Many modern BI tools are self-service, meaning that they’re designed to be as user-friendly as possible. In particular, these BI tools are designed to make it as easy as possible for users to build dashboards.
Dashboards are the basis of any effective BI strategy. With dashboards, users can visualize their data analytics and connect with them on a more intuitive level. Through graphs and charts that track business data, users can leverage the power of dashboards to make more effective, data-driven decisions.
Dashboards are the main building blocks of a BI strategy, and visualizations are the main building blocks of dashboards. Every dashboard is made up of a handful of different data visualizations, which take a certain data stream or source and represent it visually so that it can be more easily understood by dashboard viewers.
These visualizations can be common types of graph or chart, like a bar graph or a line chart, or they can be more unique and specialized representations of data. Regardless of what the visualization actually looks like, they all share one key element—they’re powered by specific streams of relevant business data.
These streams of business data are called ‘metrics.’ Ideally, metrics provide valuable insight into a business’s operations and its strategic success. The more essential and valuable a metric is, the more valuable the visualizations it powers are.
The dashboard building process begins when data analysts find valuable metrics, connect them to their BI tool if they’re not already connected, and figure out the most effective and intuitive ways to visualize them, so that dashboard viewers can easily understand their implications. However, there’s a massive problem at the beginning of this process.
How do dashboard builders know which metrics to include in their dashboards? The average business has hundreds, if not thousands, of different data sources connected to their BI tool. How can dashboard designers and data analysts sift through all this data to find the half dozen or so metrics that answer their business questions?
Data analysts have all sorts of tricks, tips, and strategies that they use to find the most valuable metrics for their dashboards. Unless a business picks up on some of these tips and figures out some metric selection strategies of their own, they’ll never build effective BI dashboards.
What do useful metrics look like?
The first step of finding useful metrics is also the first step of building a dashboard in general—figuring out the business question that the dashboard (and the metrics that make up the dashboard) should answer.
This is a complex topic, and something that we’ve published a guide for in the past. In general, though, it’s best to ask questions that can be meaningfully measured and don’t rely on making value judgements about data. This will make finding the appropriate metrics to answer the question much easier.
There’s a sort of back-and-forth tension between the central business question that drives a dashboard and the KPIs that power it. Dashboard builders want to answer useful business questions with useful metrics, but in many cases, the availability of certain metrics impacts the sorts of questions that they can actually answer.
In short, it’s important to realize that a business question might shift, broaden or narrow in scope, or need to be discarded entirely, based on what sort of metrics data analysts can find and connect.
What do valuable metrics actually look like? The specific metrics that are useful for tracking business performance will change from business to business and industry to industry, but there are some general guidelines for finding valuable metrics.
First, look for metrics that have specific applications to the business question at hand. It’s best for a metric to have generally the same scope as the question it’s trying to answer. A broad metric like business-wide revenue might be useful in some situations, but it doesn’t have a ton of utility in more specific use cases.
Along the same vein, don’t use metrics that are too specific. Specificity is good, because it allows a metric to be applicable to the business question it’s trying to answer, but it’s easy to narrow a metric so much that it doesn’t present any implications for the broader business question.
Next, look for metrics that actually measure something that’s valuable to measure. Many metrics relate to the business question well enough but don’t measure anything valuable. This is often what separates KPIs from less valuable metrics—they measure things that provide important real-world implications about business health.
Often, the value of a statistic changes from business to business. One marketing team might only care about conversion rate at the exclusion of other metrics, while another might focus only on lead generation and not worry about whether those leads are converting. When finding metrics, it’s important to think about what sort of data a business might value.
Lastly, it’s best to look for data streams that update frequently. Data has a shelf life, and it doesn’t take too long for it to go stale and become completely useless for any sort of real-world analysis.
In the past, businesses could often get away with using data sets that updated very slowly. Business operations didn’t move as fast, and it wasn’t too harmful to use data that was days, weeks, or months out of date. Nowadays, that’s not the case.
In many cases, it’s risky to use data that’s even an hour out of date. In some industries, even data that’s minutes late doesn’t cut it. Manufacturing companies, for example, need to know what’s happening on their assembly lines in real time. Anything else, and they could waste thousands of dollars of product on outages and issues that they didn’t know about.
Even businesses where timeliness is less critical, it’s better to use metrics that update frequently. This helps to keep the implications of the dashboard more timely, which makes the dashboard more valuable. Dashboards that don’t update as frequently will be less valuable.
How do data analysts find good metrics?
Even if a data analyst knows what sort of metrics they should be looking for, it’s still a challenge to actually sift through the mountains of data that a business collects to find relevant metrics and then settle on the half-dozen or so most valuable metrics from among those.
The first challenge is to actually separate the signal from the noise—delve into a business’s millions of rows of collected data and find the metrics that are actually relevant to the business question at hand.
In some ways, this is the easiest part of the process—once a dashboard builder has a general idea of the metrics they want to use to power their dashboard, they’ll have a fairly good idea of where to look for those metrics. For example, if they want internal financial data, it makes sense to look at the data coming in from their accounting tool.
Of course, it’s not always this easy. Sometimes, a dashboard builder might not have the correct credentials or permissions to access the data they want to use. Other times, they might not be able to connect to the data they want because integrations aren’t working or are unavailable.
In general, though, actually finding the data sources necessary for a dashboard within a BI tool or database isn’t the challenging part of choosing metrics. The difficult part is choosing the handful of metrics that can actually build an effective dashboard.
Usually, a dashboard will only use about four to eight different metrics. This is best practice—anything more and the dashboard won’t be narrow enough to gain quick insight from. Anything less, and the dashboard won’t be able to create a holistic view of the topic at hand.
There are a few ways to separate the most valuable metrics from thoe that can be discarded. First, look for the metrics that are the ‘most’ something. The most valuable metrics are the ones that provide the most valuable answers to business questions.
This might be the metric that’s most specific to the business question, or the one that’s most valuable based on the business’s perceptions of what a valuable metric looks like.
In addition, key performance indicators are extremely valuable. Metrics that have a clear, direct connection to business health or progress towards a goal are especially useful, especially in dashboards that want to monitor goal progress or keep tabs on operations.
Also, it’s always good practice to communicate with the target audience of a dashboard and ask them what sort of metrics they’d like to see. Often, they’ll have clear suggestions for specific metrics that they think they’ll find useful.
Sometimes, dashboard builders can think of a metric that would be valuable to have, but there isn’t a data source that can provide this metric. In these cases, they might have to transform other metrics to derive the target metric.
For instance, a dashboard designer might want to see the average of two different metrics that they already have. However, they can’t find a data source that actually provides this average. If they want this metric, they’ll have to use data transformation or a calculated field to derive it. These techniques allow for more valuable metrics than might be available otherwise.
Selecting the best business metrics
Selecting the most effective business metrics to power a dashboard is still an inexact science, and even expert dashboard designers might end up including metrics that aren’t particularly useful or leaving off metrics that are.
However, there are some best practices that allow dashboard builders to generally find metrics that are useful to the people who will actually be using the dashboard. Metrics that are specific to the business question at hand, that have a clear connection to business success, and that are updated frequently are generally the best to use.
The exact metrics that a dashboard builder should include in a given dashboard will change from dashboard to dashboard and from business to business. There’s no way to build an exact model of what a ‘good’ dashboard should have, but best practices and guidelines can help.
Check out some related resources:
Harnessing the Power of Data to become a better Credit Union
Nucleus Research: Domo’s ROI as a Data Platform
Domo for Financial Services Playbook
Try Domo for yourself. Completely free.
Domo transforms the way these companies manage business.