There is no meaningful single price for business automation. A workflow that sends a form into a CRM is not comparable with a document-processing system that connects five platforms, handles sensitive data, includes a review dashboard and has dozens of exception paths.
That is why “how much does automation cost?” should be followed by a second question: what business process are we changing, and what is that improvement worth?
Illustrative UK automation project ranges
| Type of project | Illustrative range | Typical scope |
|---|---|---|
| Simple workflow/integration | £1,000–£3,000 | Form-to-CRM, notifications, straightforward data transfer |
| Email or document workflow | £2,000–£7,500 | Classification/extraction, validation, routing and system updates |
| Multi-step operational workflow | £5,000–£15,000 | Several systems, rules, approvals, exception handling and reporting |
| Bespoke automation system | £10,000–£30,000+ | Custom logic, data pipeline, interfaces and multiple workflows |
| Larger operational platform | £25,000–£50,000+ | Broader custom environment with significant development, testing and support |
There is deliberate overlap. Complexity does not follow neat labels. A “small” project connecting an old desktop application with no supported API can require more engineering than a larger workflow built entirely on modern cloud software.
What are you actually paying for?
A professional automation project contains more than development hours. Depending on scope, work may include process discovery, solution architecture, integration research, data mapping, security design, development, test cases, exception handling, deployment, documentation, training, monitoring and post-launch support.
Cheap quotes sometimes look cheap because several of those activities are simply absent.
Cost driver 1: the number and quality of integrations
Every system connection introduces questions: how is authentication handled, what data can be read or written, are there rate limits, what fields map to each other, how are errors returned and what happens if the service is unavailable?
Modern SaaS products with documented APIs are usually easier to integrate. Legacy systems, restricted APIs, desktop software or vendor-specific export/import processes can increase complexity substantially.
If a workflow touches Microsoft 365, a CRM, accounting platform, document store and bespoke internal database, each connection needs to be designed and tested.
Cost driver 2: workflow branches and exceptions
A straight line is cheaper than a decision tree. The words that increase scope are often “unless”, “except”, “only when”, “for these customers” and “if the document is missing X”.
Exceptions are not bad. They are part of real business operations. But they need to be surfaced during discovery. A workflow with 25 exception categories requires more logic, test data and review than one with two.
Cost driver 3: whether AI is genuinely required
Traditional automation follows explicit rules. AI is useful when input is variable or requires interpretation: free-text emails, different document layouts, summarisation or classification.
If fixed rules can solve the problem, they are often cheaper and more predictable. Using AI where it is not needed can add model usage costs, evaluation work and additional monitoring.
Read AI automation vs traditional automation for a detailed explanation of when each approach makes sense.
Cost driver 4: data quality and readiness
Automation depends on data. If a customer’s unique ID is reliable and consistent, system matching is straightforward. If the same customer appears under different names in three spreadsheets and the CRM contains duplicates, the project may need data-cleaning and matching logic before automation can work safely.
Data readiness is one reason process discovery matters. An automation quote based only on a high-level idea can miss substantial preparation work.
Cost driver 5: user interfaces
Some automations run entirely in the background. Others need a dashboard, review queue, approval screen, audit history or administration interface. Custom interfaces introduce design, development, permissions and browser/device testing.
A good question is whether the business genuinely needs a new interface or whether existing tools can remain the place where staff work.
Cost driver 6: security and privacy
A generic lead-routing workflow is different from a system processing payroll, legal documents, financial information or personal data. Higher sensitivity can require stronger access controls, audit logging, data minimisation, retention rules, encryption and supplier due diligence.
The NCSC’s secure AI guidance recommends considering security during design, development, deployment and ongoing operation. It specifically highlights controls around external APIs, sensitive data, supply chains, model actions and monitoring. Those controls are part of the system, not optional polish.
Cost driver 7: testing and assurance
A reliable workflow needs more than “it worked in the demo”. Testing should cover normal cases, missing information, malformed data, unavailable systems, duplicate records, permissions failures and other edge cases.
