How I used Hex’s Notebook Agent to make pricing analysis a snap

You may have seen that we just launched our game-changing (and absolutely delightful) Notebook Agent. It can tackle an endless range of questions — but here’s how I used it to crack one of the trickiest: pricing analysis.

If you’ve ever dug into pricing, you know it’s a beast. Getting it right means wrangling a messy set of questions:

• What are people buying?

• How are they buying it?

• What plan do they land on first — and how does that evolve?

• What features do they actually use?

Each of those requires its own deep dive. And behind the scenes, that means lots of time series work, complex joins across multiple sources, and a nuanced understanding of sales and customer behavior.

At most companies, this kind of work takes months of analyst back-and-forth or expensive consultants. Even in the best case, there’s always “translation cost” — my context as a product leader doesn’t always make it perfectly into a ticket or quick sync, which means iteration cycles and missed nuance.

How the Notebook Agent changed the equation

This time, I tried something different: I opened a Hex Notebook and started working directly with our new agent.

The agent is embedded directly into the core Notebook experience. It can write SQL or Python, pull data from our warehouse, and build charts — all guided by natural language prompts.

I started asking questions: How are customers distributed across our plans? What’s the average time to upgrade? What features are used on each tier?

And the agent answered. It wrote the queries. It built the visualizations. It joined the right tables — even using window functions when needed. I don’t write much SQL anymore, and I don’t have the entire data model in my head, but I could still follow what the agent was doing.

It made the work approachable and fast and because I could see the agent’s work, I felt confident in where I was headed. If I felt the need to take charge and edit directly, I could do so with the Notebook functionality - drag-and-drop viz or editing SQL was also directly at my finger tips.

I could easily duplicate cells and ask for light edits to see new cuts of the same data to better understand the types of customers who are on each plan.

The result?

In just a few hours, I had a solid draft of the analysis: a full view of how customers move through our pricing tiers, what features they use, and how their teams grow over time. This is the kind of work that used to take weeks — not because I couldn't think through the questions, but because I couldn’t execute them alone.

Before finalizing, I shared the analysis with our data team to validate the approach — and they had no notes. Zero. I expected at least a few nits, maybe a suggestion for a more performant join. That’s when I knew our product had experienced a significant shift.

The impact: accelerating analysis

Since then, I’ve been using that pricing analysis as a foundation for interviews with customers and our sales team to refine how we think about packaging and what different personas need from Hex. The insights are already shaping how we approach pricing going forward.

There’s still a place for deep analyst collaboration, especially in reviewing the output to ensure that it can be trusted. But having the power to go from question to draft, solo and fast, was a game-changer. It let me focus on the fun part: making sense of the story in the data.

And in this case, the story will help us build a better product offering.