Dashboards are dead - 3 reasons to go metrics-first

If you've ever used a BI tool, you've probably grappled with a common hurdle that's the reality in most BI tools - dashboard chaos. Your data team builds dashboards to give users data and insights, but once a dashboard has been shipped, a followup request rolls in. This results in an endless loop of ad-hoc work, and what's even worse, people start using different variations of dashboards, and no one knows which numbers to trust. Efficient? Not really.

1. Abstraction - nobody cares about your data model

Do you want to know the harsh truth? Nobody cares about your data model. Data teams might obsess about star schemas and consistent column names, but in reality, the whole point of analysts' work is making data useful to people, and business users rarely want to learn about table structures.

That's one of the reasons why metrics-first is great. Metrics and dimensions give a higher-level of abstraction than of tables and columns. It allows people to engage with data using a language they already understand. “Hey, let's look at transaction volume, card payments only, and how we are growing in the US vs Europe?”. Pretty straightforward.

These are the terms your business already uses, and by packaging your data model in the same way makes it immediately available to everyone. So, instead of doing another pass on your data dictionary, get started defining your company key metrics. That's a change with some big implications - it means that the data team can take a step back and focus on the semantic layer, and that all end users can do much more BI work themselves.

2. Composability - BI building blocks

Once you've defined your data into metrics, they're like building blocs. Anyone can stack them up in endless combinations again and again, without breaking the foundational component.

Let's look at a classic BI example: Creating a combined graph of New user accounts and Ad CTR. In a traditional tool, you would start by first getting both these values into the same dataset, in the same query, or “explore” to build your vizualisation on the combined data. This task can range from pretty easy to fairly complex, and typically requires the attention of an analyst.

With metrics-first BI, no prep work is needed. Every metric has it's own identity, and can be queried independently, while the result is combined from multiple, separate calls. This means that you can allow end users to mix and match any metrics without ever having to care about joins and queries. With the right tool, you can also use dimensions across metrics. So, if several metrics have a dimension in common, say, country, then you can analyze the country dimension on any level.

3. Built-in consistency, finally

Metrics provide a single source of truth, ensuring that key business figures, like revenue, churn, and customer acquisition are centrally defined, governed, and reused across all reporting and analysis. If you're trying to serve a growing number of data-hungry users in an efficient way, going metrics-first eliminates a classic dilemma where using dashboards commonly leads to either of these two scenarios:

Scenario 1: trying to own and govern all dashboards

Data teams trying to make sure people are using the “good dashboards” (the ones where the latest definitions and metric names are up to date) is a common one. It usually leads to business users turning to the data team for even the simplest questions, adding up to more dashboards, eventually creating a “dashboard hell” and an endless request log. Needless to say, it's nearly impossible to govern.

Scenario 2: going self-service without governance

On the other end of the spectrum, you let teams create their own content using your data - without any aspect of governance. Most analysts would agree that this is going from bad to worse. If the consistency problem is bad in scenario 1, then this is the wild west.

Ironically, some reach the conclusion that self-service doesn't work and that they're are stuck trying to govern dashboards and restrict flexibility.

Luckily, there's a third option available.

What's great about metrics-first is that it gives us strong governance at the right level of abstraction. Business users can use metrics freely - combine, slice, and dice them - and should be encouraged to create their own content. The key is that every chart and table is simply referencing centralized metric definitions, so that names and numbers are the same no matter who creates the chart. And when definitions change centrally, it updates all of the content that was ever created by anyone. Which means you get this beautiful built-in consistency, and can safely encourage folks to both explore and create content together.

The future of BI is metrics-first

The age of dashboard-driven BI is fading. Static reports, duplicate metrics, and governance struggles slow businesses down. Metrics-first BI solves this by enabling composable, intuitive, and governed data exploration without sacrificing flexibility.

It's time to ditch the dashboard chaos. Define your metrics, empower your users, and embrace a BI model that actually scales.