An operations-focused argument that AI implementations fail not because of the technology but because organizations underinvest in process customization and adoption work. McKinsey data cited: for every $2 spent on AI technology, $3 goes to process customization and $5 to scaling and adoption; 89% of companies fail to get value from AI deployments, and S&P Global found 42% of companies abandoned most AI initiatives in 2025, up from 17% the year before. Michigan health system mergers (Beaumont/Spectrum into Corewell, Henry Ford absorbing Ascension Michigan hospitals) are used as an extreme case of duplicated systems and definitions that AI models paper over rather than reconcile. Staffing benchmarks (25 people per 1,000 workers, half technologists and half operations) and case studies (Cleveland Clinic sleep-risk tool, Fetch.ai's event bus rebuild) illustrate the point. Four questions are proposed to surface hidden execution costs before funding AI projects.
Table of contents
Michigan Health Systems Don’t Have Legacy Systems. They Have Two of Everything.Half of a Good AI Team Doesn’t Work in AI.A Pilot Is a Demo. A System Is Something Else.Here Are 4 Questions Worth Putting in Front of the Room.Questions this post answers
What percentage of companies fail to get value from their AI implementations?
Roughly 89% of companies fail to get value from their AI implementations, according to McKinsey data. Separately, S&P Global found that 42% of companies abandoned most of their AI initiatives in 2025, up from 17% the year before, meaning the abandonment rate doubled in twelve months even as the underlying models improved. Teams weighing whether to fund an AI initiative can track adoption data like this through daily.dev before committing budget.
How should a company budget spend between AI technology and the surrounding implementation work?
A McKinsey spend breakdown for AI deployments that succeed suggests roughly $2 on the AI technology itself, $3 on process customization, and $5 on scaling and adoption, for every $10 spent overall. In other words, for every dollar spent buying AI, budget roughly four more dollars to make it fit the organization. Anyone building a business case for AI spend can follow ongoing coverage of adoption economics on daily.dev.
What staffing ratio do McKinsey benchmarks recommend for organizations running AI projects successfully?
Successful organizations staff roughly 25 people per 1,000 workers on AI projects, split evenly between technologists and operations staff. The operations half comes from the functions actually doing the work today, such as clinical operations, revenue cycle, and supply chain, rather than from an innovation office or outside vendor consultants. Operations leaders scoping AI staffing plans can follow related workforce and adoption trends on daily.dev.