Building Meaningful Dashboards for Your Data Pipelines

The goal of data analytics is not to inform analysts, it’s to inform decision-makers. This means making data actionable, easy to understand and readily available so that your end-users can access it anytime and make smart decisions. One of the primary tools for accomplishing this is the data dashboard. Here are a few tips for making your dashboard a go-to resource for your organization’s decision makers.


Want to learn more about data pipelines, data engineering and related topics?

Get the Eckerson Report, When a Data Catalog is Not Enough - free with compliments of Promethium.


Tie Your Data to Your KPIs

For data professionals, creating a data dashboard can leave you feeling like a kid in a candy store--so many interesting options to choose from. However, out of the many fascinating data outputs you may be seeing, it is likely that only a few are of any importance to your end-user audience.

You can start by asking the end-user, but if they’re not too data savvy you may not get a useful answer. If you don’t, back up and try to discover their key performance indicators (KPIs). If they can’t tell you those, then take an additional step back and find out what their objectives are. Once you’ve identified those, then determine which metrics map most closely to those.

For example, if their objective is to reach a certain sales goal, start with some measure of total sales as the chief KPI, and then determine what factors might be affecting that, such as advertising metrics that directly influence sales, and use them as secondary KPIs. If your goal is reducing defective products in a supply chain, your chief KPI is a measure of defective projects, with the secondary KPIs being factors that might contribute to those defects. Once you’ve identified those KPI metrics, build your dashboard around them.

Leverage Domain Expertise to Refine Visualizations

When you integrate a data pipeline, you’ll make certain decisions as to what gets displayed. For example, you may decide to clip outliers. But how do you know whether those outliers are actually meaningful data, or just noise? Subject matter experts can tell you, for example, whether an unusually high reading on equipment is most likely a malfunctioning sensor, or if it indicates that something is seriously wrong. Similarly, a marketing expert could tell you whether an unusually high response to an ad is simply a random anomaly, or something that needs to be investigated further.

In addition to identifying potential problems with the dashboard, the SME can help guide you as to what questions you should be asking of the data. They also might have insight into how the data is collected that might be useful. For example, if a high number of fields are coming up NULL, could this be the result of a manual collection process? If you’re relying on sales associates to fill out forms, for example, could this process be automated? Or, perhaps you’ve got a lead capture form that allows users to input duplicitous answers? If so, this could be addressed to get more reliable data.

Make it Attractive and Easy to Quickly Interpret

The goal here is to make the dashboard something that your target users want to check, and not something that they feel like they have to check. So think through it from their perspective. If it’s unlikely that your audience understands statistics, things like distribution curves and p values are not going to be helpful.

Do audience tests to gauge how quickly they can draw conclusions based on your charts. If it takes more than a minute, they’re probably not going to be very engaged.

Offer Real-time Insights and Alerts

Dashboard users need to be able to observe trends over time. But they also need to know when something unusual has happened. Most of the top dashboard software platforms allow you to set alerts based on conditions--when sales peak or drop significantly in a certain area, when machines are running dangerously close to capacity, and so forth.

Find out what your users need to know in real-time, and take advantage of those alerting features. But make sure they’re not too easily triggered--if your users are constantly getting alerts over trivial matters, they’ll turn into white noise.

Select the Dashboard Tools that Work Best for You and Your Team

There are plenty of options out there when it comes to creating dashboards, and each has their advantages.

Tableau is probably the most popular one, and so there is no shortage of training out there on how to use it, but there are others like Looker that also have free usage tiers and provide similar functionality. Domo touts itself as being easier to use for business users, for example. And, if you’re collecting a lot of data from Google properties--as is the case in many marketing departments--Google Data Studio may also be an attractive and relatively low-priced option.

If you want to know how you can easily connect your dashboard to multiple data sources across the organization, check out the Promethium Data Fabric.


Recent Posts

See All