Where business intelligence (BI) tools can take huge swaths of data and parse that into digestible data points, data visualization is the presentation portion of that equation.
Think of it as the pie chart function of your favorite spreadsheet, only much more powerful.
The purpose of such imagery is to quickly transfer information from the machine to the human brain, not only efficiently but also in the most meaningful manner possible.
Therefore, it is not the aesthetic value of a visualization that counts but the clarity of the message it conveys.
However, the conciseness necessary for clarity does not preclude complexity in the message.
Since much of the information humans must consume is complex and nuanced, data visualizations are configured alone and in groups to tell a larger story through images.
An example of a single configuration is any visualization that reveals more granular or related information when the viewer clicks on or performs a mouseover on a section of the illustration.
Examples of group visualizations include just about every BI app dashboard ever made.
Indeed, data visualization is such an integral part of self-service BI tools that the tools to make and publish them largely share common feature sets.
As expected, in our recent review roundup of the best self-service BI products, we found the vast majority to be capable of data visualization operations.
However, customers looking to really exploit data visualization should look at these tools carefully and exclusively through that lens before making a buying decision.
After all, sometimes the right tool to parse your data may not offer a sufficient visualization palette for your needs.
For example, you may want the ability to build a custom infographic or create interactive visualizations, but not all BI apps provide those options.
You may need to invest in a combination of tools to get both the analytics and the visualization tools you need.
What Is Data Visualization?
In short, data visualization is a visual depiction of information.
It is imagery dedicated exclusively to messaging or presenting information.
Data visualization tools can automatically create visualizations, enable you to create your own, or offer both capabilities.
At the lower end are simpler and even free data visualization tools dedicated to building infographics rather than performing sophisticated data analytics.
Some of these tools include Tableau Gallery and even Microsoft Power BI. In January 2018, Tableau introduced a new data engine called Hyper that the company claims gives users up to five times faster querying speed over previous versions.
Meanwhile, in July 2018, Microsoft rolled out new features for Microsoft Power BI, such as integration of Big Data directly into the Power BI web service.
At the higher end are tools that can change visualizations on the fly, in the same way that outputs from sophisticated algorithms change after repeated direct querying of real-time data (i.e., streaming data) and across multiple data sources.
The tools occupying the middle of the spectrum do not represent real-time data but still produce visualizations from advanced analytics outputs.
The self-service BI apps we reviewed contain average to higher-end visualization tools. Some of the tools contain strong natural language query capabilities like Sisense, and others bring real-time analytics for the Internet of Things (IoT), like SAP Analytics Cloud. In short, you cannot judge the quality of the underlying analytics engine by the cover of its art package.
Some very powerful analytics come with pitiful to passing visualization capabilities.
Conversely, some pitiful to passing analytics come with some pretty impressive visualization features.
Since we originally reviewed these BI tools, IBM has discontinued offering IBM Watson Analytics for purchase.
Instead, IBM introduced Cognos Analytics 11.1, which offers guided data discovery, automated predictive analytics, and the ability to interact with data conversationally.
There is a wide range of art depictions that data visualization tools can create.
Some depictions are simple, some are complicated.
Some are beautiful, some are crude.
And there are some that are truly individual creations.
But most spring from templates in the traditional forms associated with statistics.
The simplest examples of data visualization are the pie and bar charts you've been able to access via Microsoft Excel for many years now.
But as BI has matured as a platform, so, too, have the options available to you for seeing your data and presenting it to others.
The tools we review here reflect the medium to higher end of the spectrum in BI; they're capable of performing sophisticated queries without the need to understand Structured Query Language (SQL) coding.
Plus, they can render analytics in a wide variety of visual formats—going far beyond the basic bar chart to include geographical mapping, heat maps, sparklines, and even more specialized visualizations such as the spider chart below.
Data visualization is not a new concept.
Pie charts and bar and line graphs have existed throughout the ages.
What's changed are the kinds and size of data that can be represented this way, and the many more sophisticated ways in which you can show it and share it.
The Importance of the Dashboard
Ultimately, data visualization capabilities are used to build dashboards.
Sometimes the dashboard represents a single, data-based story that is significant to many viewers.
Or the dashboard may contain many stories for the benefit of a single user.
Dashboards sometimes come with visualizations that are preset and fixed in place.
Other times the dashboard's visualizations come with various display options or images that are customizable.
Sharing can often be customized too, such as according to permissions, per business line, per job role pertinence, or even by personal preferences. In any case, the dashboard typically contains two or more data visualizations meant to inform and sometimes even prompt a business action or decision.
Prior to the advent of self-service BI tools, executives had to present their questions to a database professional who would then try to understand it as best he or she could, write a SQL query, and representing that question against a database or data warehouse.
