AI Doesn’t Start in Your IT Department. It Starts with You.

AI leadership in small business comes down to one variable: whether the owner personally uses and champions the tools. Not the technology stack. Not the budget. Not the IT department. McKinsey’s 2025 survey of 1,933 companies confirms it — the businesses getting real ROI from AI are 3x more likely to have senior leadership that doesn’t just approve AI spending but visibly owns it.

We’ve watched this pattern play out across every client we work with at SOHOB. The companies pulling ahead aren’t the ones with the biggest tech budgets. They’re the ones where the founder opens ChatGPT before they open their email.

94.5% of Companies Are Getting This Wrong

Here’s the number that should keep every business owner up at night: 88% of companies now use AI in some form, according to McKinsey — but only 5.5% achieve meaningful EBIT impact from it. That’s not a technology failure. That’s a leadership failure.

PwC surveyed 4,454 CEOs in 2026 and found 56% report zero ROI from AI — no increased revenue, no reduced costs. Only 12% achieved both. These aren’t companies that skipped AI. They bought it. They approved the budget. They hired the consultants. They just didn’t lead it.

The signals are everywhere:

• BCG: 72% of CEOs are now the primary AI decision-maker — doubled in one year. The C-suite is realising AI isn’t something you delegate.
• HBR: 93% of AI leaders cite human factors, not technology, as the primary barrier to adoption.
• UK DSIT: 71% of non-adopting businesses say “no identified need.” That’s not a tech gap — it’s a leadership imagination gap.

We wrote about why most AI automation projects fail before they start last month. The root cause is the same: unclear ownership at the top.

What AI Leadership Actually Looks Like

We talk about “AI leadership” as if it’s a strategy document. It’s not. It’s behaviour.

Anthony, a property business owner featured on Goldman Sachs’ 10,000 Small Businesses panel in March 2026, started by using AI for his own emails. Not a company-wide rollout. Not a six-month implementation plan. His emails. Then he ran weekly one-on-one sessions with every employee, walking them through how he was using the tools. His business doubled its turnover in six months.

BCG calls this archetype the Trailblazer. In their 2026 survey of 2,360 executives, Trailblazers spend 8+ hours per week on personal AI upskilling and allocate 60% of their AI budgets to workforce training. Followers allocate just 24%. Trailblazers don’t outsource AI literacy. They model it.

Joe, founder of Marmalade Marketing, had a client demand double the output at a third of the price using AI. She said no. Then she enrolled in Oxford’s AI programme. Now she guest-lectures on AI adoption at Manchester University. The path from sceptic to advocate ran through personal experience — not a vendor pitch deck.

The pattern across every case is the same: visibility and personal commitment matter more than technical expertise. Your team doesn’t need you to understand transformer architectures. They need to see you using the tools daily.

When CEO Enthusiasm Backfires

Here’s where we part ways with the “just get excited about AI” crowd. Enthusiasm without fundamentals is a PowerPoint deck, not a strategy.

That PwC figure — 56% zero ROI — includes plenty of enthusiastic CEOs. They weren’t sceptics. They were believers who confused buying tools with adopting them.

The perception gap is worse than the spending gap. HBR found that 76% of executives think their employees are excited about AI. The actual number? 31%. When leadership mandates AI without listening to the people who’ll actually use it, you get what researchers call performative compliance — people going through the motions while quietly reverting to old workflows.

Allison, an operations leader on that same Goldman Sachs panel, put it plainly: “If the team haven’t identified the problem, the problem doesn’t exist.” The model that actually works:

• CEO sets direction and demonstrates personal commitment
• Teams identify where AI applies to their specific workflows
• Everyone iterates together — co-creation, not mandates

That’s not soft leadership. That’s the only kind that produces what Deloitte calls “deep transformation” — something only 34% of companies have achieved despite tool access increasing by 50% year over year.

What To Do Monday Morning

This isn’t a twelve-month roadmap. It’s a week.

Start using AI yourself. Draft your next three emails with it. Write a proposal. Summarise a report. Don’t delegate this to an intern or an IT manager. The point isn’t efficiency — it’s visibility. Your team notices what you use.

Run a weekly 30-minute session with your team. Pick one workflow. Explore it together. Anthony didn’t roll out a company-wide AI platform — he sat with each person, one at a time. 98% of Goldman Sachs’ 10KSB programme graduates are now using AI because the programme made leaders go first. Once you’re ready for your first real workflow, our guide to automating client onboarding with n8n is a practical starting point.

Ask your team one question: “What’s the most tedious part of your week?” That’s your first AI use case. Not the most strategic. Not the most impressive. The most tedious. The BCC reports that 54% of UK SMEs now use AI — and the ones getting value started with grunt work, not grand strategy.

As Rishi Sunak framed it at that Goldman Sachs session: all you need is a laptop, an internet connection, and a growth mindset. We’d add one thing — the willingness to go first.

The 94.5% Are Missing a Founder, Not a Framework

The companies getting nothing from AI aren’t under-tooled. They’re under-led.

We hold a position at SOHOB that some CTOs will disagree with: the single highest-leverage AI investment a small business can make isn’t a platform, a consultant, or a training programme. It’s the founder spending thirty minutes a day using the tools themselves — visibly, imperfectly, in front of their team.

The 5.5% who’ve cracked meaningful returns aren’t smarter or better-funded. They have a leader who decided AI adoption was their job, not someone else’s. The rest are waiting for a technology solution to what is, and always has been, a leadership problem.

Frequently Asked Questions

Does the CEO need technical skills to lead AI adoption?

No. BCG’s Trailblazer CEOs differentiate themselves through commitment and visibility, not coding ability. Anthony from Groundup Property started with AI for emails — no technical background required. What matters is using the tools daily and being seen doing it.

How much time should a business owner spend on AI each week?

BCG’s research shows the most successful AI leaders spend 8+ hours per week on personal AI upskilling. Start with 30 minutes daily — use AI for your own work, then share what you learn with your team in a weekly session.

Why do most companies fail to get ROI from AI?

McKinsey found only 5.5% of companies achieve meaningful financial impact from AI. The primary barrier isn’t technology — 93% of AI leaders cite human factors. Most organisations buy AI tools but don’t invest in leadership commitment or workforce training.

Should AI adoption be led by IT or the CEO?

The data says CEO-led. McKinsey’s survey found businesses with strong senior leadership ownership are 3x more likely to be AI high performers. Delegating AI to IT without visible executive commitment produces what HBR calls performative compliance — usage without genuine adoption.