Key takeaways
- Most small firms still use no AI at all; simple if-this-then-that rules deliver much of the time saving with less risk.
- Start with the repeat jobs: service reminders, booking confirmations, payment chasers and review requests.
- Every automation needs an owner, a test and a way to switch it off.
- Keep service messages and marketing separate so reminders stay within PECR.
- Automate the routine and leave judgement calls, such as emergencies and complaints, to people.
Rules before robots
There is a lot of talk about AI in trade businesses, but most small firms have not started. Among UK businesses with fewer than ten employees, 65.7% used no AI at all in autumn 2026[1]. That is not a failing. Much of the admin that eats an office manager's week can be handled by plain rules: when a job is closed, send the invoice; when a service is due in four weeks, send a reminder; when an invoice is 14 days overdue, send a polite chaser.
Rules-based automation is predictable. It does the same thing every time, you can read exactly what it will do, and when it goes wrong you can see why. That makes it the right first step before anything more ambitious.
Automations worth setting up
| Trigger | Automatic action | Usually set up in |
|---|---|---|
| Service or safety check due in four weeks | Reminder email or text with a booking link | Job software |
| Booking made | Confirmation with date, window and what to prepare | Job or booking software |
| Day before a visit | Reminder with a reschedule link | Job software |
| Engineer marks on the way | Arrival message to the customer | Job software |
| Job closed | Invoice and service record sent | Job software linked to accounts |
| Invoice overdue | Polite chaser, then a firmer one | Accounts software |
| Website form submitted | Enquiry created and assigned, auto-reply sent | Website form plus job software or email rules |
| Job completed with no issues | Review request a day later | Job or review software |
| Quote sent | Follow-up task for the office after two working days | Job software or CRM |
| New starter or leaver | Checklist of accounts to create or remove | Shared task list |
Most job management systems include many of these already, switched off by default. Before buying another tool, go through your current software's settings and switch on what you are paying for.
Service reminders: the highest-value automation
Repeat work is built in to heating. Landlords must have each gas appliance and flue checked at intervals of not more than 12 months[2], and homeowners with a boiler warranty usually need an annual service. A reminder that goes out automatically, four weeks before the due date and again two weeks later if nobody has booked, turns that into predictable diary-filling work.
Keep reminders factual. Under the Privacy and Electronic Communications Regulations, direct-marketing emails and texts to individuals need prior consent, unless the soft opt-in applies to existing customers[3]. A message saying a customer's service is due is service information. A message offering a discount on a new boiler is marketing and needs the right basis. Do not slip offers into reminder messages.
Reminder template: “Hello Sam, your boiler at 12 Elm Road is due its annual service in March. Book a time that suits you here, or reply and we'll ring you. Thanks, Northside Heating.” (Example names.)
Payment chasers that stay polite
Late payment is a widespread problem. In 2025, late payment was a problem for 63% of SME employers that gave trade credit[4]. Automated chasers do not fix that on their own, but they stop overdue invoices being forgotten while the office is busy.
On the due date
A friendly note that payment is due today, with the invoice attached and how to pay.
Seven days overdue
A polite reminder, repeating that your bank details never change by email.
Fourteen days overdue
A firmer reminder and a task for the office to ring.
Thirty days overdue
Stop the automation. A person decides the next step.
For business customers, you can charge statutory interest of 8% plus the Bank of England base rate on late payments[5]. Whether you use that right is a relationship decision, which is exactly why the automation stops and hands over to a person.
Forms, emails and missed calls
- Route website enquiries to one shared inbox or straight into the job system, with an automatic reply saying when you will respond.
- Use email rules to file merchant invoices, remittances and agent instructions into folders, so they are not lost in the main inbox.
- Send voicemail recordings to the shared inbox as audio attachments, so whoever is free can call back.
- Create a daily call-back list from missed calls, cleared by a named person before the end of the day.
- Auto-forward landlord safety record requests to the person who sends records.
Building an automation safely
Write the rule in one sentence
When this happens, do that, for these customers, except these.
Name an owner
One person who checks it works and changes it when needed.
Test on yourself
Create a test customer with your own email and phone, and run it end to end.
