Most automation projects fail before they start — not because the technology doesn't work, but because people try to automate the wrong things first. They go after the flashy use case, the ambitious transformation, the "wouldn't it be cool if…" idea. Then they hit complexity, lose momentum, and quietly shelve the whole initiative. The smarter approach is almost embarrassingly simple: start where the pain is loudest, where the work is most repetitive, and where a mistake costs you real money or time. This guide will help you find that spot.
The Delegation Test: Four Questions to Ask About Any Task
Before you automate anything, run it through this filter. Think of a task you or your team does regularly — answering enquiries, logging data, chasing invoices, updating records. Ask yourself four questions:
1. Does this task follow a predictable pattern? Automation thrives on repetition. If the task looks roughly the same each time — same inputs, same outputs, same steps — it's a strong candidate. Responding to "What are your opening hours?" is predictable. Negotiating a contract is not.
2. How often does it happen? A task you do twice a year isn't worth automating yet. A task that happens twenty times a day absolutely is. Frequency multiplies your return. Even saving three minutes per occurrence adds up to an hour of recovered time daily across twenty repetitions.
3. What's the cost of a mistake? Some errors are embarrassing but recoverable. Others cost you a client, a compliance fine, or a missed deadline. Tasks with high error costs — and where errors currently happen because humans are tired or rushed — are prime automation targets.
4. How much skilled human judgment does it really need? Be honest here. We often assume our work requires more judgment than it actually does. Booking a meeting, sending a follow-up email, generating a weekly report — these feel like they need a human touch, but they rarely do.
Score any task well on three or four of these, and it should be near the top of your automation list.
Where to Look First: The Hidden Time Sinks in Your Business
Most businesses are sitting on the same set of high-frequency, low-value tasks that eat into productive hours without anyone fully noticing. They get absorbed into the day and normalised. Here are the most common offenders — and rough estimates of what they're actually costing you.
Appointment and booking admin. A busy physiotherapy clinic found that their front desk was spending approximately 90 minutes a day managing appointment confirmations, cancellations, and rebooking requests via phone and email. After implementing an AI-powered scheduling assistant integrated with their booking system, that dropped to under 15 minutes of oversight per day. That's roughly 6 hours recovered per week — time that was redirected to patient onboarding and billing queries.
Data entry and transfer between tools. If you're copying information from one system to another — a form submission into a CRM, an invoice into a spreadsheet, a booking into a calendar — you're paying a person to do something a machine does better, faster, and without the occasional typo. Studies from McKinsey consistently show that data entry and collection tasks consume between 15–20% of a knowledge worker's week. For a team of five, that could be 10+ hours of lost productivity every week.
Follow-up communications. Chasing unpaid invoices, reminding clients about upcoming appointments, nudging prospects who went quiet after a proposal — these tasks are simple, repetitive, and chronically underdone because people feel awkward about them or simply forget. Automating a three-touch follow-up sequence for overdue invoices alone can recover thousands of pounds per year in late payments for a business turning over £500k.
Reporting and status updates. Pulling together a weekly report from three different tools, formatting it, and sending it to a manager or client is exactly the kind of task that sounds quick but isn't. It involves context-switching, formatting decisions, and copy-pasting — often taking 30–45 minutes that could be entirely automated.
Building Your Automation Priority List
Once you've identified your candidates, you need to rank them. The most useful framework here is a simple two-axis grid: impact against implementation effort.
High impact, low effort tasks should be automated first. These are your quick wins — the ones that will free up time within weeks and build confidence in the approach. High impact, high effort tasks come second; they're worth pursuing but need proper planning. Low impact tasks, regardless of effort, should wait or be reconsidered entirely.
To estimate impact, multiply time saved per occurrence by frequency, then assign a rough hourly cost. If your office manager earns £30,000 a year, their time costs you around £15 per hour. A task that takes 20 minutes and happens 15 times a week costs you £75 a week, or nearly £4,000 a year — before you factor in the opportunity cost of what they could be doing instead.
To estimate effort, think about how many tools are involved, whether the data is clean and consistent, and whether you need to make any decisions mid-process. A single-tool, single-step task (auto-reply to a specific email type) is far easier than a multi-tool workflow (capture lead → enrich data → assign to sales rep → log in CRM → trigger onboarding sequence).
Start simple. Prove the model. Then expand.
The One Mistake That Derails Most Automation Projects
Here's the trap: automating a broken process. If a workflow is messy, inconsistently followed, or reliant on informal knowledge that lives in someone's head, automating it will just make the mess faster and harder to unpick.
Before you automate anything, document it. Write out every step. Note where the exceptions are, where things currently go wrong, and what a "good" outcome looks like. This documentation step often reveals that the process itself needs simplifying first. A task that looked like a ten-step automation turns out to need only four steps if you tighten the process beforehand.
This is also where involving your team matters. The people doing these tasks daily know exactly where the friction is. Ask them: "What's the thing you do every week that you wish you didn't have to?" The answers will build your automation backlog faster than any consultant's audit.
Conclusion
Deciding what to automate first isn't a technology decision — it's a business prioritisation decision. The best starting point is almost never the most ambitious one. It's the task that happens most often, follows a clear pattern, and quietly drains your team's time and attention every single day. Find that task, document it cleanly, and automate it well. One successful automation builds the confidence, the process knowledge, and the appetite for the next one. That's how you build a business where AI is genuinely working for you — rather than a half-finished project sitting in a folder somewhere, waiting to be revisited.