AI components need additional evaluation because the same kind of input can be expressed in many ways. For higher-impact use cases, human review and formal assurance may be necessary.
Cost driver 8: support and maintenance
After launch, systems and APIs change. Staff discover new exceptions. Credentials expire. Vendors alter fields. AI providers release new model versions. A production workflow therefore has an operating lifecycle.
Before signing a proposal, clarify:
- what warranty period is included;
- how support is charged;
- response times for critical and non-critical issues;
- whether monitoring is included;
- how enhancements are priced;
- who owns documentation and credentials;
- what happens if a third-party service changes.
Software subscriptions vs implementation cost
Automation may also use paid platforms, cloud services, AI model APIs or connector tools. These recurring charges are separate from implementation and can vary with usage.
A low-code workflow might have a modest monthly platform cost but significant one-off discovery and development work. An AI-heavy process might have lower build complexity but more variable usage cost. Good proposals show both.
How to calculate ROI
Start with the current cost of the process. Staff time is the easiest component to estimate:
But do not stop there. Depending on the process, value may also come from:
- fewer errors and less rework;
- faster response to leads;
- reduced customer wait times;
- higher throughput without additional headcount;
- less overtime;
- fewer missed deadlines;
- better auditability;
- management time released from supervision and reporting.
A worked example
Imagine a professional-services firm where three employees collectively spend 32 hours per week triaging documents, updating records and chasing missing information. At an average loaded employment cost of £24 per hour, that represents roughly £36,864 of annual staff input over 48 weeks.
If a system reduced manual handling by 55%, the theoretical capacity released would be about £20,000 a year. But that is not automatically £20,000 of cash savings. The business might use the capacity to serve more clients, reduce backlog, improve service or avoid a future hire.
If implementation cost £10,000 and ongoing costs were £2,000 a year, the project could still have a strong commercial case — but only if the organisation has a realistic plan for the capacity it releases.
Payback period
A simple payback calculation is:
Do not force every project into a six-month payback. Higher-risk, strategic or capacity-building systems may justify a longer horizon. The point is to make the assumption explicit.
Should you buy an off-the-shelf tool instead?
Yes, when an established product already solves the workflow well. Bespoke automation is not automatically superior.
Custom work is most compelling when value comes from:
- connecting several tools you already use;
- implementing organisation-specific rules;
- handling a workflow that standard software only partly covers;
- removing repeated hand-offs between systems;
- building an internal capability that is strategically differentiating.
For example, if a specialist property-management platform already handles maintenance perfectly, there may be no reason to rebuild it. The bespoke opportunity may be connecting that system to email, finance, contractor onboarding or reporting.
Questions to ask every automation supplier
- What problem are you actually solving?
- Which systems and workflow steps are included?
- Which items are explicitly excluded?
- How are errors and exceptions handled?
- Where is AI used, and why?
- What happens when AI confidence is low?
- How will the solution be tested?
- How is data secured?
- What documentation will we receive?
- Who owns the code, workflow and IP?
- What support is included after launch?
- What recurring third-party costs should we expect?
How to get a more accurate quote
The quality of a quote depends on the quality of the brief. Before speaking to a supplier, document:
- what starts the process;
- what staff do today;
- how many cases occur;
- which systems are involved;
- important exceptions;
- sensitive data or compliance requirements;
- the result that defines success.
Our Automation Assessment is designed to collect this information without requiring a technical specification.
Why the cheapest build can become the most expensive
A brittle automation that fails silently, cannot handle exceptions and has no documentation can cost more to support than it saved. Price should therefore be evaluated alongside architecture, maintainability and business risk.
UK SME adoption research repeatedly emphasises that successful technology adoption is a journey rather than a single purchase. Implementation, employee confidence and ongoing support matter alongside the initial product or build.
Sources and further reading
- Department for Business and Trade — Understanding technology adoption among UK SMEs
- DSIT — AI Adoption Research (2026)
- NCSC — Guidelines for secure AI system development
- SME Digital Adoption Taskforce — Final report
Could this apply to your business?
Tell us how the workflow operates today and we'll assess where automation could make a practical difference.
Start your free assessment