Most articles about AI automation pricing will tell you "it depends" and then ask you to fill in a contact form. That's not helpful when you're trying to work out whether this is worth pursuing before you talk to anyone.
So here's an honest breakdown of what AI automation actually costs for small businesses in 2026, what drives the price up or down, and how to work out your ROI before you spend a penny.
Why AI automation pricing is so hard to find online
There are two reasons pricing is so opaque in this space.
First, most agencies don't want to anchor you to a number before they've had a chance to upsell. Vague ranges protect them, not you.
Second, automation genuinely does vary a lot. A simple email triage workflow is a different beast from a full transaction coordination system with CRM integrations, document generation, and exception handling. Pricing them the same way doesn't make sense.
But that's not a reason to give you nothing. Most small business owners come to us having read three articles that said "anywhere from $500 to $500,000" and are no clearer than when they started. I'd rather give you honest ranges with context than protect myself with a contact form.
The three cost models: platform, custom build, and hybrid
Before we get into numbers, it helps to understand the three fundamentally different ways you can approach AI automation. They have different cost structures, different ceilings, and suit different problems.
Approach | Upfront cost | Monthly cost | Best fit |
|---|---|---|---|
Platform (Make, Zapier) | £500–£3,000 setup | £50–£400 (subscriptions) | Simple, standard workflows |
Custom AI build | £8,000–£35,000+ | £500–£2,000+ (support/hosting) | Complex, bespoke operations |
Hybrid | £4,000–£15,000 | £200–£800 | Moderate complexity, budget-conscious |
Platform-based automation (Make, Zapier) — what you actually pay
Zapier and Make are the entry point for most small businesses. They connect your existing tools and automate handoffs between them without any real coding required.
The tool subscriptions themselves are relatively cheap. Make starts at around £9/month for low volume. Zapier's paid plans start around £19/month but the costs climb quickly once you start running more complex automations at volume. Most small businesses end up spending £50–£200/month on the tools once they've built a few workflows.
Setup is where the real cost sits. If you're doing it yourself, it's time not money. If you're paying someone to configure it, expect £500–£3,000 depending on complexity. That's a reasonable entry point for straightforward use cases.
The honest limitation: these platforms hit a ceiling fairly quickly. They're built around templates and connectors. When your operations have edge cases, custom logic, or data that doesn't fit neatly into a standard field, you're fighting the tool instead of using it. That's when people come to us after six months of frustration.
Custom AI builds: what drives the price up or down
A custom build means someone is writing actual code to fit your operations exactly. At AMPL, that's Claude Code. It's not a template shop. The system does specifically what your business needs, not what a template was designed for.
A focused first build, say an email-to-quote automation that reads incoming enquiries, extracts the relevant details, checks against a pricing matrix, and drafts a quote for review, typically runs £8,000–£15,000 for a well-scoped project. That range covers design, build, testing, integration, and handover.
What pushes the cost up:
Number of integrations (each external system adds complexity)
Data quality issues (messy or inconsistent source data costs time to handle)
Edge case volume (the more exceptions the system needs to handle, the more logic it needs)
Document generation with legal or compliance requirements
Multi-step approval workflows with human touchpoints
What keeps the cost down:
A clearly defined, well-scoped problem rather than scope creep mid-build
Clean source data
Starting with one workflow rather than trying to automate everything at once
A client who knows their own process well (you'd be surprised how often this isn't the case)
A transaction coordination system for a property business, managing documents, timeline reminders, and stakeholder communications across a deal, is a more complex build. You're looking at £20,000–£35,000 depending on how many systems it needs to connect with and how much exception handling is required.
Hybrid approaches: when to use each
Hybrid means using platform tools like Make or Zapier for the straightforward connective tissue, while custom code handles the parts that require actual intelligence or complex logic.
This can be a sensible middle ground. You're not paying for custom code where off-the-shelf works fine, but you're not trying to force complex logic through a template tool where it doesn't fit.
The downside is maintenance. You now have two different systems to manage, and when something breaks, diagnosing where the problem sits takes longer. For straightforward businesses, hybrid works well. For anything with genuinely complex operations, I'd usually recommend going custom from the start rather than hitting the hybrid ceiling six months later.
What small businesses typically spend at each stage
Here's how the investment usually sequences out for small businesses we work with.
Discovery and audit phase
Before any build, you need to know what's actually worth automating and what the ROI looks like. A proper audit maps your manual processes, quantifies the time cost, identifies the highest-value automation opportunities, and produces a prioritised roadmap.
At AMPL, the audit costs £1,500–£2,500 depending on the complexity of the business. It's refundable against the first build. Most businesses find it valuable as a standalone exercise because it tells them what they're actually spending on manual operations, which is often more than they thought.
Trying to skip this stage and jump straight to a build is usually a false economy. Without a clear map, scope creep kills budgets.
First build (single workflow)
The first build is typically the highest-value intervention. You're automating the biggest time sink, the process that eats the most hours or causes the most friction. If the audit has been done properly, you know exactly what that is.
For a focused single workflow, budget £8,000–£15,000. That's a meaningful investment for a small business. The question isn't whether that's expensive in absolute terms. It's what it costs relative to the time it saves. We'll get to the ROI calculation below.
Expanded automation (multiple workflows)
Once the first build is live and you've seen it work, most clients want to expand. The second and third workflows are usually faster and cheaper to build because the infrastructure is already in place and we know your systems.
