The hot take that won't die
"Dashboards are dead." It's the data world's favorite provocation right now. Somebody posts it on LinkedIn every week. Lots of nodding. Lots of engagement. Very little nuance.
Here's where I land: dashboards aren't dead. But dashboarding as a function? The way most companies treat it? That's broken. It's been broken for a while.
The "dashboards are dead" crowd isn't wrong about the symptoms. They're just diagnosing the wrong disease.
The real problem with dashboards
I've built a lot of dashboards. I've also watched a lot of dashboards get ignored.
The pattern is always the same. Someone asks for a dashboard. You build it. They look at it for a week. Then they come back and ask if you can add one more filter. Then another slice. Then a different date range. Then a slightly different metric definition. Then they stop looking at it entirely.
This isn't because dashboards are useless. It's because people always want to slice the answer just slightly different than they did the last time. That's the nature of real questions. They're not static. They shift based on what you just learned.
Dashboards were built for fixed questions. Business runs on questions that move.
Where dashboards actually are dead
There are places where traditional dashboards genuinely don't work. And I'd argue they never really did - we just didn't have better options.
Operational reporting is the big one. When a sales rep needs to know which accounts are at risk, they shouldn't be logging into a BI tool. That data should be in their CRM. When a CS team needs to see usage trends for a renewal conversation, that should show up in their workflow - not in a separate tab they forget to check.
This is what reverse ETL tools like Census and Hightouch are solving. Move the data to where people already work. Don't make them come to you. Don't make them learn a new tool. Put the number in the field, in the tool, in the moment they need it.
I've seen this work really well. The data team stops being a dashboard factory and starts being a data delivery service. The insights actually get used because they show up in context - not in some BI tool that half the company has never logged into.
Where dashboards still matter
But here's the part the "dashboards are dead" crowd glosses over: you still need a centralized source of truth.
When the CEO asks "how did we do last quarter?" - that answer needs to come from one place. When finance and sales disagree on revenue numbers, there needs to be a canonical version. When someone new joins and wants to understand the business, they need somewhere to look.
Dashboards are great at this. A well-built executive dashboard that answers the five questions leadership asks every week? That's not dead. That's infrastructure.
The problem is we've been asking dashboards to do everything. Strategic reporting, operational reporting, ad hoc exploration, alerting, storytelling. No single format can do all of that well. We stretched the concept until it broke and then blamed the concept.
The dynamic problem
The deeper issue is one of flexibility versus accuracy. And this is where it gets tricky.
People want dashboards to be more dynamic. They want to change the filters, adjust the dimensions, ask follow-up questions. That's a reasonable ask. Static charts answering last month's question aren't helpful when you're trying to figure out what's happening right now.
But the more dynamic you make something, the harder it is to guarantee it's accurate. Every new filter combination is a new potential edge case. Every self-serve slice is a chance for someone to misread the data because they don't know that "active users" excludes trial accounts, or that revenue is recognized differently in Q4.
This is the tension nobody wants to talk about. Making dashboards more flexible means accepting more risk. And in data, wrong answers that look right are worse than no answer at all.
What I actually think works
I lean heavily toward operational analytics. Get the data into the tools people use. Stop expecting a marketing manager to open Looker every morning. Push the lead score into HubSpot. Push the usage metric into Salesforce. Push the churn signal into the CS platform.
Then use dashboards for what they're actually good at:
- A small set of canonical company metrics that everyone agrees on
- Board-level and leadership reporting that doesn't change shape every week
- Starting points for exploration - not the exploration itself
That last one matters. I liked a point someone made recently - that dashboards work best as launchpads. You look at the dashboard, you see something interesting, and then you dig in using whatever tool makes sense for deeper analysis. The dashboard doesn't answer every question. It tells you which questions to ask.
Stop building dashboard factories
The biggest shift I'd push for isn't killing dashboards. It's killing the expectation that every data request ends with one.
Sometimes the right output is a dashboard. Sometimes it's a Slack alert. Sometimes it's a field in the CRM. Sometimes it's a scheduled email. Sometimes it's a dbt model that feeds three different downstream tools. The format should match the use case, not default to "let's build a dashboard."
Data teams that only know how to output dashboards will always be bottlenecked. Not because dashboards are bad, but because they're only one delivery mechanism. And usually not the one closest to the decision.
Dashboards aren't dead. But the idea that every data problem ends with one? That's what needs to go.