TL;DR

Roughly seven in ten executives judge their AI deployments a success. Among the practitioners actually operating those tools, barely half agree — 69% against 53% in research from Certinia. Only a fifth of organisations say results have beaten what they expected, and employee distrust of AI output now ranks as a leading adoption barrier for a third of firms.

The split within professional services

The sector breakdown is the most useful part. Consulting firms were judged successful by 55% of respondents; in audit, accounting and tax that falls to 43%. The explanation is not mysterious. Advisory work is largely unregulated territory where a flawed output gets caught in review, whereas in assurance a hallucination carries consequences with a regulator attached. Lower satisfaction in the higher-stakes discipline is a rational response, not a technology failure.

Certinia’s chief executive DJ Paoni made a related point about what actually separates firms: identical tools are producing wildly divergent outcomes, and what separates the organisations extracting value from those generating noise is barely related to purchase volume at all.

Distrust as a measurable cost

The barriers cited are trust, governance and skills — notably not budget. Where programmes have stalled, the report puts more weight on absent skills and staff distrust than on either funding or leadership backing. That reframes the problem: a stalled rollout is usually a people problem being misdiagnosed as a procurement one.

Certinia’s proposed remedy, unsurprisingly for a systems vendor, is connected systems across sales, delivery, finance and customer success. The claimed returns are real enough as correlations — 57% reporting better customer satisfaction, 56% higher productivity — but the survey does not disclose its sample size, and a supplier finding that the answer is its own product category deserves the usual discount.

Looking forward

This sits alongside a run of similar UK findings, including this week’s evidence that executives and practitioners diverge on what AI is delivering. The pattern is consistent enough now to act on. UK firms measuring AI success purely through leadership self-assessment are measuring the most optimistic available number — the practitioners running the tools daily are the better instrument, and the 16-point gap is roughly the size of the error being made.