The numbers we are working against

McKinsey surveyed 1,793 executives in January 2018. A further 7% said performance improved but the improvement did not hold — meaning roughly three quarters saw neither lasting gain nor a durable capability.

An analysis of 1,471 IT projects found an average cost overrun of 27%. But one in six ran 200% over on cost and almost 70% over on schedule. The risk is not the average; it is the tail.

Gartner attributes the abandonment to poor data quality, inadequate risk controls, escalating costs and unclear business value — none of which are model problems.

Gartner cites escalating costs, unclear business value and inadequate risk controls. The pattern repeats each hype cycle, which is an argument for scoping to a recoverable failure mode.

Based on 7,600 consumers across 14 European countries. Up from 63% in 2020 — the highest abandonment recorded since the series began in 2016, despite steady investment in digital onboarding.

An average calculated across 50 separate studies of ecommerce checkout abandonment. The figure has barely moved in a decade, which suggests the causes are structural rather than cosmetic.

An automated scan found 56.1 errors per page on average, up 10.1% year on year. Six recurring issues — led by low-contrast text and missing alternative text — account for the great majority.

The same report recorded the first increase in wasted cloud spend in five years, which Flexera links to the cost complexity introduced by AI workloads and new managed services.

The 85% target decomposes as 90% availability × 95% performance × 99.9% quality. The gap between that and a typical plant is mostly invisible until the measurement itself is trustworthy.

From a survey of nearly 5,000 practitioners, 90% of whom use AI at work. DORA found AI adoption correlates positively with delivery throughput and negatively with delivery stability.

DORA reports a direct correlation between the quality of that platform and an organisation's ability to get value from AI — adoption alone is not the differentiator.