SpanForge Cost Model™
The true cost of pilot purgatory extends far beyond what appears on a budget report. Organisations typically account for only the direct costs of abandoned pilots — compute, vendor fees, and contractor hours. These visible costs represent only a fraction of the total impact.
Most AI programme cost is invisible
The direct costs of an abandoned AI pilot — compute, vendor licences, contractor fees, and internal staff time — are rarely trivial. For mid-market organisations these costs are measured in hundreds of thousands of dollars; for enterprise-scale programmes, in millions. But direct costs are only part of the picture.
The larger burden sits below the surface, rarely appearing on a budget report. This invisibility is precisely why it keeps compounding: what does not appear on a financial statement does not get managed.
“Most AI programme cost is invisible on financial statements. That invisibility is why it keeps compounding.”
Visible costs vs. hidden costs
The SpanForge Cost Model™ separates what appears on a budget from what actually determines the outcome of an AI programme. Cost categories are illustrative, drawn from publicly available industry benchmarks.
Cost categories and ranges are illustrative, drawn from publicly available industry benchmarks. Estimates; individual results will vary.
The four hidden cost categories
Deferred Business Value
Every month a pilot spends in purgatory is a month the business value it was designed to capture is not being realised. A 12-month delay on a customer-facing initiative targeting a 15% reduction in service handling time compounds losses far beyond the direct programme cost.
Credibility Rebuild Investment
When AI pilots fail visibly and repeatedly, investment appetite contracts. AI becomes politically associated with broken promises rather than competitive advantage. Rebuilding sponsor confidence after a series of high-profile failures takes time and organisational effort that is rarely budgeted for. The cost is real, even if it does not appear on a balance sheet.
Competitive Displacement
Competitors who have solved the delivery problem are compounding production-system value while others cycle through abandoned pilots. Closing the resulting gap becomes more expensive with each passing cycle.
Internal Staff Opportunity Cost
Senior engineers and data scientists assigned to failing pilots are not available for initiatives that could ship. The cost is not the salary — it is the foregone value of work that did not happen.
Sunk cost is not the primary risk
Sunk cost is not the primary risk of a failed pilot. The primary risk is the credibility damage that makes the next initiative harder to fund, staff, and deliver. Each failed pilot erodes appetite for the next, reduces tolerance for ambiguity, and increases political resistance to AI investment.
“The majority of AI programme cost is invisible — and therefore unmanaged.”