Saks Global Holdings, the American parent company of luxury department stores and commercial properties, was formed in November 2024 after the spin-off of Hudson’s Bay Company’s U.S. assets and expanded with the acquisition of the Neiman Marcus Group a month later. With this transformation, Saks Global is setting the stage for a new era of retail innovation powered by data and AI.

In a business landscape increasingly enamored with the promises of artificial intelligence, one foundational truth remains unchanged: Without strong data underpinnings, even the most advanced AI initiatives are destined to falter.

In this three‑part series, Veronika Durgin, Vice President of Data at Saks Global, engages in an incisive conversation with Gautam Singh, Head of Analytics, Data, and AI at WNS Analytics, to unravel the strategic essentials of building AI‑ready enterprises.

In Part One, Durgin brings a pragmatic lens to the hype and zeroes in on what she calls the “cake layers” of AI readiness, with data foundations as the essential base.

The myth of the AI shortcut

Durgin jumps straight into one of the most common pitfalls she sees in enterprises: the urge to “just do AI.”

“So many organizations rush into AI, where companies decide, ‘Oh, we’re just going to do AI,’ and then they just try to put in the use case for it.”

While it may feel agile, she warns, it typically results in “just salt pilots and endless vendor POCs and just a lot of frustration.”

She draws parallels to previous tech waves, when the rush to cloud and data science both showed how organizations frequently leap without laying the groundwork. Durgin comments:

“Data science initiatives struggled to scale beyond proof of concept, not because the algorithms were wrong; it’s because the data foundations weren’t there.”

For AI, she warns, the stakes and the hype are even higher. The danger is not AI itself; it is the underlying data readiness.

“AI is not the magic that we wish it was, but it should fit into the real workflows and deliver real value,” she adds.

The layered cake of AI readiness

To articulate her framework, Durgin turns to a cake metaphor. The layers, she explains, represent the essential building blocks of AI maturity:

• Data foundation: “Do we trust our data? Is it governed? Is it modeled? Is it documented well enough that teams know what they’re looking at?” Without this, fragmented or conflicting definitions get amplified by AI.
• Accessibility and infrastructure: She stresses the need for secure and scalable platforms, adding, “It might mean modern data platforms. It has to have clear ownership.”
• Business alignment: “Do we know what problems we’re trying to solve? Can we tie those to measurable business outcomes?”
• Culture and skills: Perhaps most critical, she argues, is the organizational capability to experiment responsibly.

“AI, just like data, is a team sport. You need data, product, engineering, compliance, and business stakeholders. They all have to be in the loop and in lockstep,” she asserts.

The silent killer of AI projects

When asked about the biggest obstacles to getting AI projects off the ground, Durgin points to fragmentation and data silos. “Most organizations underestimate how hard it is to bring data together in a way that anybody can use it,” she notes.

Despite decades in the field, the problem persists across industries. “Data lives in dozens, if not hundreds, of systems. Each system has its own rules, formats, and putting it together without losing meaning and context is hard.”

The other traps are trust and quality. “If data isn’t consistent, if it’s stale, if it’s poorly defined, AI will just learn those flaws and then amplify them.” She emphasizes the danger of “garbage in, garbage out, with AI it’s just now at scale.”

Context is king, and so is control

Another vital layer is semantics and business context. “AI needs to understand what data means so that the output is more appropriate to the use case,” says Durgin.

This could involve building semantic layers or clearly defining business terms. Privacy and access controls are also non‑negotiable.

“You need to balance accessibility with compliance and privacy. It’s the balance between data democracy and data dictatorship.”

Durgin summarizes her stance succinctly: “You need a real business problem to solve, ensure data is accessible for a specific use case, and cross‐functional teams.”

Centralised data, decentralised analytics

When it comes to the perennial challenge of unified data access in fragmented environments, Durgin takes a clear position. “I am a believer in centralized data, decentralized analytics,” she says.

In her view, success comes when one team owns the data — responsible for creating common definitions and semantic clarity — while business units are responsible for how they use it.

“Everybody owns data, but nobody’s responsible for it. When we centralise data and there is a team responsible for it, it’s easier to ensure common definitions, therefore, easier to define guardrails and access rules.”

Resisting the tech hype

Speaking about navigating a fast‑changing tech landscape, Durgin offers a word of caution.

“We should stop focusing on technology. We all chase hypes, but sometimes there’s a lot of promise, other times, there’s FOMO.”

She introduces a nuanced idea: “bridge solutions.” These are short‑term solutions that deliver value now while giving companies time to decide on long‑term architecture.

Ultimately, Durgin says, it’s not about tools. “It’s more about having a growth learning mindset, building a culture of experimentation and innovation, and staying hyper‑focused on solving business problems.”

Only once those cultural and foundational pieces are in place should companies “bring the next new thing,” she concludes.

Disclaimer: The interviewee’s insights are personal and not representative of any current or past employer.

CDO Magazine appreciates Veronika Durgin for sharing her insights with our global community.