video
No timeouts in retail
In athletic retail, every hour delivers a verdict. Timely insights can be the difference between a colorway that converts or sits, store shelves that stay stocked or run thin.
“We work in a business that’s super fast-paced,” says Jenny Moore, Under Armour’s Director of Regional Planning, “especially within our direct-to-consumer business, where things are happening sometimes every hour.”
Behind that pace sits a sprawling enterprise that spans wholesale, e-commerce and owned retail around the globe. At its core is a vast data surface — from inventory, supply chain and demand planning to sales and product creation — that feeds an even larger appetite for answers. When a missed insight can mean lost revenue or a market moment gone, visibility into all this data is a critical competitive advantage.
And AI has only turbocharged the speed.
“AI is moving at a speed no one’s ever seen before. Having a platform, an ecosystem and a capability to meet that speed is important — because the business is moving just as fast.”Patrick Duroseau, Chief Data and AI Officer, Under Armour
But in 2020 — long before Under Armour employees had access to conversational AI — the company’s data couldn’t keep up with the velocity of the business.
A typical workday looked something like this: An analyst opens a workbook. Then another. Then a third. They’re trying to reconcile numbers that should match but don’t, because every team has built its own way of pulling data. A leader is asking for a read on last week’s business — but the answer requires a half-day of data prep. By the time the report is in hand, the moment has passed.
Teams were hampered by a fragmented landscape and slow, inflexible systems that couldn’t scale, meaning silos stacked up while shadow IT proliferated in the gaps. Different teams arrived at different answers to the same question.
“A lot of the time was spent in data preparation as opposed to reading the information to drive a decision,” says Denny Ward, Director of Enterprise Data Analytics at Under Armour. “An executive needs insights at a specific point in time. If you miss the opportunity to have the right information, you lose the momentum when everyone is engaged to make a decision.”
A monument in the making
When Under Armour set out to build a new data foundation, the team started with the business, not just the technology.
“As a technologist, I understand why people tend to lead with technology,” says Under Armour’s Chief Data and AI Officer Patrick Duroseau. “But the critical component for us was starting from a business lens, asking how we get the technology to achieve our business outcomes.”
This perspective shaped how Duroseau approached the Snowflake partnership, treating it as what Under Armour calls a “monument” — a pivotal investment that paves the way for future innovation.
“If you’re a monument for us, it means you have the foundation we want to build on, continually adding features and functionality. Snowflake is one of our monuments: Instead of being locked in from a technology perspective, we are able to grow and scale as technology evolves.”Patrick Duroseau, Chief Data and AI Officer, Under Armour
Snowflake mapped to Under Armour’s continually expanding ambitions. “What elevated Snowflake to being a partner for us was their interoperability and extensibility,” Duroseau says. “Not only in the technology, but as a partner.”
From that decision came a single, unbending architectural commitment: one data platform.
“Snowflake has provided us with a single source of truth for analytics, AI and business intelligence,” Duroseau says. “Capabilities like secure data sharing and data clean rooms have been a huge unlock for us to drive value and speed. With this consistent foundation, we’ve been able to deliver faster responses and better results while staying true to our tenet of not moving data.”
Daily reps replace monthly marathons
Once on Snowflake, the first thing to go was the wait.
With Snowflake as its data foundation, Under Armour overhauled critical processes like product margin reporting. Before, this report required a week of manual work — pulling, reconciling, formatting — that was so labor-intensive it only ran once a month. Thirty days would pass before anyone could review current margins, meaning product creators were calling plays off last month’s info. Today, this automated process runs daily, enabling faster pricing and assortment decisions that protect revenue.
Leadership reporting, which used to take a day or two to compile by hand, now arrives in executives’ inboxes every morning, giving leaders a reliable pulse on the business to inform key timely decisions.
But one of the most critical shifts has been to put Under Armour’s data within reach of anyone who needed it, not just the people who knew how to wrangle it.
| Process | Before |  Snowflake |
|---|---|---|
| Product margin reporting | A week to produce. Run once a month. |
Automated. Delivered daily. |
| Leadership reporting | Days to compile by hand |
In execs’ inboxes by morning |
| Data access | Static dashboards | Live, shared data |
The downstream effect: Inventory, supply chain, demand planning, point-of-sale and sales teams now operate from the same numbers, helping departments surface insights that streamline operations and ultimately boost revenue. “Previously everyone was trading static workbooks and trying to reverse-engineer formulas,” Ward says. “Now we’re looking at live, real data sets and growing data literacy across the business as a result.”
And the persistent problem across regions — different teams, different numbers, same question — vanished. “By creating this streamlined data structure and reporting with Snowflake,” Moore says, “we have a one-stop shop for our employees to all be looking at the same information and the same answers.”
