From Noise to Narrative: Making Sense of Data Products

Last month, we went live with our latest Ask Me Anything (AMA) session - this time diving deep into the world of data products.

Our hosts Simon and Ust took to the virtual stage with a few guiding slides, a few spicy takes, and a shared mission: to demystify what data products really are, why they matter, and how to make them actually work in the real world.

If you missed the session or just want to revisit the highlights, here’s your quick-fire recap complete with practical advice, incorporating answers to questions from the audience.

What Is a Data Product?

Let’s get this straight first. When we talk about data products, especially in the analytical space, we don’t just mean a dashboard or a database table with a fancy name.

A data product is:

• Designed for a specific purpose with users in mind

• Backed by a curated, trusted data model

• Intended to produce outcomes, not just outputs

• Owned, supported, documented, and measured like any other product

In other words, it’s not just a prettier data mart. It's the fusion of analytics, usability, and intent.

Analytical data products, our primary focus, are typically semantic models with clean interfaces that enable business users to answer real questions. They’re treated like software: built, launched, supported, evolved.

Why Bother?

Because the traditional way of doing things is broken.

Spinning up dashboards or one-off reports in response to ad hoc requests creates slow, fragile, unscalable messes. Adopting a product mindset shifts the game entirely:

• From outputs → to outcomes

• From requests → to ownership

• From dashboards → to decisions

This mindset is already second nature in software development. We just forgot to apply it to data.

What Makes a Great Data Product?

Simple: the same things that make any product great.

• Valuable – It solves a real problem or unlocks measurable value.

• Usable – People can intuitively interact with it.

• Feasible – We have the tools and skills to build and maintain it.

• Viable – It respects budget, legal, and ethical constraints.

Building great data products is less about the tech and more about designing for your users, not just servicing their requests.

How Do You Build One?

With intention—and structure. At Advancing Analytics, that means using the SunBeam framework. Think of it like agile design thinking for data:

1. Start with “why”: What’s the business goal?

2. Map the canvas: Who are the users, what’s the event flow, where’s the value?

3. Design with questions: What do users need to know to make decisions?

4. Build iteratively: MVP first, value fast, refine over time.

You’re not throwing together a star schema and calling it a day. You’re delivering a packaged experience, complete with launch, comms, training, and ongoing support.

How Do You Measure Success?

Not just by counting report views.

A good data product measures:

• Engagement – Who’s using it, how often, and are they coming back?

• Outcome delivery – Are decisions better or faster?

• ROI – Through scenario-based value forecasts (and no, it’s not all in pounds or dollars).

If your users never open the model, or still ask the same questions in Teams, your data product didn’t land. A good product gets used, and gets talked about.

What About Conversational Interfaces?

Glad you asked.

Chat-based analytics tools like Databricks Genie or Microsoft Copilot aren’t the product they’re interfaces into the product. Think of them as new windows into your curated semantic models, designed to boost adoption and accessibility.

Dashboards may be precise. But conversations are flexible and familiar. Sometimes, to build trust in your data, it helps if your data can talk back.

Do You Need to Do Everything?

No. You don’t need a fully governed, contract-heavy, delivery pipeline-automated, agile-squad-spun-up everything-from-scratch data product platform to start.

You just need:

• A commitment to value

• A willingness to design for users

• A bit of time and space to iterate

Start small. Start with one product. Show value. The culture shift will follow.

Want to Become a Data Product Manager?

You don’t need a specific degree. The best data product managers we see either:

• Come from a data background and learn product thinking, or

• Come from a product background and pick up enough data literacy to apply it well

You need to understand value, users, and data modelling plus how to sell ideas internally. Most of all, you need empathy: for the user, the business problem, and the data team.

So, Where Do You Start?

It’s easy to get overwhelmed. But you don’t need to overhaul everything on day one. Here’s a simple path to begin:

1. Run a two-hour workshop to identify a high-value use case.

2. Design your first data product using the SunBeam framework.

3. Launch it with impactful internal comms, storytelling, demos.

4. Measure success and evolve from there.

We offer free workshops to help teams take that first step. And our design and implementation accelerators can help you move fast without compromising quality.

Final Thought

A data product is more than an artifact. It’s a mindset shift. From “data for data’s sake” to “data in service of action.”

If you’ve got topics you want us to explore, or questions we didn’t get to, drop us a comment or email us directly. Until then - keep building products people actually want to use.