I had a call last month with the founder of a 45-person logistics company in the Midlands. A Big 4 firm had quoted her £48,000 for a 12-week “AI readiness assessment.” The deliverable? A PowerPoint deck telling her where AI could add value. Not the AI itself — just the deck explaining where it might go.
She was right to be sceptical. For £48,000, we implemented an AI chatbot that handles 55% of her customer queries, automated her delivery scheduling using predictive routing, and set up AI-assisted invoice processing. All three were live within 8 weeks. Total cost including my advisory time: under £12,000 for the first year.
This isn’t an unusual story. The enterprise consulting model was built for FTSE 250 companies with seven-figure technology budgets. It doesn’t scale down to SMEs, and it shouldn’t. UK small businesses need a fundamentally different approach to AI — one that starts with practical tools, not theoretical frameworks.
The consultancy trap: why big firms get it wrong for SMEs
Enterprise AI consultancies follow a predictable pattern. First, a discovery phase (4 to 6 weeks, £15,000+) where junior consultants interview your team and document your processes. Then, a strategy phase (4 to 6 weeks, £15,000+) where they produce recommendations. Finally, an implementation phase (3 to 12 months, £50,000+) where they build what they recommended.
For a company with £500 million in revenue and 5,000 employees, this makes sense. For a company with £5 million in revenue and 40 employees, it’s absurd. Your processes aren’t that complex. Your data isn’t that messy. And the AI tools available today are mature enough that you don’t need someone to spend 12 weeks telling you where to use them.
The SME AI playbook: start small, prove value, scale
Here’s the approach I use with every SME client. It’s not complicated, but it works.
Week 1: Identify three pain points.Not “areas where AI could theoretically add value” — actual pain points where your team wastes time or your customers get a poor experience. Common ones: answering the same support questions repeatedly, manually entering data from documents, writing the same types of emails or proposals, producing reports from multiple systems. If you want a structured way to identify these, the AI Readiness Quiz walks you through it in 5 minutes.
Week 2 to 3: Implement the easiest win. Pick the pain point with the simplest solution. Usually, this is either customer support (deploy an AI chatbot trained on your FAQs and documentation) or document processing (connect an AI extraction tool to your existing workflow). Get it live, measure the impact, and share the results with your team.
Month 2 to 3: Tackle the second and third projects. With one win under your belt, your team believes in the approach. Now you can take on slightly more complex projects: AI-assisted proposal writing, predictive analytics on your sales data, or intelligent workflow automation that handles exceptions instead of just following rules.
Four AI use cases that actually work for SMEs
I’m deliberately being specific here, because vague promises about “AI transformation” are part of the problem.
AI customer support. An AI-powered chatbot trained on your knowledge base can handle 40 to 60% of incoming queries without human involvement. Not the robotic chatbots of 2020 that frustrated everyone — modern AI chatbots understand context, hold natural conversations, and know when to escalate to a human. Typical cost: £100 to £500 per month. Typical saving: 20 to 30 hours of staff time per month. That’s a 3 to 5x return on day one.
Document processing.If your team manually extracts data from invoices, contracts, application forms, or delivery notes, AI can do it in seconds with 95%+ accuracy. The data flows directly into your accounting system, CRM, or spreadsheet. Implementation typically costs £1,000 to £3,000 and saves 10 to 20 hours per month for a business processing 200+ documents.
Sales intelligence.AI can analyse your CRM data to predict which deals are most likely to close, which customers are at risk of churning, and which leads deserve your team’s attention first. This doesn’t require a custom machine learning model — tools like HubSpot and Pipedrive have built-in AI scoring that works out of the box. The key is having clean data, which is where a fractional CTO adds value.
Content and communication. AI writing assistants can draft customer emails, proposals, job descriptions, and social media posts in your brand voice. The output still needs human review, but it cuts first-draft time by 60 to 80%. For a business that produces 20+ pieces of content per month, this alone can free up an entire day per week.
What it actually costs: an honest breakdown
Here’s what a realistic first-year AI investment looks like for a UK SME with 20 to 80 employees:
AI tools and subscriptions:£200 to £800 per month (chatbot platform, AI-enhanced CRM or helpdesk, document processing tool).
Implementation and integration:£3,000 to £8,000 one-off (setting up workflows, training AI on your data, connecting to existing systems).
Advisory and oversight:£1,500 to £3,000 per month if you work with a fractional CTO who manages the AI roadmap alongside your broader technology strategy.
Total first-year cost:£8,000 to £25,000. Compare that to the £48,000+ for a consultancy assessment that doesn’t include implementation. The maths speak for themselves.
Three mistakes to avoid
Don’t start with custom AI models.Unless your competitive advantage depends on proprietary algorithms, you don’t need custom machine learning. Start with off-the-shelf AI tools. You can always go custom later once you understand what drives value.
Don’t ignore data quality. AI is only as good as the data you feed it. Before implementing AI on your sales pipeline, make sure your CRM data is actually clean and up to date. A week of data cleanup often delivers more value than the AI tool itself.
Don’t skip the measurement.Every AI project should have a clear metric: tickets deflected, hours saved, accuracy rate, response time reduced. If you can’t measure the impact, you can’t justify the investment — or know when to scale it up.
Getting started this week
You don’t need a 12-week discovery phase to start using AI. You need 15 minutes to identify your biggest time sink, an afternoon to research the tools that address it, and a week to run a pilot. The AI readiness assessment gives you a prioritised list of opportunities specific to your business, and I’m happy to spend 15 minutes on a call walking through the results — no pitch, no obligation.
The gap between AI-ready SMEs and AI-lagging SMEs is widening every month. The good news is that catching up doesn’t require a big budget. It requires a willingness to start small, measure honestly, and build from what works.