Ninety-five per cent of generative AI pilot projects are failing to deliver measurable business impact because leaders are allowing “fear of missing out” to drive disjointed strategies, a global technology executive warned today.
In comments released at the World Economic Forum Annual Meeting in Davos, Cai Ting, Chief AI and Data Officer at Rakuten, cited a recent MIT study to highlight the “concerning gap” between investment and returns that is plaguing the industry.
“Too many companies today allow the ‘fear of missing out’ to guide their AI strategies, resulting in costly, disjointed AI programmes,” Ting said. “AI transformation succeeds only when it moves beyond pilots into integrated operations.”
To escape this “pilot purgatory”, Ting unveiled a radical operational framework his company is using to force technology out of the lab and into the balance sheet.
The ‘Triple 20’ mandate
To ensure AI contributes to the bottom line, the Japanese e-commerce giant implemented a “Triple 20” initiative, a rigid set of targets demanding specific efficiency gains across the business.
The mandate requires a 20 per cent efficiency gain in marketing spend, a 20 per cent improvement in internal operational processes, and a 20 per cent efficiency boost for merchant and business partners.
“AI transformation requires specific, quantifiable targets that span divisions horizontally and teams vertically,” Ting explained. “Each BU, division and team can set their own KPIs to align with these North Star metrics.”
Hospitality meets hardware
Ting argued that successful transformation requires “AI-nisation”, a strategy that blends hard metrics with soft cultural values.
He pointed to the Japanese concept of omotenashi — hospitality that anticipates a guest’s needs before they ask — as a guiding principle for AI development. Rather than simply automating tasks, AI agents should “mindfully anticipate” user desires.
To achieve this, the company uses “forward-deployed engineers” — specialists sent directly into business units to break down silos — and a “data flywheel” strategy where human expertise validates machine outputs in a continuous loop.
“The true measure of AI’s success isn’t just in its technical sophistication… but in its ability to amplify human potential,” Ting concluded. “The true winners of the AI era will be businesses that master both innovation and operations.”
