Helping AI succeed in retail: Turn pilots into measurable impact

New AI tech might be smart, strong and intriguing. But for retail organizations, even the most promising tools can fall short of expectations if they aren’t supported by the culture, human judgment and trust needed to make AI work in practice. 

“Bigger budget, more people, none of that’s going to solve this,” said Julie Averill, chief impact officer, Gold Thread LLC, and author of the recent book, “Chief Impact Officer.” 

Averill was speaking to retail digital, marketing and technology leaders at NRF Nexus 2026 about why AI pilots fail and what organizations can do to create measurable impact, and highlighted data showing why AI success remains elusive for many organizations. MIT reports that 95% of all generative AI pilots fail to make any measurable impact. RAND reports that more than 80% of AI projects fail. Companies are quietly abandoning their AI initiatives and asking how others seem to be succeeding. 

The secret? No matter how strong the technology, she said, it’s still no match for human insight, judgment based on experience, and solid culture. It’s less about the tools, and more about the conditions that allow for greatest impact. 

Averill is a former global CIO and executive vice president at lululemon, and helped the company grow from $2 billion to $10 billion over eight years. She also led a technology transformation at REI and helped create the omnichannel playbook at Nordstrom. At Nexus, she presented three succinct steps toward creating innovation at scale and at speed. 

Her message for retailers was direct: AI impact depends less on adopting the newest tools and more on creating the organizational conditions that allow people to use them well. 

1. Build psychological safety before scaling AI 

AI is providing some companies the “perfect excuse to justify big layoffs,” she said. “It makes it seem like, ‘It’s not our decision — it’s the technology.’” This creates fear. Averill pointed to more than 100,000 jobs lost to AI-related layoffs in the first six months of 2026. 

“I wonder, in our organizations today, how many people are feeling this incredible pressure to keep up, to do more and to stay quiet?” she asked. “But I contend, if you create an environment where those fears can be spoken out loud, and can be talked about in a way that they can be solved, and you create community, that’s the foundation for being able to scale an organization.” 

Averill encouraged those in the audience to “audit for fear” as a first step. “Ask a difficult question in your team. Ask, ‘What are the things that we’re not saying out loud that are preventing us from being successful?’ And then the first person who speaks, thank them for that. Create an environment where people can share their fears and feel safe in doing so.” 

2. Use influence to drive AI change 

Stop waiting for formal authority to drive change. “The authority that we’re waiting for is never going to come,” Averill said. 

Instead, focus on influence. “Influence is what helps us be successful,” she said. “I know that’s difficult for many of us who studied engineering or computer science. I mean, why do we have to worry so much about influencing human beings? We came from the technology. That is no longer the job. The job is influence.”  

The action step here is a relationship audit. “Take your top three initiatives. And name every person who could kill them. And then say, ‘What’s my relationship with each one of those people?’ Spend your influence budget before you need it.” 

3. Recognize culture as the infrastructure for AI success 

Technology may be increasingly commoditized, but culture still stands alone. 

"The technology that we create today is no longer special,” Averill said. “We’re in a competition. We’re in a race, and we’re consuming the same models.” Culture, however, can operate at scale. “I have seen it transform organizations into curious, nimble, risk-taking, success-hungry organizations,” she said. 

“Nobody can have this. Nobody can copy that. It is special. It is unique. And you need to understand that this is the infrastructure. Culture is not soft. It is an infrastructure that your entire organization runs on.” 

She encouraged those in the audience to name culture as an asset — and received a question in return: How to get other leaders to embrace it, too. 

“Proof,” she responded. Leaders persuade others by demonstrating culture’s results. “Walking is so much better than talking.” For retailers trying to move AI from pilot to impact, culture must show up in measurable outcomes.