Why do AI rollouts fail to pay off?
Short answer. An AI investment pays back when people change how they work, and that change happens team by team, under each team’s manager. Rollouts that stall tend to be run as technology projects: IT deploys the tools, HR runs the training, and nobody owns the business result.
What the evidence shows
- Access is high. BCG’s AI at Work survey (June 2026, close to 12,000 respondents) found that 74% of frontline employees use AI every day or a few times a week, up 23 points in a year.
- Direction is weak. Only a third of them say leadership’s communications about AI are clear, and 28% see a strong connection between what leaders say and what the organization does. Employees with a clear strategy and limited tools outperform those with strong access and no direction.
- Leaders’ support changes how people feel about AI. In BCG’s 2025 survey, positive feelings among frontline employees rose from 15% to 55% when leaders strongly supported AI use. About a quarter of them got that support.
- Managers decide how far tools reach. In a Hubble_s analysis of a sales force of about 20,000 people with the same tools and training, 56% of the sales reps who used the tools the way the best teams do reported to 13% of the supervisors.
What the teams that get a return do differently
In seven Hubble_s analyses (sales forces, a manufacturing plant and medical sales teams, at two global companies), two behaviors separated the best teams from the rest every time: transparency and trust, and focus on a few priorities. Tool knowledge was about the same: in Europe, 95% of the best teams said they knew how to use the tool, against 92% of the other teams.
Who should own an AI rollout?
Three owners, each for one part. IT owns the technical deployment. HR owns the human deployment: training, and a context that makes the right behaviors easy. The general manager, or the head of the function, owns the result and keeps asking whether AI is improving business results.
How should a company measure AI adoption?
By what AI lets each team produce, in three categories: doing the same work better, doing more of it in the same time, and doing things the team couldn’t do before. Logins say little. To find the teams worth copying, read adoption from intensive use of the features linked to results, and read performance against each team’s potential once external factors are removed.
How long until an AI rollout pays back?
Judge it over several years. AI is improving faster than most companies can absorb it, and few users get close to what today’s tools can do. The best single read of the return is whether your teams are keeping pace with the technology.
Read more: “Your AI rollout runs through a handful of your managers”.
Sources: BCG, “AI at Work: Strategy Matters More Than Tools” (June 2026) and “AI at Work: Momentum Builds, but Gaps Remain” (June 2025). Hubble_s analysis: a global consumer-goods company’s sales force, 2025; seven perform-to-potential analyses, 2024 to 2026.