Why AI projects stall in the UK mid-market, and how vendors position around it

Almost every UK mid-market business has already had an AI project fail. In aibl’s survey of 755 UK mid-market leaders, failure is close to universal, and the causes are structural rather than technical. For a vendor, understanding why projects stall is the difference between selling into that pain and adding to it.

Failure is near-universal, and it is not the tools

Most buyers carry a recent AI failure, and it colours how they buy next. In the survey, the failures cluster around integration, returns that never materialise, and AI that never reaches a real decision, not around the technology itself falling over.

That matters for positioning. A pitch that assumes the buyer is a blank slate misreads the room. The buyer has been burned once and is wary of the next promise, so proof beats enthusiasm.

The policy trap stalls more projects than bad software

A written policy that no one enforces can return less than having none at all. In the survey, functions with a defined-but-inconsistent framework often score below those with no governance, because a framework on the intranet feels like the job is done and the money to enforce it never gets spent.

Vendors who sell a tool into that trap inherit it. The ones who help a buyer enforce, measure and own their AI are solving the thing that actually stalls projects.

The most active functions fail the most

Activity without proof is its own failure mode. Marketing, sales and customer experience are the most active AI adopters and report the highest failure rate, at 85%, alongside the lowest measurable ROI of the functional pillars at 45%.

For a vendor, that flags where the pain is sharpest and the attribution gap widest. Selling measurement into the busiest, least-proven functions meets a real and felt need.

No owner is the clearest predictor of failure

When no one owns AI, it does not pay. In the survey, AI with no single owner returns just 18%, against 62% where the CEO or C-suite owns it, a 44-point gap and the widest structural lever in the data.

A vendor can use this directly: helping a buyer name an owner and stand up accountability is often the unlock that makes the rest of the deployment work.

How vendors position around the stall

Lead with proof, not promise, because the buyer has failed before and knows the tools are rarely the problem. Position around the structural causes: enforcement, measurement and ownership.

Sell the conditions under which AI reliably pays, not another platform to add to the pile. In a market defined by near-universal failure, the vendor who prevents the next one is the one that earns trust and expansion.

Frequently asked questions

Get the free vendor guide: The AI vendor’s guide to the UK mid-market buyer →