Creating Effective Business Intelligence
Much of the discussion surrounding business intelligence (BI) is focused on exciting new developments and technical details, so it can be easy to lose focus on the fundamental purpose of BI: to clearly communicate about data.
As BI is created and consumed by ever larger audiences, it’s important for those with BI experience to consider ways to create content that doesn’t require special expertise to understand. It’s also important for those new to BI to learn ways to effectively share insights based on data.
As you work on any BI project, keep its purpose clearly in your mind.
You should be able to state the purpose in a sentence or two. Usually, the purpose relates to either solving a known business problem or enabling the discovery of new opportunities.
Try to state the purpose in terms of what a document will enable users to do, rather than what information it will contain or what it will look like. The purpose should drive those decisions, and ideally be based on input from the target audience. The purpose will also help determine which BI tool to use.
Without clarity of purpose, you may be tempted to create a single solution to serve users who don’t share common needs, or to provide information that isn’t necessary just because it is available.
In SAP BI tools, the statement of purpose you create to guide you as you create a BI document can also help users find and interpret it. Save the statement in the document description so users can read it as they browse content in SAP BusinessObjects BI launch pad or Mobile. Words in the description also function as keywords in content searches.
Above are two tables showing sales results for a number of store locations. The table on the left has all the same columns as the one on the right, along with several additional ones. Which table is more useful?
In BI, more often doesn’t mean better.
If the purpose of the tables is to help users to assess the overall performance of different store locations, the additional columns about products in the table on the left are irrelevant. The presence of these columns means that the sales revenue results are calculated by product rather than by store, so users are left with additional work to do before they can get the store performance information they actually need from the table.
By including information that isn’t relevant, the chart on the left invites users to spend time considering information that doesn’t matter and fails to directly answer the question that does matter.
In SAP BI tools, you can add interactive features such as filters, input controls, prompts, and sections to documents to enable users to view only the results that are relevant to them.
If you were to shuffle the paragraphs in a story, you probably wouldn’t be surprised to find that readers have trouble understanding it. The same thing can happen when the elements in a BI document are not arranged with care.
A well composed document should tell a story or make an argument. Elements should flow from one to the next.
For example, in the dashboard shown above, the panel at the bottom shows the elements that comprise a company’s revenue and expenses, and how these numbers relate to net income. The sliders at the top facilitate what-if analysis, enabling users to explore how changes in any revenue or expense category would affect the results in other areas to result in a change to Net income. Without the grouping of items at the top and bottom into revenue and expenses, and without the flowchart showing relationships between various revenue and expense items and net income, the entire dashboard would be much more difficult to interpret.
SAP BI tools such as Crystal Reports, Web Intelligence, and Dashboards enable extensive customization and were developed for the creation of content to be shared, but newer data discovery and exploration tools such as Lumira and Explorer also have features that enable the creation of carefully designed documents to share insights. Exploration Views in Explorer allow you to arrange multiple visualizations and interact with them all simultaneously. Lumira enables you to create multi-page interactive boards and infographics to tell the story behind the data.
As you organize a document, you need to consider the elements that you want to enable users to compare. Elements intended for comparison should be close together.
For example, in a column chart, the easiest columns to compare are right beside each other. When columns are arranged in alphabetical order, as above, it is difficult to compare the performance of any but the top product lines. Sorting by revenue makes comparison much easier, as shown below.
The need for comparison between charts is also an issue. If two charts can’t be viewed simultaneously, such as when they are on different tabs of a dashboard or different pages of a report, it is much more difficult for users to effectively compare them.
You can help to enable users to make comparisons between related elements by using consistent chart types, scales, and formatting.
Above are two charts showing sales results by product category at two different store locations. It is not clear that the charts are meant to be compared because they use different chart types, are reproduced at different sizes, and have different units of measurement and color schemes. In the example below, the matching formatting of a series of charts in a Crystal report makes it clear that the charts are meant for comparison.
Visualizations such as sparklines in SAP BusinessObjects Mobile and trellis charts in SAP Lumira enable multiple visualizations with identical formatting to be displayed in a limited amount of space, making comparisons easy.
The prominence of an element in a BI document should relate to its importance. For example, you might place the most important chart first to ensure that it gets appropriate attention. Within a chart, the most important elements should be the ones that stand out the most.
You can draw attention to important elements by displaying them in a larger size, using bolder headings or colors, or by placing them in more prominent areas of the screen, such as the top left rather than the bottom right.
It can be tempting to use color to spice up a document simply for visual appeal. Unfortunately, using color for strictly aesthetic reasons can reduce a document’s communicative power rather than enhance it.
Colors should generally be neutral, with brighter or darker shades used to draw attention to important information. You can use different intensities of the same color to show results for the same object through time, or to convey more positive or negative results.
A change in color should indicate a change in meaning. For example, the chart above shows sales revenue by store, with each store displayed in a different color. The colors don’t communicate anything meaningful. When the same color is used for all stores, as in the chart below, distractions are reduced. Color can then be used as a communicative element, to highlight the results for a store that failed to meet sales targets.
Ideally, the same colors should have the same meaning throughout a document. For example, in the two charts below, the same set of colors represent product lines in one chart and product categories in the other. This makes it more difficult to understand the charts, and might lead users to interpret the charts incorrectly if the users do not read both legends.
The appropriate level of complexity for a document depends on the audience. However, complex should never mean confusing.
Avoid overstuffing a document or putting too much in any single chart or table. For example, the left chart above is difficult to interpret because it contains so many bars and they are arranged alphabetically. When ranking is applied and only the top five results are shown, as in the chart on the right, the main message is conveyed more clearly.
When using visualizations to communicate with others, it is often helpful to stick to common chart types such as bar, line, scatter, and geographic charts unless there is a specific reason for using a different chart type that may be more difficult for others to interpret.
Other strategies to avoid making a document overly complex are to initially present high-level results and enable users to drill down to see more details if necessary, or to present the details within subreports or in separate documents that are accessible via OpenDocument links.
Finally, it is important to consider whether or not users will understand the meaning of the information in a BI document. This is not just an issue for complex documents using technical terms.
For example, what does the simple table below mean?
Clearly the table is about sales by country, but beyond that it could mean many different things. Without additional context, we don’t know if the results are for a single product or service, or for a range of items. We don’t know if Country means country of sale, country of customer, country of distributor, or country of manufacture. We don’t know what time period the results cover. It could be a day, a month, a year, or several years. There is no unit of measurement, and we don’t even know if the sales figures represent sales revenue or sales quantities.
It’s not always safe to assume that the results speak for themselves. You can limit possible misinterpretations by providing descriptive names for objects; using descriptive titles for reports, charts, and tables; including units of measurement; inserting text explanations within documents; and saving explanatory details in document descriptions.
Please feel free to use the discussion below to share your own tips for effective communication using BI. They will be greatly appreciated!