How to Plan Your First Automation Project

Start Small or Go Home

The single biggest mistake businesses make with their first automation project is trying to do too much. We call it the "boil the ocean" approach: let us automate everything at once, transform the whole business, go big or go home.

It fails almost every time. And it fails spectacularly, leaving businesses with overspent budgets, frustrated teams, and a deep scepticism about automation that takes years to shake.

The approach that actually works is the opposite. Start with one process. A single, contained, well-understood process that your team does frequently. Automate it properly. Measure the results. Celebrate the win. Then use that success to fund and justify the next one.

This is not conservative thinking — it is strategic thinking. Your first automation project is not just about saving time on one task. It is about building organisational capability. Your team learns how automation works. You learn what good implementation looks like. You build the internal momentum and confidence that makes every subsequent project easier and faster.

The businesses that are furthest along their automation journey today are not the ones that started with the biggest projects. They are the ones that started with small wins two years ago and compounded them. If you are still in the planning phase, that compounding has not started yet — and that is time you cannot get back.

Picking the Right First Process

Your first automation project needs to tick several boxes simultaneously. It needs to be impactful enough to prove the value of automation, but simple enough to succeed without too much complexity. Here is the criteria we use with clients:

Common first projects that tick all these boxes: email triage and routing, invoice processing, appointment confirmation and reminders, data entry from forms or documents, weekly report generation, and customer enquiry acknowledgements.

Avoid these for your first project: anything involving financial transactions with customers, processes that change frequently, tasks where exceptions are the norm rather than the rule, or anything where an error could have legal or safety implications.

Mapping the Current Process

Before you automate anything, you need to understand exactly how it works today. Not how you think it works. Not how the manual says it should work. How it actually works, in practice, including all the workarounds, shortcuts, and unofficial steps your team has added over the years.

A proper process mapping exercise involves sitting with the people who actually do the work and documenting every step. This typically reveals surprises: steps you did not know existed, workarounds for problems nobody reported, and variations between team members doing "the same" task differently.

Here is a simple approach you can use:

1. Ask the person who does the task to talk you through it while they do it. Take notes.
2. Document every step, every decision point, and every exception they mention.
3. Ask "what happens when things go wrong?" for each step. The exception paths are where automation projects often stumble.
4. Repeat with a different team member doing the same task. Note the differences.
5. Write up the process as a clear, step-by-step flow, including decision points and exception paths.

This documentation is valuable regardless of what happens next. Even if you decide not to automate right away, you now have a clear map of how your business actually operates. Many businesses discover significant inefficiencies during this step alone.

Defining Success

Before you write a single line of code or configure a single tool, define what success looks like. This is critical, because without clear metrics, you will never know if the project was worth it, and you will not have the evidence you need to justify the next one.

Measure the current state first. How long does the process take? How many times per week does it happen? How many errors occur? What does it cost in staff time? These are your "before" numbers. Write them down. You will compare against them later.

Then set specific targets for the automation:

• Time reduction: "Reduce time spent on invoice processing from 10 hours per week to 2 hours per week"
• Error reduction: "Reduce data entry errors from 5% to under 1%"
• Speed improvement: "Reduce customer enquiry response time from 4 hours to 15 minutes"
• Cost savings: "Save £15,000 per year in staff time on this task"

Be specific. Be realistic. And be honest about what "good enough" looks like. Automation does not need to be perfect to be valuable. If it handles 80% of cases automatically and your team handles the remaining 20%, that is still a massive win. For detailed guidance on calculating returns, see our ROI of automation guide.

Choosing Your Approach

You have three broad options for implementing your first automation. The right choice depends on the complexity of the process, your budget, and your team's technical confidence.

DIY with off-the-shelf tools. Platforms like Zapier, Make (formerly Integromat), and Power Automate let you build automations by connecting tools together with a visual interface. No coding required. Best for simple, linear processes that connect two or three tools you already use. Budget: £50–£300/month for the tools, plus your team's time.

Work with a consultant. An experienced AI consultancy handles the design, build, testing, and handover. They bring methodology, experience, and objectivity. Best for medium-complexity processes where getting it right first time matters. Budget: £2,000–£15,000 depending on complexity, plus monthly running costs.

Custom development. A bespoke solution built specifically for your business. Best for complex, unique processes that off-the-shelf tools cannot handle. Budget: £15,000+ for development, plus ongoing maintenance. This is rarely the right choice for a first project.

Running the Project

A well-run first automation project follows a predictable timeline. Here is what to expect:

Weeks 1–2: Discovery and documentation. Map the process, define success metrics, and agree on the scope. This is the foundation. Do not rush it.

Weeks 3–5: Build and configure. The automation gets built. This might be you configuring tools, or your consultant building the solution. Either way, expect iterative development — build a bit, test a bit, refine a bit.

Weeks 5–7: Testing. Run the automation alongside your existing manual process. Compare results. Fix issues. This parallel running period is essential — never switch off the manual process until the automation has proven itself.

Weeks 7–8: Go-live and handover. Switch to the automated process. Train any team members who need to interact with it. Document everything. Set up monitoring so you know when something needs attention.

Throughout the project, keep these principles in mind:

• Involve your team from day one. The people who do the work know the exceptions, the edge cases, and the unwritten rules. Their input prevents costly rework.
• Communicate constantly. No surprises. Keep everyone informed about what is changing and why.
• Accept imperfection. Your first automation will not handle every edge case. That is fine. Handle the 80% automatically and improve over time.
• Document everything. When you scale up, you will thank yourself for having clear records of decisions, configurations, and lessons learned.

Measuring Results

Once your automation has been running for at least four weeks, measure the results against the "before" numbers you recorded. Be thorough and honest.

Compare time spent before and after. Calculate the cost saving based on the hourly rate of the people whose time has been freed up. Count errors before and after. Measure speed improvements. Ask your team how it feels — qualitative feedback matters alongside the numbers.

Present the results clearly. A simple before-and-after summary is powerful: "Invoice processing took 10 hours per week. It now takes 2 hours. That saves 8 hours per week, or approximately £12,000 per year, with a one-time setup cost of £3,500." Numbers like that make the next project an easy sell.

Be honest about what did not work as well as expected. If the automation handles 75% of cases instead of the 90% you hoped for, say so — and explain what would be needed to close the gap. This honesty builds trust and makes your recommendations more credible.

Scaling Up

Your first successful automation project is not the finish line — it is the starting gun. You now have evidence that automation works in your business, a team that understands the process, and the credibility to invest further.

The next step is to identify your second project. Use the same criteria as your first: high frequency, rule-based, measurable, visible, low risk. But now you have more options, because your team has experience and your organisation has appetite.

Many businesses find that their second and third projects are faster and cheaper than the first, because the foundational learning has already happened. The team knows what to expect. The infrastructure (tool subscriptions, integration patterns, documentation standards) is already in place.

Take our AI Readiness Checklist again after your first project. You will likely score higher than before, because the process of doing your first automation improves your readiness for everything that follows.

Here are the key takeaways:

• Start with one small, contained process — do not try to transform everything at once
• Pick a process that is frequent, rule-based, measurable, visible, and low risk
• Map the actual process before you automate — not what you think happens, but what really happens
• Define clear success metrics before you start building
• Choose DIY tools for simple processes, a consultant for medium complexity
• Allow 6–8 weeks from discovery to go-live for a first project
• Measure results honestly and use them to justify the next project
• Every month you delay is compound time lost to manual work