Are dashboards dead? Not quite. They just need to evolve

In recent discussions across the tech and data communities, a provocative idea has been gaining traction: the notion that dashboards are dead. The first time I came across this was in an article by Taylor Brownlow, "Dashboards are Dead". A worthwhile read. The article suggests that dashboards, as we known them, no longer serve the needs of modern data-driven organizations. Not through their own fault, more through misuse or over-asking.

Working where I do, the article caught my eye and I can't say I disagree with Taylor's sentiment. People want more than just static numbers, they want to explore and drill-in, without learning SQL. Our products have always been more than a dashboard, in those terms, but I can forgive someone for coming to that conclusion about SquaredUp when "dashboard" is still the best term.

What's wrong with dashboards?

When people say “I need a dashboard,” what they’re really saying is “show me the data.” No organization in the world has moved beyond the need to use data to make informed decisions. The challenge is that the old way of delivering that data is slow and views are typically static. People don't want that anymore. These last generation tools have tarnished the word.

Data today is distributed across SaaS tools, data lakes, and a whole host of other places, and it's not necessarily stored with anything resembling structure or consistency. We need a new generation of tools that can cater to the reality that data comes from everywhere, and shelve those solutions that require all that data to be consolidated or transformed just to make a useful line graph.

Equally, dashboard tools need to work. Not only for the datasets they're connected to or an army of data scientists, but for their end users. And for organizations to get true value from a dashboard tool, those end users should be everyone, and they should be empowed to self-serve and explore.

Ultimately, the next generation of dashboard tools need to provide operational intelligence for all, and that is a whole world away from the dashboards of the past. No more static views based on stale data ingested days ago, no days of waiting for the data team to build a new query or add a new filter, but real-time discoverable insights delivered at scale.

Complexity and sprawl

Let’s consider Grafana, a very popular and widely-adopted tool in the open-source community. While powerful in certain scenarios (show me a Prometheus deployment without Grafana on top...), it can be overwhelmingly complex for the average user. Its narrow core use case makes it less than ideal for organizations that need flexibility for a wide audience. It just isn't everyone's first choice, because it isn't built for everyone. (Check out our list of Grafana alternatives.)

I don't intend that to be a slight, but like many trend-setters in the tech world, opening up demand to a mass market should come with the drive for continual improvement, to keep leading the way. Grafana has its niche, and it seems they're sticking with it.

Moreover, Grafana tends to suffer from sprawl. It often ends up deployed multiple times into the data/user silos it serves. This fragmentation can lead to isolated instances where insights can't be shared across teams. This siloed approach, while perfectly serving those core users, is counterproductive to achieving true operational intelligence where a modern single pane of glass would win out. Grafana isn't the organization-wide choice, so it isn't implemented with that as a consideration.

The flexibility inherent in Grafana also introduces variability. The vast array of plugins from various third-party developers can sometimes result in inconsistency. Users find themselves working with plugins that behave differently to others, adding a layer of complexity in maintaining and standardizing dashboards. This can lead to ‘dead dashboards’ - dashboards that are no longer actively maintained or used, diminishing their value over time.

Flexibility doesn't need to come at the cost of simplicity.

The limitations of BI tools

On the other end of the spectrum we have tools like Power BI and Tableau. Typically very robust solutions that do not struggle at all to derive insights from data. Well, assuming you've got a PhD that is... While offering extensive analytics capabilities, they come with significant caveats. Power BI, for example, kills at processing large amounts of data, but this strength is also its weakness. The tool’s heavy reliance on ingestion can be a stumbling block for organizations that need real-time insights. (See our take on Grafana vs Power BI.)

Additionally, BI tools are often lacking integrations, requiring a substantial amount of technical knowledge and time to roll your own. While BI tools can work well (when supported by analysts), they tend not to cater directly to the needs of your teams. Plug-and-play is not something you'd typically associate with tools of this nature. This inevitably leads to tool sprawl, where people drift, often in secret, to tools they can manage themselves. No one team can really benefit from the sea of data that should be at their fingertips.

Moreover, the cost of BI tools extends beyond the software licenses. A significant consideration is the cost of data storage. BI tools often require substantial storage to handle the volumes of data they process. This can lead to increased costs for maintaining the necessary infrastructure, especially as data volumes grow. For larger organizations with more complex needs, there’s a considerable human cost involved too.

Organizations frequently need to employ data analysts / scientists / engineers, whose expertise is essential for interpreting complex data sets and extracting those needed insights. Their involvement is crucial but it can significantly add to the overall cost of traditional BI.

The need for evolution

The evolution of dashboard tools is crucial to addressing these challenges. Next generation tools need to integrate seamlessly with diverse data sources, offering real-time insights without requiring all your data to be ingested into a single repository. They should provide a consistent user experience regardless of the data source or audience, and they should be extensible through common frameworks to cater for niche or legacy systems.

They should also work for everyone, whether that's an occasional user who just needs a quick visual cue, or the highly technical who are more comfortable on the command line than in a point-and-click UI. These tools need to acknowledge that not every user wants to "see" something, they want to connect with their data through other means. A rich API that allows data to be used elsewhere, or contextual, timely notifications delivered into messaging tools, for example.

A next generation tool should also prevent sprawl and add new value by allowing relationships between data and dashboards to form, to share and surface their insights without the need for repetition. A semantic understanding the data and its subject is also key, for example by indexing and correlating entities, making them explorable and helping users understand dependencies. This approach ensures that dashboards themselves become part of your data fabric, not just a layer on top. The irony that data visualization tools suck at making dashboards visible isn't lost on me.

Dashboards are far from dead

My not entirely unbiased opinion is that the perceived issue lies not with the concept, but with these last generation tools like Grafana and PowerBI that have failed to evolve with demand. Nobody has moved away from the idea that being data-driven is the key to good decision-making, but it's certainly easy to feel like you're drowning in a tidal wave of data when you're stuck with legacy tools that aren't helping you keep afloat.

Data is one of your most valuable resources, but we're all guilty of overlooking that. And in fact, the actions we take when hitting these challenges are often part of the problem. Simply shovelling data around between tools that all promise to change everything only goes to grow our negative sentiment.

Instead, look to the next generation of tools that work with your data regardless of where it is, and more importantly, work for you. Tools like SquaredUp, naturally.