Your Dashboard Situation

You, my friend, have a dashboard situation.

With near certainty, I can tell that somewhere in your company, your dashboards are sprawling. They’re stale, cluttered, and inconsistent with one another. This is one of the most common challenges that data teams face, often leading to confusion and lack of trust among data consumers.

How Did We Get Here?

Data teams build dashboards to provide easy, self-service access to the data the business needs. But things change. Data analysts come and go, KPIs evolve, and reporting needs shift along with business strategies. What was important last year may no longer be relevant today. And that’s perfectly okay. But it does mean that our collections of dashboards will continue to need our attention.

Why Is It Important?

I could talk about ROI: dashboard sprawl leads to inconsistent metrics, maintenance headaches, and a loss of trust in data. It drains resources, reduces productivity, and creates compliance risks. But let’s talk about something more personal - your reputation.

If you have a title like Head of Data or Chief Data Officer, dashboards are part of your product portfolio. They might not be sexy, but they are the "front door" to your data team for many business stakeholders. First impressions matter. If you want executives to see the value of analytics and give you funding for aspirational projects, you need to do something about your dashboard situation.

What Should We Do About It? (My Dream Solution)

If I could freeze time, and if resources weren’t a concern, here’s what I’d do to fix your dashboard situation:

1. Contain sprawl: Delete all dashboards below a certain usage or value threshold. Develop a system to regularly clean up unused dashboards. If a dashboard doesn’t have a named owner, delete it. Scorched earth!
2. Rationalize dashboards: For the dashboards that remain, pare back content by deleting all cuts of data that are not essential. If there are similar dashboards, merge them. Prioritize keeping content that is directly aligned with business outcomes. Your dashboard surface area should slim down considerably.
3. Rationalize data sources: Inventory the data pipelines powering your dashboards. Ensure similar dashboards are using the same pipelines, and consolidate if you need to.
4. Set alerts everywhere: Put alerts on the pipelines that power the dashboards, and on the KPIs that exist within the dashboards. These alerts should go directly to the data analysts who own the dashboards. You never want to find out that a dashboard is broken by hearing about it from your stakeholders.
5. Standardize style: Reduce cognitive load for your stakeholders by standardizing design, including colors, layout, nomenclature, and definitions. Make it easy for people to flip from one dashboard to another with the expectation of consistent design.
6. Make everything easy to find: Create a portal to search, browse, and curate dashboards across the whole business, regardless of source or vendor.

The key is to make both backend and frontend improvements. If you only focus on frontend improvements, it’s just lipstick on a pig.

Back to Reality

At this point you might be thinking, “Nice plan, June, but that’s just a dream. I live in the real world. Help me out!” Obviously the real-world complexity lies in finding scalable, resource-efficient ways to address sprawl. In reality, we all must make choices about what to do (and what to drop) when time and budgets are constrained.

In 2018, I wrote an article for CIO Applications magazine called Best Advice for BI Leaders. One of my favorite points was about the cost of perfection. In BI, striving for perfection can be expensive and often unnecessary. Sometimes, "good enough" is the right approach for business needs. This holds true for dashboards. Aim for what’s useful and maintainable, not for a flawless data ecosystem.

Whatever you do, I will offer these words of caution: don’t let your dashboard-organizing work get bogged down in heavy processes. Too much process slows people down, and they will find ways to bypass it. Opt for light, sustainable processes instead.

What Does This Mean for You?

Data leaders: Pay attention to your dashboard collections. Set clear expectations with your team and reward them for maintaining quality. Predictive models and AI follies may steal the spotlight, but dashboards are still your team’s front door.

Data analysts and data engineers: Don’t just focus on the specific dashboards or pipelines you’re building. Look up and out. Treat dashboards like products, not “launch and leave” endeavors. Senior team members should lead by proposing solutions to the dashboard sprawl - not adding to the chaos.

Conclusion

If you aim to achieve excellence in analytics, your dashboards must be excellent too. (P.S. You just read an article about data governance. Hah! Take your medicine.)