Numbers mean nothing without context and conviction

Nell Thomas fell in love with data because of the human brain. In college, she was obsessed with how people made decisions, chasing that fascination into cognitive neuroscience, social psychology, philosophy, and computer science. Her sophomore year, she ended up in a psychology lab running experiments on humans.

"I swear it sounds weirder than it actually is," she says.

One day, a researcher found something strange in some EKG data. It was an unexpected pattern that seemed to challenge an existing theory. They developed an elegant explanation. It was compelling, counterintuitive, and publishable.

It was also wrong. A bug in the Visual Basic script had scrambled the row order. The finding was an artifact of broken code.

"People tell very powerful stories with data. So you’d better make sure the story is true," says Nell, who now leads Shopify's data organization—the team responsible for the infrastructure, tooling, and data systems behind Shopify and its millions of merchants. "You need confidence in the integrity of the methodology because it's easy to run with the wrong interpretation."

That early lesson from twenty years ago perfectly encapsulates how Shopify thinks about data: being "data-driven" isn't the goal. Because data doesn't make decisions. People do. The right data just makes their intuition sharper.

The power of context

Shopify has a deliberate philosophy when it comes to data. The company thinks in decades, not just in quarters. There’s a belief here that the overuse of data leads to people chasing the wrong things, micro wins that feel productive but ultimately pull you off course.

“We take such a long-term view of the company here, and there has been a real allergy to short-term optimizations," she says.

But long-term thinking and data don't have to conflict. You can take the long view while charting the current terrain. For Shopify, the goal isn't to swap intuition for dashboards, it's to strengthen intuition with better information.

"Data does not replace intuition, it sharpens it," she says. "And data doesn't kill nuance, it reveals it."

What does that mean in practice for Shopify? Every good decision needs to be grounded in context. Context is a very AI-laden word nowadays, but it’s a word that’s been loved by Shopify since its founding. Context gives you the actual circumstances of what's happening. And data is a way of providing context. But it should never become a crutch.

"Even if you build the most sophisticated analytics in the world, you don't want to deflect your own responsibility as a decision-maker onto the data,” she says. “The numbers don’t get to make the call. You still have to have conviction."

The targets trap

Shopify’s leadership has a deep respect for Goodhart's Law. Coined by economist Charles Goodhart in 1975, it states that when a measure becomes a target, it ceases to be a good measure. In other words, if you tell someone they'll be judged by a number, they'll game the system to hit it—even if it doesn't move the needle on what actually matters. They become more focused on winning than playing the game well.

"It's a gift to work with a founder like Tobi who has such a deep respect for this effect," she says. It shapes how measurement works within the data team. Nell’s careful about how she talks about targets, knowing the power they hold once they're in play.

"Constructing metrics correctly is difficult because it holds the power to dictate how people behave. If you want to build a 100-year company, you’re not going to get there if you only care about hitting a quarterly target," Nell says.

"Metrics can tell you what’s true today, but no metric is going to tell you what will be true 10 years from now, especially as AI reorders our world.” That gap is where human judgment lives.

So if metrics alone shouldn't drive decisions, and targets can be a trap, what does Shopify use data for?Infrastructure.

The foundations underneath 

Shopify has invested heavily in the infrastructure of its data—the foundations, the data quality, the canonical sources of truth. The company's data infrastructure processes an average of 13 petabytes per day; that's 13 million gigabytes.

"I've been maniacally focused on the infrastructure of our data," Nell says. “If we don't focus on the quality of our data and the systems that produce it, we won’t have great outputs. Everything we build is only as good as what it’s built on top of.”

Those foundations enabled something significant: data at Shopify isn't gated behind analysts. Nell’s team has democratized access to data across the company—not by removing safeguards, but by building tools that make data internally accessible without compromising security or privacy. 

Only a month after Anthropic announced Model Context Protocol (MCP) technology, the team had built an MCP for Shopify's data warehouse. It lets anyone at Shopify query internal data using natural language. Within a week, people who'd never written a query were using it to build entire slide decks. Now nearly 90% of Shopify employees not in the data discipline use data tools monthly, and 40% use data tools daily.

"More people using data means more people who scrutinize the work, put pressure on it, make it better," she says. "It gives people more context to do their jobs, and it's a great feedback loop for the data systems themselves."

And what that infrastructure can reveal is extraordinary. The data team is the custodian of decades of entrepreneurial journeys: millions of businesses launched, scaled, abandoned, and restarted by Shopify merchants. In 2025, there were billions of order transactions through Shopify. That data is precious, and Shopify treats it that way.

One pattern Nell loves: only half of the people who sign up for Shopify are new. The rest are repeat entrepreneurs, people who keep coming back to build something different. Merchants who have previously launched a business on Shopify and return to build a second one earn more than twice the sales per shop on average, compared to first-time founders. "This idea that great entrepreneurs often try and fail and then try again before they succeed, we can actually see that in the data," she says. 

Everyone says entrepreneurship takes persistence. Shopify's data shows exactly how true that is. It sharpened an intuition into a fact.

Why this matters now

The current moment in data is unlike anything in the past two decades.

"It is a game-changing level of intelligence that we have access to," she says. "We have unlocked a whole new level of what is possible."

In a world where AI models are becoming commodities, the underlying data and context become the real differentiator. Shopping is already shifting from scrolling and searching to conversation and intent. AI agents are beginning to shop on behalf of buyers. On Shopify, that means navigating a catalog of billions of products from millions of merchants. And that only works if the data underneath—the product catalog, the merchant information, the buying patterns—is reliable. Agentic commerce doesn't run on models alone. It runs on trusted data.

That's what Shopify has been building toward. And it circles back to a lesson learned twenty years ago in a college lab, watching a researcher build an elegant theory on broken code.

"Data people live and thrive on trust," she says. “As fast as things are changing, some things still hold true: having trust in the underlying data is critical. And correct, objective interpretation is critical."

Nothing is more merchant-obsessed than understanding how merchants actually use Shopify’s tools and applying that insight to build better ones. That's what data does for Shopify. It doesn't tell us what to think. It tells us what's real. What we do with that is still up to us.

That's the balance. Scrutinize the data. Understand the territory. Have conviction in your decisions. And never stop asking whether the story is true.