“One of the most rewarding things about being on Snowflake is building out a self-service ecosystem so coders and non-coders alike have access to data to make products of their own, in a governed way.”Jeff Leineweber, Director of Data Platform Engineering, Under Armour
Protect this house — with governance
With reporting reinvented and data within reach, Under Armour set its sights higher: conversational AI every employee could use — and trust.
“Data governance is a core pillar of what we do,” Duroseau says. “That means securing the data, ensuring it’s properly structured and accessible, but always predictable when anyone accesses it.”
To accomplish this, the team spent five years building observability, tightening governance and creating data models reliable enough that a leader could pull up a number and act on it without second-guessing the source. “Snowflake has allowed us to build that governance around our data and, at the same time, enable AI capabilities for our end users,” says Jeff Leineweber, Director of Data Platform Engineering.
This governance strategy centers on a distinct pattern: the golden report. “We line up any AI-generated insight to a golden report that the business uses on a regular basis,” Ward says. “Everything behind that insight is anchored in the metrics and definitions of that golden report, which gives us the trust that, as they dig deeper and get a recommendation, they know it’s starting from a trusted source.”
The golden report
The trusted source every AI answer is anchored to.
When trusted data becomes a daily advantage
Ask, and the data answers
“Making data accessible is no longer just about creating more dashboards,” Duroseau says. “It’s about allowing teammates to interrogate that data in a way a static dashboard couldn’t answer — and deliver insights in a way they can take action against more holistically.”
Thanks to Snowflake CoWork, Under Armour knowledge workers now have a personal agent that gives them instant access to AI intelligence, on top of governed sources to prioritize accuracy.
“Snowflake CoWork has been a game changer for us to truly enable our end users with conversational AI in a controlled manner,” Leineweber says. “It also has allowed us not to build our own conversational interface from scratch, which would have required a lot of time and maintenance.”
On the technical side, Snowflake CoCo provides developers with their own coding agent, helping them build faster and serve the business sooner. “Six months ago, CoCo was not even in existence. But as of three months ago, it has transformed how we operate and deliver internally,” Duroseau says. “CoCo has simplified the way we engage with our data, interact and deliver outcomes. We’re able to distribute that capability to teammates across the entire organization.”
“By leveraging CoWork and CoCo together, we’ve opened up access to insights at all levels — from the front lines all the way up to the executive suite. It’s a foundation for us to adapt quickly, meeting the business where it is while moving at the speed of AI.”Patrick Duroseau, Chief Data and AI Officer, Under Armour
These days, a regional planner doesn’t open a workbook or wait for a query to run. She simply types a question into CoWork. By the time the first meeting starts, she already knows where the business stood yesterday — and what to do about it today.
“Using CoWork instead of manual processes has freed up an enormous amount of time for our team,” Moore says. “Instead of spending time pulling data, we can focus on finding the white space market opportunities for Under Armour to continue to grow.”
This access to insights reverberates from Under Armour’s C-suite to its newest hires. “CoWork empowers our leadership team to ask that first initial question based on metrics,” Ward says. “The question that gets pushed to an analyst is a much deeper insight, where they can actually get to the real root cause of the problem.”
Meanwhile, employee onboarding — once a multi-week process of orientations, documentation and meetings — now takes just hours or days with the ability to ask questions, learn workflows and understand the company’s operating model. “The faster teammates get on board, the faster they can actually contribute to the organization,” Duroseau says.
Agents in the lineup
Ask anyone on the team what’s next, and the answer is the same: agents.
Under Armour’s first agentic workflow — the AI-powered internal help desk — is already paying dividends. “Our agentic workflows are driving financial benefit operationally while also providing a better self-serve user experience for our teammates,” Duroseau says. “Snowflake was a key enabler for success, providing us with the interoperability to leverage unstructured data and add capabilities on top of our foundation.”
“The next chapter isn’t about technology. It’s about the speed of business. As Snowflake continues delivering on key partnerships and innovating at the speed the technology is moving, we’ll be well positioned to serve our organization and be a couple steps ahead.”Patrick Duroseau, Chief Data and AI Officer, Under Armour
What’s coming: more agents, across more functions. “Any agent we build will require good data, whether structured or unstructured, as reference points. Snowflake helps us add agents and assistants at speed and scale, leveraging our foundation and learnings,” Duroseau says.
For Ward, the excitement is in the velocity of the work itself. “Conversational AI with rich context behind it — that’s going to enable speed in our business as well as decision-making. And being able to bring a data product to life in a matter of days instead of weeks or months.”
As Under Armour maps out its next strategic plays, Snowflake is key to staying agile and competitive.
