Beyond Tools: Building True Data Culture

Want the full story? Listen to my complete interview with Sundar Swaminathan, where we discuss how seven slides saved Uber $30M in marketing spend and more.

Ever notice how every company claims to be “data-driven” these days? It’s become one of those buzzwords that’s lost almost all meaning right up there with “synergy” and “paradigm shift.” (I can practically hear you rolling your eyes.)

But here’s the thing: there’s a massive difference between having data tools and building a true data culture. And nobody demonstrates this better than Uber, where a junior analyst could shut down a $35 million marketing channel with just seven slides if they had the data to back it up.

Let’s dive into what actually makes a data culture work (and no, buying Tableau licenses doesn’t count).

The Myth of Tool-Driven Transformation

First, let’s bust a common myth: better tools = better data culture.

As Sundar points out: “You can do all of that with SQL and Google Sheets… Throughout my entire career at Uber, for the five years I was there, SQL and Google Sheets were the only two tools I used.”

(BTW, if investment bankers can make multi-billion dollar M&A decisions with Excel, you can probably analyze your marketing data without that fancy new BI tool.)

The truth is, tools are just tools. What matters is how you use them and more importantly, whether your organization actually trusts and acts on the data they surface.

The Four Pillars of Data Culture

Through my conversation with Sundar, four key elements emerged that separate true data culture from performative data theater:

1. Ego-Less Decision Making

“Data driven data obsession, whatever you want to call it, what it results in is an ego-less decision making,” Sundar explains. This means that good data trumps hierarchy, period.

At Uber, junior analysts could (and did) challenge million-dollar decisions if they had the data to back it up. This wasn’t just allowed; it was expected.

2. Process Over Perfection

Here’s something that might surprise you: data science isn’t about being perfect. As Sundar notes, “There’s a misconception that data science is designed to create perfection… it’s statistics, which has inherent uncertainty into it.”

Good data culture embraces this uncertainty. It’s not about having perfect information (you never will), but about having a reliable process for making decisions with the information you have.

3. Psychological Safety for Failure

Want to know a dirty secret about experimentation? “90% of experiments fail,” says Sundar. That means you need a culture where people feel safe failing nine times out of ten.

This isn’t just about being nice, it’s about being effective. If people are afraid to fail, they’ll only run safe experiments with predictable outcomes. And you can’t innovate playing it safe.

4. Democratized Access

At Uber, they had a tool called Query Builder that let anyone access and share SQL queries. But the magic wasn’t in the tool itself; it was in what it represented: complete transparency and accessibility of data.

“SQL queries would go viral at Uber,” Sundar recalls. Think about that for a second. When’s the last time you saw people getting excited about sharing SQL queries in your organization?

Building Your Own Data Culture

Okay, so how do you actually build this in your organization? Here’s a practical framework:

Step 1: Start with Leadership

Sorry, but there’s no getting around this one. “I truly think it has to be founder led,” Sundar emphasizes. Without buy-in from the top, you’re fighting an uphill battle.

If you’re not the founder, your job is to sell up. Focus on concrete wins and ROI rather than abstract benefits. (Those seven slides that saved $35M? That’s the kind of story that gets attention.)

Step 2: Build the Process

Remember PACE? Prioritization, Analysis, Communication, and Execution. Here’s how Sundar breaks down the time allocation:

• Prioritization: 1 hour/week
• Analysis: 3-4 days
• Communication: 1 full day
• Execution follow-up: 1-2 hours

Step 3: Create Feedback Loops

Data culture isn’t just about analyzing data, it’s about acting on it. Create regular touchpoints between analysts and stakeholders. As Sundar puts it, “I would force the stakeholders and my analysts to be in a room once a week at the beginning of the week.”

Step 4: Embrace Simplicity

Don’t get caught up in perfect attribution models or complex analytics stacks. Uber used last-click attribution far longer than most people would guess. Why? Because “last click is wrong, but it’s consistently wrong.”

Sometimes, good enough is good enough.

The Real Test of Data Culture

Want to know if you have a real data culture? Here’s the test: can a junior analyst with good data override a senior executive’s opinion?

If the answer is “no” or “it depends,” you don’t have a data culture yet. You have data theater.

But hey, that’s okay! Building a true data culture takes time. Start small, focus on process over tools, and remember: it’s not about having perfect data. It’s about making better decisions with the data you have.

The beauty of this approach? It scales. Whether you’re a two-person startup or a Fortune 500 company, the principles remain the same. Trust the process, embrace uncertainty, and let the data lead the way.

Just don’t expect to get it right the first time. Remember: 90% of experiments fail, and that’s exactly as it should be.

Want to hear more insights from Sundar Swaminathan, including how Uber built their brand data science team and measured ROI on a billion-dollar brand budget? Listen to the full interview here.