Technology is moving fast, but is fashion keeping up? Too often tech strategies fail not because the systems themselves are lacking but because companies aren’t taking a holistic, connected approach. To leverage AI’s full capabilities, the apparel industry must make the right information visible to the entire organization and bridge the processes across product, planning and merchandising.
A panel of retail technology leaders from Anaplan, VibeIQ, Centric Software, Optimum Retailing and Impact Analytics came together to unpack how AI and data are reshaping the future of merchandising and how to overcome the challenges companies face in implementation.
One major roadblock is legacy systems. In fact, tech platforms’ biggest competition doesn’t come from competitor software solutions, but the spreadsheet.
“I’m more and more surprised at how many retailers are still on spreadsheets; retailers with 1,000s of stores,” Kyle Kaechle, VP of planning & pricing, North America & Asia, Centric Software said, noting the consequences go far beyond inefficiency. This is particularly problematic in a large company with different departments, where they each end up in their own silo. “So when you’re trying to make decisions, you come together, and everyone has a different set of data and a different set of processes, and so it’s almost impossible to get to a decision.”
In today’s volatile retail environment, that lack of alignment becomes a serious risk. “You can’t pivot unless you have a platform and you can make quick decisions,” he said. And it takes a “single source of truth” to allow you to connect your strategy to execution.
Once that’s established, the focus becomes capturing data in real time.
“[You need] the ability to have a good ingestion system that can bring data in from different sources—internal, external, first, second, third party, whatever—and then with the ability to do that in real time,” Bob Debicki, GM, Consumer Strategy, Anaplan, said. Traditional batch processing, done weekly or nightly, is no longer sufficient. “We need [data] to be dynamic.”
But structured data is one thing; handling unstructured inputs is another challenge, he said. “You start getting into things like imaging, content that maybe doesn’t map cleanly, and so you need a way of not only storing that information, but grabbing the attributes and making sense of that in your planning process.”
AI and machine learning are also evolving. And while everybody’s thinking about AI and ML, it’s not just about bringing that information into your environment, “but making sense of it and allowing that information to grow and to learn,” he said.
Breaking down silos
Just gathering data isn’t enough, of course, especially if the quality isn’t good, isn’t applied properly or the users can’t understand it to make the tools deliver.
“If you want to do anything with AI or planning, you need to create data your company can use,” Brian Lindauer, co-founder and CEO, VibeIQ, said, noting that so much data is inaccessible, or “wrapped up in places that your company cannot use.”
He added companies need to also pay attention to user experience when it comes to selecting a new tech platform. The easier they are to use, the more likely the company will embrace them. Moreover, Lindauer pointed out teams are often working with tools like Airtable or SmartSheet, which are familiar and easy. The problem is they’re disconnected from the systems that manage your enterprise data. “If we want to do more with AI and more with planning and more with automation, we need to address that gap. Where we work, where people think, and capture ideas needs to be a lot closer to the enterprise systems that manage structured business data.”
A major challenge of connecting planning, product and merchandising teams is that siloed fractured systems create fractured results.
“This is an age-old problem,” said Jennisa Visram, vice president of retail strategy & partnerships, Optimum Retailing. “Tools aren’t speaking to each other.”
Rather than advocating for a single system, she emphasized integration so that each team can lean into the tools that are optimized for their roles without losing the necessary collaboration and data flow across the organization. “I don’t think there actually is one tool that every [team] should be using, but rather, how do we start to think about this from a connectivity standpoint?”
This extends to the retail floor, where valuable customer information is gleaned but then lost. “The store team [is] the frontline… but where does that feedback go? Oftentimes it’s anecdotal, it’s untimely, and the way it goes back up [through the company, it’s] not really triaged to the right place,” Visram said.
Retailers need an active feedback loop to move from strategy to execution to results.
Rethinking Agility in Planning
Another value-add for AI in merchandising is the need to shift agility and move beyond seasonal planning.
“We’re running into the topic of season versus seasonless,” said Lindauer, noting how deeply seasonal structures are embedded in retail systems. “There’s so much technology out there that’s built on top of the past.”
He suggested a new model may be needed, although no one has truly figured it out yet. “I think there’s something there around merchandising and planning, having one way of representing the world and the data and what their market plan is, and making sure that you have allowance for development to move at a different pace, and then finding a way to bridge them together.”
These issues exist even as companies continue to push into new markets, which require tech solutions that don’t increase operation complexity.
The first consideration, according to Jeff Gadawski, chief revenue officer, Impact Analytics, is data attribution, specifically identifying attributes that can give a company’s models a good starting point for learning what that forecast should look like.
Next is figuring out how to manage differences between markets spanning North America, EMEA and Asia-Pacific. Where there are commonalities across regions, brands and retailers are able to leverage processes, solutions and workflows. Beyond that it demands a more bespoke approach. “I would say 85% of it’s the same, but then there’s going to be nuances,” he said. “Those market nuances are more of a challenge from a standpoint of how you’re successfully delivering value across markets for a global brand.”
Global brands, he emphasized, must balance consistency with localization. “It’s not the one size fits all.”
Getting buy-in
Getting the C-suite and executive leaders to adopt tech is just the first step. Getting the rest of the team to implement new technology is also challenging—and essential.
Leadership has a major role beyond budget approval, agreed panelists. But leadership must also help users knock overcome the pitfalls they’re bound to face. Without executives forcing accountability, the end users won’t adopt the process and implementation is going to fail.
“The magic is when leadership agrees, and the passion of the end users kicks in,” said Debicki. “Then you’re off to the races.”
But no matter how good the intentions, technology can fail if a company is too rigid. “Quite often I’m seeing tech come in and just try to put one footprint and one process on top of everyone, and that can reject the system,” said Lindauer.
Visram added that misalignment is another major issue. Adopting technology for technology sake isn’t the answer. She said the organization must start with a problem-solution standpoint and the key players must be incorporated into the decision making.
“Sometimes the technology fails because it doesn’t end up being used by the people or in the manner in which the problem actually needed to be solved,” she said. “[You have to ensure] that we’re actually solving the problem for them rather than creating new ones.”