Businesses that move to a broader automation programme across multiple workflows typically invest £30,000–£60,000 over 12–18 months. Run the opportunity cost calculation below on three or four workflows and that number usually looks very different.
How to calculate your automation ROI before you spend anything
This is where most pricing articles fall down. They tell you what things cost but not how to work out whether they're worth it. Here's how to do the calculation yourself.
The opportunity cost formula
Start with the process you're considering automating. Answer four questions:
How many hours per week does this process consume across your team?
What's the blended hourly cost of the people doing it (salary plus employer costs, divided by working hours)?
What's the annual cost of that time? (hours per week multiplied by hourly cost multiplied by 52)
What percentage could realistically be automated? For most workflows that's 60–80%, rarely 100%.
The formula: annual hours saved x hourly cost x automation percentage = annual value recovered
That's your savings figure. Compare it to the build cost to get your payback period.
Payback period calculator (worked example)
A removals company has an operations coordinator spending 15 hours a week processing job enquiries: checking availability, calculating quotes, sending confirmations, and updating the CRM. At a blended cost of £22/hour, that's £17,160 per year in staff time on a single process.
A custom email-to-quote automation handles the intake, checks the job calendar, applies the pricing logic, drafts the quote, and updates the CRM. The coordinator reviews and approves rather than building from scratch. That's roughly 70% of the time recovered.
£17,160 x 70% = £12,012 per year saved. Build cost: £12,000. Payback period: just under 12 months. Every year after that is pure recovery.
And that's the conservative version. It doesn't account for faster quote turnaround increasing conversion, or the coordinator's time being redeployed to higher-value work.
What makes a build more expensive (and how to avoid it)
A few things reliably push costs up. Knowing them in advance means you can come into the process better prepared.
Poorly documented processes. If your team can't clearly describe how a process works, including edge cases, the builder has to reverse-engineer it. That takes time, and time costs money. Before any scoping conversation, document the process you want to automate in detail: what triggers it, what happens at each step, what happens when something goes wrong.
Scope changes mid-build. Starting with "automate our quote process" and expanding to "and also sync with our accounting system and send WhatsApp confirmations" halfway through a build is expensive. Get the scope agreed and locked before work starts.
Legacy or unusual systems. If your business runs on a bespoke CRM from 2009 or a system with no API, integration is harder. Sometimes significantly harder. Flag this early. Any decent automation consultant will ask about your tech stack upfront, but volunteer the information if it's unusual.
Trying to automate everything at once. Automating five processes in one go increases complexity exponentially. Start with one. Get it working. Expand from there. It's faster, cheaper, and less risky.
Questions to ask before signing any automation contract
Whether you work with AMPL or anyone else, these are worth asking before you commit.
What's included in the quote and what isn't? Specifically: is testing included? What about post-launch bug fixes? Ongoing support? Be clear on what happens if something breaks in month three.
What does the handover look like? Will your team be able to understand and manage this system after delivery, or will you be dependent on the builder indefinitely? That dependency has a cost.
How is scope managed? What happens if requirements change mid-build? Is there a change control process? Unclear scope management is where projects go over budget.
Can you see examples of similar builds? Not case studies with vague outcomes. Actual examples of the type of system they'd be building for you, with some indication of complexity and timelines.
What's the support arrangement after go-live? AI systems need maintenance. Prompts drift, APIs change, edge cases emerge. A one-off build with no support is often a false economy.
FAQ
How much does AI automation cost for a small business?
For platform-based automation using Make or Zapier, expect £500–£3,000 to set up plus £50–£400/month in subscriptions. For a custom AI build on a single focused workflow, typical projects run £8,000–£15,000. More complex multi-workflow builds range from £20,000–£60,000 depending on scope. Starting with an audit (£1,500–£2,500) gives you a clear picture before committing to any build cost.
Is AI automation worth it for a small business?
It depends on how much manual work you're carrying. If your team spends 10 or more hours a week on a repeatable process, the maths usually works. Take the hours spent per week, multiply by hourly staff cost, multiply by 52. If that number is bigger than the build cost and the payback falls within 18–24 months, it's generally worth doing.
What's the difference between AI automation and standard automation?
Standard automation tools like Zapier move data between systems based on fixed rules. AI automation can handle unstructured inputs like emails and documents, and make judgements rather than just follow a rigid flowchart. That's what makes it useful for complex, real-world operations rather than simple data transfers between tidy systems.
How long does an AI automation build take?
A well-scoped single workflow typically takes four to eight weeks from kickoff to go-live. Larger multi-workflow builds can run three to six months. Timelines stretch when scope changes mid-project or when client-side dependencies like data access or stakeholder sign-off cause delays. The audit and scoping phase itself usually takes one to two weeks.
Can I build AI automation myself to save money?
For simple workflows on Make or Zapier, yes. The learning curve is manageable and there are good tutorials. For anything involving custom logic, multiple integrations, or unstructured data, DIY usually costs more in time than it saves in fees. The hidden cost is your own time: hours spent debugging a workflow are hours not spent running the business.
What should AI consulting fees cover?
At minimum: discovery and scoping, build and testing, integration with your existing systems, documentation, and post-launch support for a defined period. If a quote doesn't specify what happens after go-live, ask directly. Ongoing maintenance including prompt updates, API changes, and edge case handling is a real cost that low-priced quotes often leave out entirely.
If you want to run the numbers on your own operations before talking to anyone, the opportunity cost formula above will get you most of the way there. And if you'd like someone to do that mapping properly, that's exactly what the AMPL audit is for. Book a free introductory call at amplconsulting.ai.
