Updated: Oct 6
Creating dashboards that drive meaningful results and not just "Interesting But Useless Information (NBUIs)" is one of the biggest challenges that businesses face when visualizing data. The amount of data sources and measurement options keeps growing exponentially, this poses a great opportunity to measure what really matters and also the threat of visualizing NBUIs.
Stephen Few, in his book "Information Dashboard Design", describes a four step framework to manage performance:
In this blog post, we will go over how you can achieve these steps when creating a KPI Dashboard. As a reference, we will be mentioning Microsoft Power BI, but the same can be achieved with your preferred Business Intelligence platform like Tableau, Qlik or MicroStrategy. (But seriously, why are you not using Power BI?)
1. Choosing the Right KPIs
The goal is to provide inspiring information that moves people to action. - Guy Kawasaki
The first is to choose the right KPIs so that users of the dashboards are given a high-level situational awareness of interesting and useful information. Here are some important recommendations when choosing KPIs:
Use Metrics as Proxies
Metrics should not become targets. In the same way that students should not fixate their education on their GPA score, we shouldn´t be counting on quantitative metrics as the only thing that matters. Because as Albert Einstein use to say "Not everything that can be counted counts and not everything that counts can be counted."
Plan for Actionable Time Frames
Imagine if every time you drove your car you only saw your average speed after you finished driving instead of having constant feedback of what your speed is. We would all have many more speeding tickets! The same thing happens nowadays in business. Long gone are the days in which managers can only see their metrics after the month or quarter is finished.
With the increases in processing power of the last decades and the speed in which reports can be created, it is now possible to have constant updates that can drive action. The key here is to decide how often you can take action upon the information. If you can only take action upon a metric every other week, then there is no need to build a data pipeline that gives you minute by minute updates.
Review your KPIs constantly
Determine a cadence for monitoring and acting on KPIs. Are KPI dashboards monitored daily or weekly? Do actions only come after reporting or are stakeholders and/or team members empowered to adjust tactics along the way? Whether you’re meeting your goals or falling behind, KPIs are tools to help you make better decisions. So don’t track just for tracking’s sake, take action.
Only pick KPIs that are aligned to your specific goals. If your objectives aren’t clear, matching KPIs against them will prove complicated.
Have a balance of KPIs and KPPs
There are multiple ways in which KPIs can be categorized. One way that can be really useful in the context of actionable metrics is dividing them in KPIs and KPPs. Key Performance Indicators (KPIs) describe what has already happened while Key Performance Predictors (KPPs) predict something that might happen (if no corrective actions are taken).
In many business areas, the same KPI can be used as a KPP. Take out-of-stock events for example, through careful inventory tracking we can know how many out-of-stock events we had in the past, but there are also ways of estimating which products will be out-of-stock over the next week and take action by transferring inventory among distribution centers or stores. Both KPI and KPP are important but the latter allows for corrective actions.
If it is not possible to predict the future status of a metric to create a KPP, what you can do is find metrics that are leading indicators of a metric and use those as proxy KPPs. In Sales, a common KPI would be Revenue Growth, this metric can be hard to estimate but there are many leading indicators that could help you estimate what Revenue will look like in the future. Some examples could be "Number of Sales Meetings Scheduled", "Number of Calls Scheduled", "Number of Open Quotations", etc.
2. Choosing the Right Visuals
The purpose of visualization is insight, not pictures. - Ben Schneiderman
When visualizing KPIs, there are many ways in which things can go awfully wrong. Here are some examples of dashboards from Stephen Few´s book that could be visualized better:
Example 1 - Variety Gone Amok
In this dashboard we see metrics being visualized in five different ways. This may seem to make the dashboard more attractive but it forces the user to try to understand each graph one by one instead of giving the user a quick and understandable high-level situational awareness. The alternative would be to visualize metrics in a one simple way.
Example 2 - Poor Order of Data
Each visual within a dashboard should have an order that takes into account how people analyze the dashboard, the importance of the metrics and how are actions are taken. In the following dashboard we see how the most important real-state in the dashboard (i.e., the upper left corner) is used for a metric that could be important for someone in the organizations but is unlikely the most important metric in this dashboard. In a similar way, the "Order Size Trend" would be better placed right next to the "Order Size" gauge chart as they are related.
Example 3 - Dashboard à la Pollock
Take a look at the following dashboard and see what catches your eye and which metrics are performing well and which ones aren´t. Everything in the dashboard is structured in the exact same way so there is nothing that really stands out that forces the user to focus on something. Apart from that, there is no context added to understand if metrics are performing well or not so the user is not given valuable actionable information to act upon.
An alternative to Gauge Charts is visualizing KPIs using Bullet Graphs. This type of graph resembles a thermometer in the way that it shows how a metric (1) is performing (2) in comparison to various qualitative ranges (3) and a symbol that shows a comparative measure (4) all in the context of quantitative scale (5).
The most important feature of bullet graph is that they pack all that information without occupying much real estate on the dashboard and they can be layered on top of each other in a horizontal or vertical way.
Power BI Tip: If you want to start using Bullet Graphs on your reports you should check out the Bullet Chart Visual from the guys over at OKViz.
Showing the status of a metric in a specific period in time is not always enough, that´s why it is important to also add graphs that show how a metric has been changing across time. That´s where Sparklines come in, they can be understood as a very small line chart without axes or coordinates.
The example below shows a (1) metric for six different (2) categories across the last 12 weeks (3) in comparison a (4) target.
Power BI Tip: If you want to start using Sparklines on your reports you should check out the Bullet Chart Visual from the guys over at OKViz.
When put together, Bullet Charts and Sparklines make a great combination to update users with high-level situational awareness and it helps them identify items that need attention.
If you want to learn more about choosing the right chart for other use cases, be sure to check out our blog post on How to Choose the Right Graph.
3. Choosing the Right Add-Ons
The last step you could do is to enrich the information on the visuals with additional information they could to determine an appropriate response. Here are some examples of things that can be done:
Adding a tooltip allows users to hover a visual and see more information related to each category. Here for example, the Tooltip displays the Category Manager and the Country of Origin for that Category.
Buttons + Details Dashboard
Another great option is to add buttons that take users to the details behind these KPIs. Giving users transparency into how metrics are calculated and what information is taken into account is a great way to convince users to trust the data and act upon it.
Feel free to play around with the live version of the dashboard and let us know how you would improve this KPI Dashboard!