What makes BI meaningful?
Business Intelligence (BI) is a broad domain and is used to accomplish a variety of goals. Whether we’re talking about predictive analytics or lagging KPIs, the common thread is data visualization that drives effective decision making.
The trick is recognizing the difference between a report that has a large quantity of data versus one that has the right information.
A cautionary tale: Why BI for contract negotiations failed.
A large, multinational enterprise was trying to leverage its size to negotiate better payment terms and create a stronger cash flow position. The team working on this initiative found creative ways to collect supplier spend data from across multiple divisions and decentralized systems of record. They were also able to pinpoint the best payment terms that any division had been able to negotiate with each supplier. It was the first time the company had access to this type of information and the team was extremely excited.
The next step, in their mind, was to give this contextualized information to the groups negotiating new contacts, thereby empowering them to secure more favorable terms. The team created clean and clear reports that they proudly shared with their decision makers.
Unfortunately, there was not enough information in the reports to allow the negotiators to make the necessary decisions. Two key pieces of information were missing:
- The decision maker didn’t know what other concessions a supplier may have made when approving better terms with another division, at least not without leaving the report and finding more details in the Contract Management System.
- They also didn’t know if there were alternate sources for the materials the particular supplier was providing—vastly influencing the negotiating leverage. For that information, the user would have to go to the ERP system and run some complex queries.
The end result? The people negotiating contracts never optimized their existing agreements to drive better payment terms. It’s not because there weren’t valuable opportunities for the company. It was a task abandonment issue. They were too busy with other things to go to three different places to find all the information necessary to make a decision.
In my experience, there are a lot of reports out there passively presenting large amounts of data but not actively driving decisions, therefore making them considerably less effective.
The right amount of information—no more, no less.
There are dangers in not providing enough information to a decision maker at the singular time and place when it is needed to make a decision.
Switching costs
The first risk is loss of productivity. When a decision maker must access information and context from multiple places, switching back and forth between reports and visualizations, their time is wasted.
Errors
The second risk is misinterpretation. When the decision is timebound and absolutely needs to happen, not only is system-switching a complete waste of time, it’s also an opportunity to misread, misalign, or otherwise misunderstand the insights due to human error.
Abandonment
The third danger is much worse. When the decision is not timebound, the complexity of the process to visit all information sources leads to task abandonment. As in the example above, the decision maker simply moves on to other tasks that are considerably easier to accomplish and never returns to the decision.
On the other hand, it can be problematic to share too much information.
Distraction
An abundance of new technologies allows companies to easily merge and combine data sets in new ways. It can be tempting to provide more information than the decision calls for just because you can. Sometimes that additional information seems relevant, but it can actually be distracting from the important facts.
Bias
The determination of what information should be included and excluded should not be taken lightly. Working with subject matter experts to fully understand the criteria for a given decision is extremely important to avoid presenting information in a way that skews or confuses a decision.
Start with the user in mind and tell them a story.
How information is laid out visually has a huge influence on how well a decision maker can comprehend it and therefore act on it. The data should tell the end user a story. There’s nothing more frustrating than peering at large tables of numbers or values that are not organized in any meaningful way, trying to figure out how the data is pertinent to your decision at hand and what it says.
Let’s say an insurance company needs to determine where it faces the most financial risk in terms of geography and type of natural disaster. The organization plans to use this information to make decisions on insurance coverage premiums. Assume in this example that the decision maker is looking specifically for where and for what they should charge their highest premiums.
The following two reports use the exact same dataset but communicate the story very differently.
Report A includes way too much data and is not designed in a way to help the conclusions jump off the page. There is clearly enough information to make a decision, but as a user, you need to think about what you’re looking at in order to make the right decision.
Report B sorts the data and provides a layout that quickly answers the questions. The viewer is guided to the conclusions that the Texas and Florida incur the most cost and that tropical storms have caused the most damage. This information is perfectly clear at a glance. Report B carries a much lower risk of misinterpretation than Report A.
Once again, information bias is a concern. The same data can tell contradicting stories depending on how it’s laid out, so it’s extremely important to collaborate with multiple subject matter experts and to uncover if there’s an important piece of data or a perspective that’s excluded. (In this oversimplified example, there would likely be several additional factors that impact the setting of premiums which are not represented in this dataset.)
Does beauty matter?
Over the course of my career, I’ve finally come to the conclusion that making a visualization appealing is in fact very important.
To be clear, if the data is not comprehensive enough for the decision at hand, aesthetics are irrelevant. But once you’ve achieved the optimal level of detail, flow, and format, it’s worth putting some thought into how the report looks.
There are a couple of reasons why looks matter:
- A stronger visual design signals to a decision maker that the underlying information is of high quality and therefore trustworthy.
- Any impediment to user adoption should be removed. Beautiful applications have proven to invite more adoption and garner more usage.
Aesthetics may seem like a squishier characteristic but providing your key decision makers with non-eyesore data insights to help drive their decisions is one simple thing you can do to reduce friction from their workflow and encourage adoption. This is especially important if users will need to leverage the visualizations on a regular basis.
Keep in mind that beauty is in the eye of the beholder and requirements will depend on the setting. The best way to determine what your target users consider to be beautiful is to work with them directly, put low fidelity mockups in front of them, get feedback, and adjust accordingly. Your users will make different choices depending on if they are holding a tablet in their hand while on the shop floor or sitting at a desk with a large monitor in front of them.
BI is meaningful if it helps people make decisions. In a visually appealing interface, the right amount of information will be enough to avoid switching costs, errors, and abandons, but not so much to create distraction, confusion, and bias. Input on prototypes from multiple subject matter experts will help to inform and shape the optimal level of detail and the most effective way to present the story of the data.
