AI Consulting for Small Businesses: What It Actually Costs and What You Get

If you run a business with 10 to 200 people, you have almost certainly had the following experience in the last twelve months. Someone — a board member, a client, a competitor's LinkedIn post — mentions AI. You nod. You know you should probably be doing something about it. You have no idea what that something is.

So you Google "AI consulting for small businesses." And what you find is either enterprise firms with offices in Canary Wharf quoting six figures for a "discovery phase," or freelancers who built a chatbot six months ago and now call themselves AI consultants.

Neither is what you need. And the gap between the two is where most small businesses get stuck. They don't need a £200,000 digital transformation programme. They don't need a chatbot. They need someone to walk in, look at how the business operates, point to the three things that are costing the most time or money, and fix them.

This piece is an attempt to fill that gap — plainly, with real numbers, and without pretending that AI consulting is more complicated than it actually is.

What AI Consulting Actually Means

Strip away the jargon and AI consulting, at its core, is someone coming into your business, understanding how it operates, identifying where AI would save you the most time or money, and then either building the solution or telling you exactly what to buy and how to configure it.

That's it. It's not mysterious. It's not a research project. It's the same thing a good operations consultant does, except the tools they recommend happen to involve machine learning instead of spreadsheets.

The reason it feels opaque is that most AI consulting firms have a financial incentive to make it sound complicated. Complexity justifies the fee. If the client doesn't fully understand what's being done, they can't challenge the price or the timeline. This is a problem across the industry and it's why most small businesses end up doing nothing — the perceived barrier to entry is artificially high.

For a business your size, the engagement is typically straightforward. There are four steps: audit, recommend, build, measure. Each one has a clear output. Each one has a fixed price. You decide at the end of each step whether to continue to the next. There is no open-ended commitment.

What It Actually Costs

This is the section most AI consultancies won't write. Publishing prices makes it harder to charge whatever the client will bear. But if the principle is transparency, then transparency has to start with cost. Here's what the UK market actually looks like for businesses with 10–200 staff.

What You Actually Get

The audit should produce a document you can act on without the consultancy's help if you choose to. This is important. If they hand you a strategy deck full of buzzwords and no specific recommendations, you've been sold smoke. A good AI strategy report for a small business tells you:

Which processes to automate first — ranked by impact and ease of implementation. Not a list of 20 things you could do. The 3 things you should do, in order.

What the expected saving is — in hours or pounds, not percentages of undefined baselines. "This process currently costs you 40 hours per month at £35/hour. Automation reduces it to 4 hours. Saving: £15,120 per year." That level of specificity.

What tools or builds are required — and whether you need custom development or off-the-shelf configuration. Some problems are solved by connecting two tools you already have. Some need something built. The audit should tell you which.

What it will cost to implement — fixed price, not a range that spans £10,000 to £100,000. If the consultant can't scope the work precisely enough to price it, they don't understand it well enough.

How long before you see results — and what "results" means in measurable terms. Not "you'll feel more efficient." Specific numbers you can check against.

The build phase should produce working automation connected to your actual systems. Not a proof of concept. Not a demo on a laptop. Software that runs inside your business, processing real data, saving real time, measured against the numbers from the audit.

The retainer, if you choose one, should produce one new automation every 4–6 weeks with a monthly report showing specific numbers. If the monthly report is vague, the retainer isn't working.

How to Choose an AI Consultant

The market is flooded with people who learned ChatGPT six months ago and now call themselves AI consultants. Some of them are brilliant. Most of them are not. Here's how to tell the difference.

Ask for case studies with numbers. Not testimonials. Not "we helped a client in the manufacturing sector." Specific numbers: hours saved, cost reduced, revenue gained. If they can't produce them, they haven't done it. Testimonials without metrics are marketing. Case studies with metrics are evidence.

Ask what happens after the audit. This is the question that separates the strategists from the builders. Some firms do brilliant audits and then hand you a report you can't act on without hiring developers. The best firms audit, build, and measure — end to end. You want the firm that says "we'll build it and prove it works," not the one that says "here's what you should build."

