data quality exists, meaning that more complicated dashboards may still require a staging area to get valid results for more complex data sets.
Realistically, implementation times will differ depending on the solution used, how broadly it is being deployed, and how complex the data is. This will also affect testing and any reiterations of dashboard design required. Overall, depending on the data required, organizations looking at a departmental deployment should realistically set aside two to three months to get a solution up and running. This includes developing the dashboard, integrating the information, managing complexities, testing, and acceptance.
Are there specific kinds of presentation methods that are better than others, like a scatter chart or bar chart versus the much-maligned pie chart of old?
I do think some presentation methods are better than others, but the method chosen depends on what you hope to achieve. Many traditional data visualization solutions are moving away from dials and gauges (picture a car dashboard) and towards scatter charts, heat maps, and other more interactive and straightforward visualizations.
In addition, many people prefer bar charts in which current information along with goals and projections can be viewed in a single visualization. Selected visualizations will continue to be the choice of designers and end users, but the main thing to remember is to "limit the noise" on any given page. To do that, select the top five metrics required and make sure they are addressed adequately; avoid filling the page with too many types of graphs.
In addition, many dashboard solutions have built-in parameters that select the best choices for the data on hand. I admit that I like heat maps because they show different levels of detail (such as sales and profit) represented in different sizes and colors, making it easier to decipher the data.
Additionally, GIS-related data is becoming much more robust for geographic-related data, expanding on the types of analytics available.
Should companies looking for a data visualization tool hold off until the market settles down?
Actually, now is a great time to get into the market. Even though new solutions and capabilities are constantly being introduced, the pace of acquisitions has slowed. Organizations can still get tremendous value out of the tools that currently exist. In addition, by being a customer, organizations are participating in the marketplace, and are more likely to have a say in new features added over time. That can help to tailor a solution to their specific business needs.
A bigger issue is that many organizations start off looking only for a data visualization solution without considering what else they need to consider, such as real-time versus trends-based approach; whether they will be drawing on multiple data sources or just Excel; or the needs for common ODBC connections.
In general, as long as a solution can meet the business and technical needs of the client, there's no sense in waiting -- after all, this technology will be in a state of continuous advancement for many years to come.