In the evolving landscape of e-commerce, the integration of artificial intelligence (AI) stands at the forefront of innovation. Owen Spencer’s experiences at Revelyst highlight a transition from initial skepticism to embracing AI as an operational enhancement. This article explores key insights from industry experts on navigating the AI implementation journey and the lessons learned along the way.
When Owen Spencer began discussing the potential of artificial intelligence for coding at sporting goods manufacturer Revelyst over a year ago, the immediate reaction was one of concern.
At the time, ChatGPT was gaining traction, and Spencer, then Revelyst’s senior director of DTC applications and AI enablement, noted that many employees worried AI might render their jobs obsolete. However, he observed a positive shift; engineers who once felt threatened are now collaborating across departments on more significant projects, such as assisting data analysts or developing applications.
“They’ve transitioned past that initial anxiety,” he remarked. “Initially, there was a sentiment like, ‘If this tool can write code, then my role is redundant.’ But they quickly realized that coding was merely one aspect of their responsibilities.”
Spencer was among hundreds of branding and retail professionals at eTail Palm Springs this week, discussing how AI is reshaping their e-commerce activities. Notable participants included J.Crew, PopSockets, and Hibbett. Revelyst itself encompasses brands such as helmet manufacturer Bell, outdoor equipment producer CamelBak, and extreme sports brand Fox.
Here are some lessons these brands have shared regarding their AI implementations thus far.
Align Teams Around AI Strategies
Tari Huddleston, global VP of digital and e-commerce at Revelyst, stated that the company is integrating AI throughout nearly all its operations, with the exception of marketing imagery. Before deploying these tools, she consulted with the head of legal to ensure a secure rollout.
“AI isn’t just a project; it’s a fundamental way of operating moving forward,” Huddleston explained.
Since adopting AI, Reveyst has broadened its internal testing and tools across various departments. Spencer, who has since departed to start a consulting firm called Friends with Robots, emphasized that involving teams early helped eliminate fears and promote unity. This was complemented by company-wide announcements, regular training sessions, and a SharePoint drive dedicated to AI learning resources.
“Part of our success in aligning teams around AI was due to the consolidation of brands within Revelyst, which was established as a standalone entity in November 2024 after separating from Vista Outdoor,” she noted.
“We didn’t unify under the same tech stack initially,” she continued. “We lacked shared categories or platforms. It’s been a year-long process to integrate everyone into the same systems, but now we have established a solid foundation to effectively leverage AI.”
Clean Your Data Before Feeding It to AI
J.Crew recently implemented AI-generated summaries of customer reviews. However, before this, the company actively sought more feedback from previous customers to enrich its data pool regarding product sizing and fit, as shared by Ruchika Jupali, J.Crew’s SVP of digital experience and technology.
“We didn’t initially realize we were gathering data for AI,” she admitted. “This principle remains vital: while technologies will continue to advance, having a robust proprietary data set is paramount.”
As AI gained traction following several A/B tests, J.Crew successfully deployed these AI-driven review summaries.
“Without investing in our review collection and enhancing those capabilities, we wouldn’t have produced high-quality summaries,” she noted. “Summarizing seven reviews is far less impactful than synthesizing seventy.”
Huddleston from Revelyst also emphasized the critical role of clean data in launching AI initiatives, noting that even minor discrepancies in file naming conventions can hinder the efficiency of information processed by AI systems.
“Invest time and effort into nurturing your data,” she advised. “Bad data won’t improve with AI intervention; it only exacerbates underlying issues.”
Prioritize Functionality Over Flash in AI Processes
The most successful AI processes are those that save time for employees or enhance customer experiences.
At Revelyst, a significant achievement involved scraping data from product manuals, video tutorials, and other site content, consolidating it into a text field interpretable by AI. This improved the chances of the brand being highlighted in ChatGPT results for user inquiries relevant to that content. The company collaborated with the AI platform Perplexity to streamline this information-gathering— a task that would typically require a dedicated team.
“Our strategy has been to identify and alleviate sources of friction within the organization from a business perspective,” Huddleston explained. “This benefits not only our customers but also our associates engaged in bringing these innovations to life.”
Jack Farrell, global VP of e-commerce at PopSockets, noted that the company scaled its Meta ad campaigns, achieving a 50% revenue and profitability increase the previous year.
Behind the scenes, PopSockets developed BEAM (Breakdown Engine for Ad Metrics), a system that evaluates which ads perform best and identifies optimal combinations of copy, images, and products.
“You can’t just instruct your design and creative teams to produce 500, 1000, or 2000 ads without a systematic approach,” he stated. “A structured feedback loop is essential.”
Spencer highlighted that the industry is currently in the “messy middle” phase of AI implementation. While clear applications for AI exist in chatbots, inventory forecasting, and website optimization, challenges remain in training teams and determining effective uses beyond superficial technology displays. His advice to companies is to slow down and reflect before taking action.
“Understand that this is just the beginning; even if you feel behind, we are only just starting out,” he said. “No matter your current status, keep progressing, remain curious, and take a moment to breathe— it’s easy to feel overwhelmed.”
In conclusion, the journey to integrate AI within e-commerce is complex yet rewarding. By aligning teams, ensuring data integrity, and prioritizing practical over flashy applications, companies can effectively harness AI’s potential to enhance operations and drive meaningful change in the industry.
