“Build nothing until you’re sure someone will use it.”
That’s not just a product mantra, it’s a survival principle for modern data teams. But while product teams obsess over customer validation, data teams often sprint into building pipelines, dashboards, and data products without checking if anyone actually needs them.
And here's the punchline: the technology is rarely the bottleneck anymore. It's the demand that’s missing.
Why You Should Read This
Most data teams are busy mitigating the wrong kind of risk. They worry about how to build something instead of why they’re building it. This leads to beautiful dashboards no one opens, pipelines no one trusts, and analytics that don’t move any metric.
This newsletter is about flipping that mindset. It will help you:
- Spot the real risk in your projects
- Avoid wasted months of engineering effort
- Make your team more valuable to the business
Let’s break down where things go wrong, and how to get it right.
The Trap: Building the Perfect Solution to the Wrong Problem
Here’s a scenario that’s far too common. And it happened to me exactly like this 11 years ago on my first data leadership job:
I was leading the data team of the world's largest company builder and my task was to build a data infrastructure blueprint that we could copy and scale across all portfolio companies.
So I did what I thought was the right thing to do:
- Hire a rockstar engineering team
- Stand up the perfect infrastructure
- Set up scalable pipelines
- Build beautiful dashboards
- Develop sophisticated predictive models
Five months later, we launched.
And people hated what we had built.
It turns out…
We built a Ferrari for a family with 3 kids who needs a minivan to go on shopping trips and holidays.
We built a perfect product that did not fulfill the needs of our users.
This is market risk. And most data teams completely ignore it.
The Two Types of Risk in Data Work
Let’s get clear on terms:
- Product risk = the risk that you can’t build what’s needed.
- Market risk = the risk that no one wants what you built.
Ten years ago, product risk was real. Today, thanks to:
- Cloud data warehouses
- Massively scalable compute and storage
- Built-in AI and data science tools
…product risk is very low in most cases.
That means market risk is the real enemy.
Yet most teams still behave like they’re in 2010, obsessing over infrastructure while ignoring the most basic question: “Will this be used?”
The Fix: De-Risk the Market Before You Build
Here’s the shift:
Before you write code, before you build a pipeline, you must validate demand.
Here’s a 3-step filter to determine whether a pipeline is even needed:
- Repeatable & well-defined use case → Is this something that happens regularly and follows a clear pattern?
- Needs multiple data sources merged → If it’s just one source, maybe you don’t need a pipeline at all.
- Mapped to a measurable business goal → Can this data product clearly influence a KPI?
Unless all three are true, I don’t build any data pipeline.
Real Talk: Why We Over-Build
So why do data teams still over-engineer?
- Pipelines feel like progress. They’re tangible and measurable.
- There’s a bias toward infrastructure. Especially in tech-driven cultures.
- Governance concerns are overblown (too early). Teams build guardrails before there's even a road.
I used to build pipelines too quickly.
Now, I never build any pipelines before I am 100% sure I understand how I will act on the data and I will not redefine or deprecate the metrics in the next 6 months.
The Bottom Line
Building data products without demand is the fastest path to irrelevance.
Before you ship anything:
- Ask: What’s the smallest step we can take to test demand?
- Focus on market risk, not product risk.
- Build only when there’s a clear, repeatable use case tied to a business outcome.
Because in the end, nobody cares how clean your pipeline is if it doesn't solve a real problem.
Let’s build less and deliver more.
Until next week,
Sebastian
