Why Most Fitness Brands Don’t See a Return on Their AI Investment

When it comes to long-term success, 95% of AI projects fail, but don’t blame the technology, says PerfectGym’s Karl Foster

Artificial intelligence continues to dominate conversations across the fitness industry, but here’s the thing: while everyone seems to be investing in AI, how many are actually seeing a measurable return?

Not many, according to reports, but that doesn’t tell the whole story, insisted Karl Foster, head of AI at PerfectGym during his presentation at the ATN Innovation Summit. He encouraged people to look beyond the “uncomfortable truth” that AI fails to produce measurable ROI 95% of the time. “AI technology is truly transformative,” he said. “It just needs to be properly unlocked.” 

The biggest challenges aren’t technical, he said. The technology has proven value. “It’s everything outside the technology itself that holds things back.” Organizational readiness, leadership commitment and data quality are the key elements that determine whether AI projects scale or fail. 

According to Foster, AI isn’t a silver bullet and operators must be pragmatic in their expectations. Unfortunately, most organizations hit “pilot purgatory” and abandon AI projects long before they reach maturity. “Real value typically takes 18 to 36 months to emerge,” he said. 

This mismatch between executive expectations and AI’s natural adoption curve has become one of the industry’s biggest challenges.

Chief among those barriers is data, with at least 70% of projects noting data integration as the number-one problem. This includes fragmented software ecosystems with separate platforms managing memberships, marketing, payments and customer communications. As integrations weaken, data quality suffers. 

“Knowledge for AI is not just the data you have on your database, it’s also the knowledge you have on your FAQs, the knowledge you have on your social media platforms,” Foster said. “Gyms aren’t generally the best at updating this stuff. A marketing executive might forget to update the website, so if an AI agent is scraping all this [outdated] information, then your APIs are degrading over time.”

Yet it’s culture, not technology, that may be the biggest challenge, with employees often resisting AI improvements for fear of replacement. “Tech is easy; people are hard. Staff resistance is a big, big, big, big hurdle,” said Foster. “I can’t emphasize this enough. And if you don’t handle this properly with a change management program, people will reject the technology.” 

Organizations that position AI as a tool to augment employees rather than eliminate them are significantly more likely to get people on board for long-term success. The best framing is that “you’ll improve your human workforce output two or three times,” he said. 

Leadership commitment is equally critical. “AI’s ROI has a much higher chance of producing returns over 10% when the CEO is involved,” he said. “And it can’t be performative support. It needs to cascade to the rest of the C-suite, transfer down to teams and have constant follow up—accountability, KPIs, commission, all that good stuff.”

Build or Buy? 

The fitness industry also struggles with AI because it’s a “passion industry,” where most people in senior leadership don’t have degrees in engineering or mathematics—they just love fitness. “They don’t really connect the dots between data technology and the power it can unlock. But you need to understand the value behind it.” 

That begs the question whether companies should build their own AI system or invest in a third-party SaaS platform. There’s no one answer, but those who build themselves need to “get their data in order — centralize it, control it, then make sure it’s clean, secure and accessible.” 

Those who go with a partner need to ensure it’s one they can trust. “Work with someone who’s done all the groundwork on the tech side for you. It’s the quickest return in terms of value.”

Foster’s recommended investment strategy reflects a 10/20/70 rule, where 10% of an AI initiative focuses on selecting models and LLMs [large language models]; 20% goes toward infrastructure; with the remaining 70% centered on people — change management, training, communication and organizational adoption. 

Unsure where to begin? Foster advises addressing a single business problem rather than multiple initiatives. But while sales automation is an enticing entry point—with easy-to-measure ROI — member retention is actually the bigger long-term opportunity. 

“You can increase conversion, you can cut some sales agents out of the loop and save some money, but it’s not really going to move the needle,” he said. “My belief is it’s all in the member journey.”