We've been working with SMEs on AI adoption for 18 months now, and the pattern we're seeing is striking enough that McKinsey's latest State of AI report (1,993 respondents across 105 countries) seems to confirm what we've been watching unfold in real time.
88% of organisations say they're using AI, but only 6% are seeing real business impact at the bottom line. The gap isn't closing, which raises questions about what's actually working and what isn't.
You'd expect larger companies with bigger budgets and dedicated data science teams to dominate, but when McKinsey looked at what actually separates winners from everyone else, budget and headcount weren't the deciding factors. Two factors stood out: workflow redesign, where high performers are 2.8 times more likely to fundamentally redesign how work gets done, and leadership commitment, where high performers are 3 times more likely to have senior leaders who truly champion AI initiatives.
Here's what we've learned working with dozens of SMEs: small companies should be better at both of these things, and yet most aren't capitalising on what makes them different.
Two structural advantages for SME's
A 200-person company has 200 processes, 15 approval layers, and entrenched systems that have been running for a decade. You have a sales process you can sketch on a whiteboard in 20 minutes. When you spot something that needs fixing, you can test the fix this week. When something doesn't work, you can pivot tomorrow rather than waiting for quarterly reviews and stakeholder alignment.
Then there's the leadership advantage, which operates differently at scale.
In an enterprise, AI transformation requires EVP buy-in, SVP alignment, VP approval, Director implementation, Manager adoption, and team execution. That's six layers between decision and action, and each layer introduces delay, translation loss, and the possibility that the original intent gets diluted.
In your business, it's probably founder decides, team executes. Maybe one layer between, which means you can move from conviction to action much faster.
The McKinsey data backs this up. High performers fundamentally redesign workflows. 55% do this compared to 20% of typical organisations. They don't bolt AI onto existing processes, which tends to produce marginal improvements at best. Instead, they rebuild processes around what AI can actually do, which often means rethinking the work itself.
If you're willing to actually redesign rather than just augment, you can move faster than most. The question is whether you'll commit to real change or just incremental tweaks.
McKinsey found this is the single biggest differentiator. High performers are 2.8 times more likely to fundamentally redesign workflows. Making old processes faster has diminishing returns. The real opportunity comes from building entirely new processes that couldn't exist without AI.
What this actually looks like
Here's what we do with clients who move from pilots to real impact.
The first conversation is always with the founder or CEO, and we focus on what you're actually trying to achieve, which often isn't what people initially say when asked.
Are you trying to grow faster? Serve customers better? Free up your team for higher-value work? One client told us they wanted to 10x their content output. We asked why. Turned out they didn't need more content at all. They needed better targeting, which was a different problem entirely and required a different solution.
You can have this conversation in an afternoon. Enterprise clients need three months of workshops to get to the same clarity, partly because more stakeholders need to weigh in and partly because the real goals are often obscured by political considerations.
Then we help you pick one or two high-ROI areas to start. Usually marketing and sales if you're focused on growth (McKinsey found 67% of companies report revenue increases from AI in this function), customer service if you're optimising operations, or knowledge management if you want quick productivity wins that build confidence in the technology.
The key is picking something concrete rather than trying to use AI everywhere.
This is where most pilots fail, and it's worth understanding why. People add AI to existing processes instead of rebuilding the process, which limits the potential impact to incremental improvements.
Most companies approach sales AI as help reps write emails faster, which treats AI as an add-on. The redesign question is different: if AI can qualify leads, generate custom proposals based on the prospect's industry and pain points, and identify high-intent buyers, what does the rep's job actually become?
The answer, in many cases, is that reps only touch high-value negotiations and everything else runs automatically. Whether that's the right answer for your business depends on your sales cycle and customer relationships, but at least it's a different question.
That's the difference between adding AI and redesigning for it. You can test this in weeks because you don't need six layers of approval to change how your sales team works, assuming you're willing to make real changes rather than cosmetic ones.
Once one workflow proves ROI, you expand. The data shows high performers invest 35% of their digital budgets in AI, compared to 7% for typical organisations. That's a nearly 5x difference, which compounds over time.
You don't need enterprise budgets, but you do need to commit meaningfully. Better to do three things properly than ten things half-heartedly. We see SMEs spreading £50k across eight different tools and getting no impact anywhere. The ones who win put that same budget into redesigning two critical workflows completely, which produces measurable results they can build on.
The opportunity is narrowing
Right now, only 29% of small companies (under £100 million revenue) are scaling AI, compared to 50% of companies over £5 billion. That gap means there's still room to move, but probably not for long.
Large companies are investing 35% of digital budgets into AI. They're redesigning workflows to be AI-native. They're slower than you, but they're not standing still. McKinsey's data shows 62% of organisations are already experimenting with AI agents, autonomous systems that can plan and execute multi-step workflows. Only 23% are actually scaling them yet, which suggests this is early. But it won't stay early for long, especially as the tooling matures.
We work with you through implementation. Actual workflow redesign, tool selection and integration, team training, and iteration as you learn what works. Some things will work immediately. Others won't, which is why the iteration matters.
If you're stuck in pilot mode or trying to figure out where to start, let's talk. The gap between companies capturing real value from AI and those burning budget on experiments is widening. You have structural advantages. You can move faster than enterprises. But you're also competing against companies investing 5 times more budget and redesigning their entire operations around AI.
Book a free assessment and we'll show you exactly where AI can drive real value in your business.
.png)
.png)