The result would be fed to an IT person who would then write the necessary code to represent it as a dashboard on the executive's team website, on a shared app, or even just as a standalone document the executive received via email.
If more than one data source was needed, then very often more than one database professional had to write separate queries (which then had to be melded together offline).
At the end of this inefficient and multistep process were analyses.
You got historical analyses (i.e., information after the fact rather than in real time).
These reports usually arrived too late for the business to change or influence the outcome of the activity it depicted. Thus, business analysts, department heads, and C-suite leaders typically received reports with delayed, overly simplistic, and vague information.
Sometimes the information was irrelevant when it finally made its way to business analysts or the C-suite because the company had changed direction or other factors emerged in the meantime.
Even so, dashboards and reports made in this way rarely changed.
Things proceeded as they always had: the same questions asked, the same data queried, the same reports and dashboards generated—day after day and week after week.
By contrast, today's self-service BI apps let business analysts bypass the middlemen and unstop many of the IT bottlenecks.
This self-service software also enables the use of data outside the company as well as from within, such as social media, the cloud, public data sets, and IoT data.
Some self-service BI apps can use real-time data, but many are limited to near-time data (frequent refreshes).
However, near-time data usually isn't a business limitation.
There are actually only a few use cases where real-time data analysis warrants the extra effort and expense.
After all, near-time refreshes can be as frequent as every minute or less.
With regards to self-service BI dashboards, the key value is typically threefold:
First, they don't require database expertise to use.
You'll probably (though not always) need your database professional's help to set them up and connect them to all of the data sources you need.
After all, compliance and security issues still remain.
IT usually gets involved at least to the point of resolving those issues, determining who gets credentialed access, and how much data they can see.Once that's done, these tools provide varying degrees of simplicity when it comes to writing your own queries.
Some still work best if you know some SQL, but others work entirely using natural language syntax, rendering SQL knowledge unnecessary.
However, most do require a good understanding of statistics.
This necessity is not strictly from an operational standpoint, but because errors can be made in the interpretation of the outputs if the user lacks a basic understanding of statistics.
Just because the software made you an excellent visualization of the machine's answer does not mean that you asked the right question.Second, almost all of them can act as a unified front end to multiple databases and data types.
This is primarily due to the rising popularity of Big Data, which is typically a combination of relational data (generally SQL-based) and unstructured data found in disparate sources both inside and outside the company's walls.
By providing support for various kinds of data, these tools allow folks without database expertise—but with direct, front-line job experience—to ask questions directly against the organization's data.This can provide immediate payback against fast-growing Big Data stores.
It also enables new insights and ways to leverage data, which might otherwise be lost when those questions percolate through data scientist and IT professional filters.A single query can span multiple databases and data types in record speeds, and the tool will take care of building the visual representation, too.
In short, a team of data scientists is not required.
That's not only faster but it's orders of magnitude easier.Third, these tools can also build live data visualizations and dashboards themselves rather than forcing a separate operation from your company's programmers or IT staffers. Those visualizations can be exported as flat graphic files or as code snippets that you can just copy and paste onto webpages or team websites.
Dashboards can also be directly shared, oftentimes even with users who are not using the BI app.Integrating them with other apps is usually easily done through connectors, depending on whether or not the self-service BI app you are using has a connector to the app on which you want to share the dashboard.
Some will still require some IT assistance, but even there the time required to perform the integration is often reduced versus starting from scratch with just a series of SQL queries.
Those code snippets will also do more than simply render a visualization; they can also maintain their connections to the live data sources referenced in the query.
This lets them change on the fly as source data changes—the primary function of any dashboard.
That certainly goes far beyond what you can get through a traditional spreadsheet.
The good news is that even some spreadsheet software such as Microsoft Excel now includes data visualization capabilities.
These tools can bring businesses of any size fresh perspectives on their data quickly and easily.
Given that most businesses are being inundated with new data from all directions, a fast path to return on investment (ROI) is often reason enough to justify a self-serve BI or data visualization software purchase.
What to Look For
Once you've made the decision to invest, you'll quickly realize that not all data visualization tools are created equal.
You'll realize they also tend to focus on different aspects of data interaction.
So, to find a solution that will fully meet your needs, you need to evaluate your selection carefully in terms of features and capabilities.
First, check carefully into the kinds of visualizations a tool supports.
Compare that not just with the kinds of data your organization collects, but with how your company likes to consume that data.
Grab a free trial and experiment with new visualizations.
Many companies have standardized on a certain method of looking at their key data.
Make sure any new tool can render data that way, and take the opportunity to try some new visualization methods.
You may find that a new view unlocks new insights.
Second, find out exactly which data formats the query tools supports.
There should be a long list that includes not just basic data formats such as SQL and NoSQL...
