Everyone Has AI, Nobody Gets Value From It
Quick answer
HBR and BCG research shows most firms have AI but can't scale it. The fix isn't more tools — it's picking one painful workflow, getting the data right, and proving value before expanding.
HBR and BCG research shows most firms have AI but can't scale it. The fix isn't more tools — it's picking one painful workflow, getting the data right, and proving value before expanding.
I see this all the time with UK SMEs. They've got Copilot switched on, maybe a chatbot on the website, someone's had a play with ChatGPT. The board gets a presentation saying "we're doing AI." Everyone nods.
But when you ask what's actually changed — what process runs faster, what decision is better, what hours have been saved — it goes a bit quiet.
That's not a failure. That's what "wide but shallow" looks like.
The research is brutal
Harvard Business Review published a piece in July comparing AI adoption across the US and Japan. The headline finding: 88% of US companies use AI in at least one business function. That sounds brilliant, right? But HBR's own analysis described the deployment as "wide but shallow" — companies struggle to create real value from it.
BCG's latest research backs this up. They found nine in ten CEOs say they're seeing initial value from AI. Brilliant. But when BCG dug into what "value" meant, most of it was productivity tinkering — drafting emails faster, summarising meetings, the odd chatbot. Almost none had redesigned an actual business process around AI.
Having AI and getting value from AI are two completely different things.
That gap — between having it and getting value from it — is where most UK SMEs are stuck right now.
Why everyone's stuck
Here's what I see, almost every time:
- Tool-first thinking. Someone buys the AI tool, then goes looking for a problem to solve with it. That's backwards. It should be: find the painful process, then see if AI can help.
- No workflow redesign. Bolting AI onto a broken process just makes the broken process a bit quicker. BCG's data shows firms that redesign workflows around AI see 3x the productivity gains of those that just add tools.
- Data's a mess. AI is only as good as the data it pulls from. If your ERP data is inconsistent, your customer records are duplicated, and your product codes don't match across sites — AI just gives you confident wrong answers faster.
What to actually do
Right, enough of the diagnosis. Here's what I'd do if I were running an SME today:
- Pick one painful, repetitive process. Not five. One. Invoice processing, quote generation, stock reconciliation — something where people are manually moving data around. Start there.
- Get the data for that process clean. Not all your data. Just the data that feeds this one workflow. Make sure part numbers match, customer records are deduplicated, the fields are consistent. This is the boring bit that makes everything else work.
- Redesign the workflow around AI, don't just add to it. If you're using AI to draft quotes, change the process so the AI drafts, a human reviews, and the quote goes out. Don't keep the old manual process running alongside "just in case." That's how you get shallow deployment.
The firms getting real value from AI aren't the ones with the most tools. They're the ones who picked a process, sorted the data, redesigned the workflow, and proved the value before expanding.
That's it. No magic. Just proper business analysis, done properly, on one thing at a time.
References & Further Reading
- Harvard Business Review, U.S. and Japanese Companies Struggle with Different Parts of AI Adoption (July 2026)
- BCG, How CEOs Can Scale AI Value Across the Enterprise (July 2026)
- McKinsey, The State of AI (2026)
- Stanford HAI, AI Index Report 2025 — Economy
Stuck in AI pilot purgatory?
I help UK SMEs figure out which process to fix first, get the data right, and actually prove AI value — not just tick a box. One workflow at a time.
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