Start small
Turn it on for one job type or one month of reminders before all customers.
Check the first results
Read every message sent in the first week.
Know the off switch
Write down how to pause it, and who can.
Engineer-side automations
Automation is not only for the office. Engineers lose time on admin at the end of every job, and small rules in the job app give some of it back.
- Pre-fill job sheets with the address, appliance details and last visit notes, so engineers only add what is new.
- Attach the right certificate template automatically by job type.
- Create a parts order or a return visit task when an engineer marks a job as waiting for parts.
- Send the next day's job list to each engineer the evening before, with access notes.
- Flag a job to the office when an engineer records a warning notice, so the follow-up call happens the same day.
Ask engineers which five minutes of each job annoy them most. The answer is usually a field they retype or a photo they send separately, and both can be fixed in an afternoon.
Messages customers actually want
More messages is not better. Customers want to know when you are coming, whether you are running late, what was done and what they owe. Anything beyond that risks being ignored or, worse, training them to ignore the message that matters. Review your automated messages once a year and remove any that nobody would miss. Keep each one short, say who it is from in the first words, and give a phone number for anyone who would rather talk.
Measuring the time saved
Before switching on a set of automations, time the manual version for two weeks: how long reminders, confirmations and chasers take each day. Repeat after a month. Also count the side effects: fewer no-shows after day-before reminders, more services booked from reminders, and fewer days that invoices sit overdue. Those figures tell you which automations to keep, and which to improve next.
Where automation goes wrong
- Reminders sent for boilers that were removed years ago, because the appliance record was never closed.
- Review requests sent to a customer who complained yesterday.
- Payment chasers sent for invoices already paid in cash on the day.
- Duplicate messages because two systems both send confirmations.
- Messages addressed to the landlord's tenant from two tenancies ago.
Almost every failure is a data problem rather than an automation problem. Clean records come first; the guide on setting a technology budget covers tidying your core tools.
Worked example: two hours back every week
Example: a fictional office manager spends Monday mornings sending service reminders from a spreadsheet, and Friday afternoons chasing invoices. She switches on the job software's built-in reminders at four and two weeks before the due date, and the accounts software's chasers at the due date, seven and 14 days. Both are tested on her own email first. After a month she reviews what went out, fixes three reminders for boilers that had been replaced and adds a rule to skip review requests on jobs marked as a complaint. Her Monday and Friday tasks shrink to checking exceptions.
When to look at AI
Once the rules are working, the remaining admin tends to be the work rules cannot handle: understanding what a caller needs, writing a job summary from an engineer's notes, or answering a question at ten at night. That is where AI tools may help. Read how trade firms are using AI for where they fit, and keep the same principles: an owner, a test and an off switch.
Frequently asked questions
What should a small trade firm automate first?
Service and safety check reminders, booking confirmations, day-before reminders and invoice chasers. They are frequent, predictable and usually built into job or accounts software already.
Are service reminders marketing under PECR?
A factual reminder that a customer's service or check is due is service information. Adding offers or promotions turns it into marketing, which needs consent or the soft opt-in.
Do I need AI to automate my admin?
No. Many quick wins are simple rules in software you already have. Consider AI for tasks rules cannot handle, once the basics run well.
How do I stop automations sending wrong messages?
Keep customer and appliance records clean, test each automation on yourself first, read the first week's messages and have an owner who can pause it.
References
Office for National Statistics (2026). Business insights and impact on the UK economy: wave 165 dataset (AI tables) (opens in a new tab) ↑
legislation.gov.uk (The National Archives) (1998). The Gas Safety (Installation and Use) Regulations 1998, regulation 36 (opens in a new tab) ↑
legislation.gov.uk (The National Archives) (2003). The Privacy and Electronic Communications (EC Directive) Regulations 2003, regulation 22 (opens in a new tab) ↑
Department for Business and Trade (GOV.UK) (2025). Longitudinal Small Business Survey 2025: SME employers (businesses with 1 to 249 employees) (opens in a new tab) ↑
GOV.UK (2026). Late commercial payments: charging interest (opens in a new tab) ↑