Everyone Has AI, Almost Nobody Gets Value From It
Quick answer
McKinsey research shows 88% of companies now use AI in some form, but most are "wide but shallow" — scattered experiments that never create real value. BCG says only firms that redesign processes end-to-end see 60%+ cost reductions. Here's what UK SMEs should do instead of bolting AI onto broken workflows.
McKinsey research shows 88% of companies use AI in some form, but most are "wide but shallow" — scattered experiments that never create real value. BCG says only firms that redesign processes end-to-end see 60%+ cost reductions. Here's what UK SMEs should do instead.
I see this all the time with UK SMEs. Somebody goes to a conference, sees a demo of an AI tool, comes back buzzing. They buy it. They switch it on. Three months later it's sitting there doing nothing, and nobody can explain what it was supposed to fix in the first place.
Sounds familiar? Yeah. It's everywhere.
The numbers are brutal
McKinsey's latest State of AI research found that 88% of companies now use AI in at least one business function. That sounds brilliant, right? Nearly everyone's on board.
But here's the kicker. Harvard Business Review's July 2026 research calls what's happening in the US "wide but shallow" — companies have AI scattered everywhere, but almost none of it is creating actual, measurable value. It's surface-level adoption. Checkbox stuff.
BCG's July 2026 research on AI-first enterprise operations backs this up. They found that AI can drive cost reductions of 60% or more — but only when organisations redesign their processes end-to-end first. Not bolt AI on top. Not sprinkle it around. Actually rip the process apart and rebuild it with AI in mind.
Wide but shallow. That's the trap. Everyone's got AI somewhere. Almost nobody's got it where it counts.
What "wide but shallow" looks like in practice
I worked with a manufacturing company that had five different AI tools running. A chatbot on the website nobody used. A Power Automate flow that summarised emails nobody read. A Copilot licence for the MD who never opened it. An forecasting tool plugged into data that was 40% wrong. And a reporting dashboard that showed numbers three people in the company disagreed with.
Five AI tools. Zero measurable value. Thousands a month in licences. That's wide but shallow.
The problem isn't the tools. The tools work. The problem is they're sitting on top of processes and data that were never designed for them. It's like putting a turbo on a clapped-out engine — more power, same old problems, just faster.
Why it matters for UK SMEs
If you're an SME, you haven't got money to waste on vanity AI. Every pound needs to show a return. The BCG research is clear: the companies seeing 60% cost reductions didn't buy more tools. They redesigned the workflow first, then slotted AI in where it actually removed manual effort.
HBR's comparison between US and Japanese firms is telling. The US has massive adoption numbers but struggles with depth. Japan has lower adoption but, in the firms that do use AI, tends to build it deeper into core operations. Two different paths, and the "adopt everything now" approach is producing a lot of expensive shelfware.
Three things to actually do
- Pick one painful, repeatable process. Not five. One. Something a human does every week that follows the same steps — invoice processing, report generation, data entry between systems. That's your target.
- Fix the data underneath before you add AI. If the source data is messy, AI just gives you confident wrong answers faster. Clean it, define it, get it consistent. This is the unglamorous bit that makes everything else work.
- Redesign the process around the AI, not the other way round. Don't just drop an AI tool into the existing workflow. Map the process, cut the steps that AI can handle, and rebuild the flow so the human only touches the exceptions.
That's it. No magic. No £50k platform. Just one process, sorted properly, with clean data and a workflow that actually makes sense.
The companies getting value from AI aren't the ones with the most tools. They're the ones who did the boring groundwork first. Same as it's always been with technology, really.
References & Further Reading
- Harvard Business Review, U.S. and Japanese Companies Struggle with Different Parts of AI Adoption (July 2026)
- BCG, AI-First Enterprise Operations: Reinventing the Operating System of Work (2026)
- McKinsey, The State of AI (2026)
- BCG, The Future of the AI-Powered Transformation Office (July 2026)
Getting value from AI?
I help UK SMEs figure out which process to fix first, get the data sorted, and build AI in where it actually pays for itself. No vanity tools, no shelfware.
Let's have a chatMatty Hatton is the founder of Digital Adaption, an ERP and data consultancy based on the Wirral. He has spent 15 years delivering ERP transformations for manufacturers, including leading the data migration on a £4.5m consolidation of four legacy systems onto a single Infor LN cloud instance for a 220-user group. He holds an MSc in Digital Transformation and IT Strategy from Manchester Metropolitan University and is Microsoft PL-200 certified.
Start with a 30-minute data risk call