Quick Answer
A 30-person electrical firm was spending an estimated £40,000 per year on avoidable admin: duplicate data entry, manual invoicing, paper timesheets, and reactive scheduling. This playbook walks through the exact audit process used to find that waste, the tools involved (ServiceM8, Xero, n8n, Make.com), and the seven automation fixes applied. You can use the same framework on your own business in under a week.
Table of Contents
- The Firm: Background and Starting Point
- The Audit Method: How We Mapped Every Admin Task
- What We Found: Six Categories of Waste
- The Tech Stack: ServiceM8, Xero, n8n, and Make.com
- The Seven Fixes Applied
- Results After 90 Days
- Further Viewing
- What Other Trades Are Saying
- Running the Audit on Your Own Business
- Verdict
- Frequently Asked Questions
The Firm: Background and Starting Point
Spark & Sons Electrical (name changed) is a 30-person firm based in the East Midlands. They run a mixed domestic and commercial operation, including EICRs, rewires, consumer unit upgrades, commercial fit-outs, and a growing solar installation division. Turnover sits around £3.2 million annually.
The MD came to us with a specific complaint: "We're growing but our margins are shrinking." Revenue was up 18% year-on-year, but net profit had barely moved. His gut feeling was that admin was the problem. He was right. But the scale of it surprised even him.
They were already using ServiceM8 for job management, Xero for accounting, and had a part-time office manager handling everything that fell between the two. There was no automation in place. Everything between job completion and invoice payment was manual.
The Audit Method: How We Mapped Every Admin Task
The audit took four days. Two days of interviews and observation, one day of data analysis, one day of prioritisation. Here is the exact process.
Day 1: Shadow the Office Manager
Spend a full day sitting next to the person who handles the most admin. Do not ask them to describe their job. Watch them do it. Every time they open a new tab, switch between apps, type something they have already typed, or pick up the phone to chase something, write it down. Note the time it takes.
At Spark & Sons, the office manager's day broke down like this:
- 08:00–09:30: Transferring job completion notes from ServiceM8 into a spreadsheet for invoicing
- 09:30–10:45: Creating invoices manually in Xero from the spreadsheet
- 11:00–12:00: Chasing outstanding purchase orders by email
- 13:00–14:30: Reconciling engineer timesheets against job records
- 14:30–15:30: Responding to customer queries about job status
- 15:30–17:00: Filing, filing, filing
That is a full day of work that produces nothing new; every task is processing information that already existed somewhere else in the business.
Day 2: Interview Every Engineer
Ask each engineer three questions. How long do you spend on paperwork at the end of each job? What do you have to do twice because the office needs it differently from how you recorded it in ServiceM8? What do you forget to do and have to be chased about?
At Spark & Sons, the average engineer spent 11 hours per week on admin, mostly post-job documentation, timesheet submission, and materials recording. With 22 engineers on the tools, that is 242 engineer-hours per week. At an average fully-loaded cost of £30/hour for on-tools time, that is £7,260 per week in admin cost from the engineers alone.
Day 3: Quantify the Waste
Pull three months of data from ServiceM8 and Xero. Count: how many jobs were completed versus how many invoices were sent within 24 hours. Count how many purchase orders had to be chased more than once. Count how many jobs were closed in ServiceM8 but never invoiced in Xero.
At Spark & Sons, 34% of completed jobs were not invoiced within 48 hours of completion. The average delay was 8.3 days. On a monthly invoicing run of roughly £250,000, that delay represented approximately £3,200 per month in working capital held up, totalling £38,400 per year.
Day 4: Score and Prioritise
Score every identified waste item on two axes: how much time does it consume per week, and how difficult is it to automate with tools you already have? Anything scoring high on time and low on difficulty goes to the top of the list. Anything scoring low on time and high on difficulty gets parked.
What We Found: Six Categories of Waste
After four days of audit work, every admin task at Spark & Sons fell into one of six waste categories. This taxonomy is useful because it applies to most trades businesses regardless of size.
1. Duplicate data entry: The same information typed into two or more systems. At Spark & Sons this was mostly job details going from ServiceM8 into Xero. Every completed job required a manual handoff. This alone accounted for 6.5 hours per week of office manager time.
