How Abercrombie & Fitch Co. Optimizes Planning with AI

Lauren Parker
·5 min read

Amid shifting trends, uncertain supply chains and margin-decimating markdowns, inventory planning and allocation has never been more challenging. AI, however, can sharpen inventory planning and save money—if applied correctly.

“AI is helping our teams be more precise on how we deploy inventory across the globe,” said Molly Kozar, senior director, omni location planning, Abercrombie & Fitch Co. in the recent Sourcing Journal webinar AI Explained: Everything Planning Teams Need to Know (now available to view on demand). “Within the last few years, really, Abercrombie has been experiencing change—brand positioning, target customer, store footprints, the mix of digital versus physical, it’s all changing. AI is one of the tools that we have been able to leverage to give us that speed and flexibility to keep up.”

More from Sourcing Journal

That said, AI must start with a brand’s strategy and needs. There’s an expression that if the only tool you have is a hammer, you tend to see everything as a nail. With artificial intelligence, brands and retailers must pause before they metaphorically start hammering in screws. “I found it important to first identify the problem,” said Kozar. “And then chase if AI can help solve that problem.”

AI solutions also pull in an overwhelming amount of data, so it’s critical for teams to stay focused. “Just because the data is there doesn’t always make it useful,” said Kozar. “If weather doesn’t impact your product, maybe that’s not data that you need.”

But useful data is highly abundant. “[AI offers planning teams] data consolidation and data clean up,” said James Theuerkauf, CEO and co-founder, Syrup Tech, which supports Abercrombie & Fitch Co.’s AI efforts related to planning, among other major retailers and brands. “It also improves forecast specificity and forecast granularity, with the ability to go down to the style, color, size and location level. You’ve got the intelligent optimization of inventory decisions and placement and buys. That leads us to the workflow side…rebalancing of inventory across nodes, replenishing of inventory across the network, and then buying and assortment planning.”

While 60 percent of employees using AI worry about its impact on their jobs , according to CNBC, Kozar is a “huge proponent that AI is here to enhance your job.” Companies should look at AI as a blend of art and science. “Utilizing humans in lockstep with AI is truly where the magic happens. In my experience, tools are never a ‘set and forget.’”

AI models are only as smart as the information fed into it, so planners should still use gut instinct for instances where they know better, like a live store event that the AI model doesn’t yet know is happening. The follow-up step is to “get that data into the AI models” so they get smarter for the next store event at a different location. “The more we understand how the two can work hand in hand, the better, so it’s not AI versus your gut,” said Kozar.

Theuerkauf pointed to AI data sets—such as customer reviews that can inform returns forecasting, or marketing activities and ad spend that can complement historical data and real-time data—infuse demand forecasting with a 360-degree view.

That well-rounded view is critical for omnichannel retailers, which must gather and utilize both in-store and e-commerce data. E-tail organizations tend to buy on a more frequent schedule, as opposed to more traditional seasonal buys, and that frequent usage allows for more learnings. “All our models are built for the omnichannel world, so they factor things in like, buy online pick up in store, buy online return in store, buy online ship from store, which have made the potential combinations explode,” said Theuerkauf. “But that’s where AI shines. The more complexity there is, the more potential upside.”

He suggests companies and planning teams ask four questions to decide if AI solutions are right for them. One, does the problem that I’m trying to solve map to tasks that AI does? Two, does the outcome I’m looking for map to AI’s value (informed decision making, diversity of inputs, clarity of outputs and speed to insight)? Three, am I set up internally to embrace change and make the most of AI? Four, can I trust the provider with a solution to deliver quality results? One AI is implemented, it’s essential to set up KPIs to measure how well it is working.

For planners, selecting a starting use case for AI depends on their needs, be it applying historical data for new store growth, or rebalancing inventory to maximize full price sell through for stores with a style scarcity model. But in-season use cases make the most sense, Theuerkauf added, as you can “measure the baseline versus uplift in in a matter of weeks.”

With more than 700 stores to plan for, gathering and parsing data in quick time is invaluable for Abercrombie & Fitch Co. “We opened 60 new experiences last year and expect to continue to be a net store opener again in 2024, and AI is a way that we can further understand those new experiences,” said Kozar. “It’s critical for those locations to be able to get that information quickly so we’re able to adapt and make sure our customer is getting exactly what they need.”

It’s also been important for Abercrombie & Fitch Co. to “determine what will or won’t work through testing, learning, reevaluating and implementing.” Additionally, added Kozar, “ensuring that your team has the resources and the support for the change is also really important. Clearly defining what part of the process and what level of the organization the information is going to benefit helps for a smooth transition.”

This might all feel like a brave new world, but one thing is clear.  AI is only growing, and retailers run the risk of getting left behind if they continue to misunderstand or fear it. Theuerkauf stressed such qualms are misguided. “AI can seem like magic, but the reality is, it’s just math.”

To watch the webinar on demand, click here.