Ask about their own AI. This is the test most consultants fail. Do they use AI inside their own business? Have they built AI products? If they're selling AI consulting but running their own operations on spreadsheets and manual processes, that tells you everything you need to know about how deeply they understand the technology.

Ask for a fixed price. AI consulting for a business your size should not be billed by the hour. Hourly billing incentivises the consultant to take longer. Fixed pricing incentivises them to be efficient and to scope the work properly upfront. If they can't scope the work and price it before starting, they don't understand it well enough.

Ask what they'd tell you NOT to automate. Any consultant who says everything should be automated is selling you a solution, not solving your problem. The good ones will tell you which processes aren't worth the investment — because the time saving is too small, the process is too variable, or the technology isn't reliable enough yet. Honesty about limitations is the strongest signal of genuine expertise.

The Real ROI Pattern

Across the businesses we've worked with, the pattern is remarkably consistent. It doesn't matter whether the business is in manufacturing, professional services, distribution, or legal. The economics follow the same curve.

The audit costs £2,500–£5,000 and identifies £80,000–£200,000+ in recoverable value. That's not a typo. The gap between what a business spends on manual processes and what those processes would cost automated is almost always larger than the leadership team expected going in. The reason is simple: nobody has ever measured it before. The cost of manual processing doesn't appear as a line item on the P&L. It just shows up as salaries, overtime, and the vague feeling that the team is stretched.

The first build costs £8,000–£25,000 and pays for itself within 6 weeks. This is the proof point. If it doesn't pay for itself within 6 weeks, either the audit was wrong or the build was wrong. Either way, you know quickly and cheaply.

The retainer produces compounding returns because each new automation builds on the infrastructure from the last. The first project might automate invoice processing. The second connects that to your reporting. The third adds client onboarding. Each one is faster and cheaper than the last because the foundation is already in place.

The businesses that get the most value are the ones that start with the audit, prove the concept with one build, and then scale. The ones that get the least are the ones that try to automate everything at once, or skip the audit entirely and go straight to buying tools they don't need.

What a Bad Engagement Looks Like

You'll know you've hired the wrong consultant if any of the following happens:

They spend four weeks "discovering" your business before recommending anything. A good consultant can identify the top 3 opportunities in two weeks. Four weeks of discovery means they're either padding the engagement or they don't know what they're looking for.

They deliver a 60-page strategy document and call that the output. A document is not a deliverable. A working system is a deliverable. If the primary output of the engagement is a PDF, you've paid for expensive advice, not a solution.

They can't name the specific hours or pounds the project will save. Vague promises about "improved efficiency" and "streamlined workflows" are a sign that they haven't done the work to understand your numbers. Good consultants deal in specifics.

They recommend tools you've never heard of and can't explain why your existing tools won't work. Most small businesses don't need new platforms. They need their existing tools connected properly. If the first recommendation is "rip out what you have and start again," that's a red flag.

They create dependency. If you can't run the system without them, they haven't built a solution — they've built a subscription to themselves. Good consultants train your team and hand over documentation.

How to Start

The first step is not buying anything. The first step is measuring. Pick the one process in your business that you know wastes the most time. Ask the person who does it how many hours it takes per week. Multiply by their hourly cost. Multiply by 52. That's the annual cost of that process.

If that number is over £20,000 — and it almost always is — then AI consulting will pay for itself on the first project. If it's under £20,000, you probably don't need a consultant yet. You need a better tool or a process redesign.

The businesses that get this right treat AI consulting the same way they treat hiring a good accountant or a good solicitor. They check credentials. They ask for evidence of results. They agree a fixed scope and price. And they measure the outcome against what was promised.

It is not more complicated than that.

Want to see what the process looks like in practice? Read about where to start with AI strategy, or take the AI Readiness Assessment — it takes 3 minutes and shows you where the biggest opportunities are.