2. Manual triggered tasks: Tasks that should start automatically when something else happens. Invoice creation when a job is marked complete. A customer email when an engineer is en route. A purchase order reminder when a quote expires. None of these had triggers at Spark & Sons. All were done on a schedule or when someone remembered.
3. Missing follow-up: Quotes, invoices, and service reminders that fell through the cracks. The audit found 47 quotes sent in the previous three months that had never been followed up. Estimated lost revenue: £28,000.
4. Paper-to-digital transcription: Engineers completing paper EICR forms on site, then re-entering data into ServiceM8 in the van or back at the office. Every paper form that gets typed up later is a transcription risk and a time cost.
5. Communication lag: Customers not being notified when engineers were running late, when jobs were complete, or when invoices were due. This generated an average of 14 inbound calls per day to the office, each taking 4–8 minutes to handle.
6. Reconciliation work: Monthly effort to match timesheets against job records, identify unbilled materials, and reconcile Xero against ServiceM8. At Spark & Sons this took the office manager two full days every month.
The Tech Stack: ServiceM8, Xero, n8n, and Make.com
Spark & Sons already owned the tools needed to fix every one of these problems. They did not need new software. They needed the software they already had to talk to each other.
ServiceM8
Xero
n8n
Make.comServiceM8 has a well-documented REST API and webhook system. Xero has one of the richest accounting APIs available. n8n is an open-source workflow automation tool that runs self-hosted (the firm already had a small VPS for their email server). Make.com handles the lighter scenarios that needed a visual interface rather than code.
The Seven Fixes Applied
Each fix corresponds directly to a waste category from the audit. They are ordered by payback speed, fastest first.
Automated job-to-invoice pipeline (ServiceM8 → Xero)
When a job is marked as "Work Order Complete" in ServiceM8, an n8n webhook fires. The workflow pulls the job details via the ServiceM8 API, pulling client ID, job items, materials used and labour time, then creates a draft invoice in Xero with the correct line items populated. The office manager reviews and approves with one click. Time saved: 6.5 hours per week. Payback: immediate on the first invoice run.
Automated invoice chasing sequence
A Make.com scenario polls Xero daily for invoices overdue by 7, 14, and 30 days. At each threshold, it sends a personalised email to the client (using their name and invoice reference pulled from Xero). At 30 days it also creates a task in ServiceM8 for the office manager to make a phone call. Average debtor days dropped from 42 to 28 within 60 days of deploying this fix.
En-route and completion customer notifications
ServiceM8 already captures when an engineer checks in and when they mark a job complete. A Make.com scenario listens for these status changes via webhook and sends SMS notifications to the client. Engineers no longer call ahead. The office no longer handles "is he still coming?" calls. Inbound call volume dropped by 60% in the first month.
Automated quote follow-up sequence
Quotes in ServiceM8 that remain in "Sent" status after 5 days trigger an n8n workflow. The workflow sends a polite follow-up email and sets a task for a call at day 10. No quotes fall through the cracks. Of the first 30 quotes that went through the new sequence, 9 converted that would previously have been lost. Estimated additional revenue in the first 90 days: £21,000.
Digital timesheet capture in ServiceM8
Engineers were shown how to log time directly against jobs in the ServiceM8 mobile app rather than completing paper timesheets. An n8n workflow runs every Sunday evening, pulling all job time logs for the week and generating a formatted timesheet report in Google Sheets. The office manager's monthly reconciliation dropped from two days to four hours.
Purchase order expiry reminders
A Make.com scenario checks outstanding purchase orders daily. Any PO approaching expiry (within 5 days) triggers an email to the supplier and a task in ServiceM8 for the office manager. Chased POs dropped from an average of 3.2 per week to 0.4 per week within the first month.
Xero-to-ServiceM8 payment status sync
When an invoice is marked as paid in Xero, an n8n workflow updates the corresponding job in ServiceM8 to "Invoiced & Paid" status. This sounds small but eliminated an entire category of confusion: engineers turning up to sites where the client thought the job was still open because ServiceM8 showed it as unpaid.
Results After 90 Days
Here is what actually changed in the 90 days after the seven fixes were deployed.
