Only 23% of HR leaders can currently demonstrate the ROI of their organisation’s AI investments using hard data, according to live polls at Keka HR Katalyst, the enterprise PeopleOS firm’s flagship virtual conference held on 23 July 2026.
That leaves nearly three in four unable to show clear returns. The rest said they can only partially measure AI’s business impact, have not yet established clear ROI, or are not measuring it at all. The Summer Edition, the sixth run of Keka’s HR Katalyst, drew close to 8,000 HR professionals and business leaders alongside more than 25 speakers from companies including the Adani Group, Reliance Jio, Raymond, PVR INOX, Ather Energy and LSG Sky Chefs.
The polls sketch a sector caught between adoption and accountability. More than half of respondents, 53%, said their organisations are still experimenting with AI while its return remains unclear. Just over half, 50.5%, named the choice of success metrics rather than the technology itself as the biggest barrier to proving ROI. And when asked how confidently they could answer their CFO on returns from HR-technology spending, 48.8% admitted they could offer only rough estimates rather than measurable outcomes.
“HR leaders are no longer asking ‘How do we deploy AI?’ They’re asking ‘How do we measure whether it’s making our organisation better?'” said Tapan Acharya, Chief Revenue Officer at Keka. “That’s the conversation that will define the next phase of workplace transformation.”
The findings land in a market where AI adoption has raced ahead of proof. Indian enterprises have poured budget into AI across hiring, performance and workforce planning, yet the Keka data suggests measurement frameworks have not kept pace. The gap matters most when finance leaders start asking harder questions, and the near-half of HR heads leaning on rough estimates points to a looming credibility test for people functions.
Governance surfaced as a parallel worry. In one session, 58.6% of respondents said their biggest concern about giving AI greater authority in leadership decisions was the loss of human accountability, while only 3.8% said they would fully trust AI to make such calls. Half the audience reported changing their position after the debate, a sign of how unsettled the question remains.