The total annualised saving across all seven fixes came to £41,200. That includes the reduction in office manager overtime, the recovered cash flow value from faster invoicing, the additional quote revenue, and the reduction in engineer admin time.
| Metric | Before Audit | After 90 Days | Change |
|---|---|---|---|
| Average invoice delay (post-job completion) | 8.3 days | 1.2 days | -85% |
| Average debtor days | 42 days | 28 days | -33% |
| Office inbound call volume (daily) | 14 calls | 6 calls | -57% |
| Monthly reconciliation time | 2 days | 4 hours | -75% |
| Quote conversion rate (followed-up quotes) | Unmeasured | 30% | New metric |
| POs chased per week (average) | 3.2 | 0.4 | -88% |
Further Viewing
Business Process Mapping 101
A solid introduction to mapping business processes before automating them. Start here if the audit framework above felt abstract.
How to Automate Any Business With AI in 3 Steps
The high-level framework for identifying automation opportunities across any business, and maps well onto the waste categories in this article.
Contractor Hacks For ChatGPT That 10x Your Productivity
Practical AI productivity tips for contractors and tradespeople, covering admin shortcuts and client communication templates.
How to Scale Your Electrical Business Past £1 Million
Covers the operational systems needed to grow an electrical firm, with admin efficiency a core theme throughout.
How To Build An AI Receptionist For A Trades Business
Goes beyond the fixes in this article to show how AI can handle inbound enquiries, reducing call volume further.
Master 80% of n8n in 36 Minutes
The fastest way to get comfortable with n8n before building the workflows described in this case study.
What Other Trades Are Saying
Running the Audit on Your Own Business
You do not need to hire a consultant to do what we did at Spark & Sons. The framework is straightforward. Here is a condensed version you can run yourself in one week.
Block one day for shadowing
Sit with whoever does the most admin. Do not ask them to describe what they do. Watch them do it for a full day. Log every task, every app switch, every time they type something that already exists somewhere else.
Survey your engineers
Ask three questions: How long does post-job admin take you? What do you have to do twice? What do you get chased about? Collate the answers; the patterns are usually obvious.
Pull three months of data
From your job management and accounting software, extract: time from job completion to invoice sent, average debtor days, quote conversion rate, number of chased POs. These are your baseline metrics.
Classify every task into the six waste categories
Use the six categories from this article: duplicate data entry, manual triggered tasks, missing follow-up, paper-to-digital transcription, communication lag, reconciliation work. Every admin task fits somewhere.
Score and prioritise
Score each waste item: hours per week consumed, difficulty to automate with your existing tools. Pick the top three. Build those first. Do not try to fix everything at once.
Verdict
The £40K saving at Spark & Sons was not the result of expensive new software or a complex digital transformation programme. It came from doing one thing: systematically mapping where time and money were leaking out of the business, and connecting the software they already owned.
Most trades businesses at the 20–50 person scale have the same structural problem: ServiceM8 or Jobber on one side, Xero or QuickBooks on the other, and a human manually carrying information between them. That human costs £30,000–£45,000 per year. The automation costs £20–£50 per month in hosting and tooling.
Run the audit. Map the waste. Start with the invoice pipeline. The numbers take care of themselves.
Frequently Asked Questions
The Make.com scenarios (invoice chasing, customer notifications, PO reminders) require no coding at all. Make.com has a visual drag-and-drop interface. The n8n workflows (job-to-invoice pipeline, timesheet sync) need a basic understanding of JSON and API concepts, but n8n's visual builder means most of the work is configuration rather than coding. If you can set up a spreadsheet formula, you can build these workflows.
The main costs: n8n self-hosted on a VPS runs from about £5–£10 per month. Make.com's Core plan (sufficient for most small trades businesses) is £9 per month. ServiceM8 and Xero you are presumably already paying for. Total new tooling cost: £14–£19 per month, or roughly £200 per year. Against a saving of £40,000, the payback period is measured in days, not months.
The same audit framework applies regardless of which job management platform you use. Jobber has a GraphQL API and webhook support, so every automation described in this article can be replicated with Jobber as the trigger source. The specific API calls differ but the logic is identical. We have a separate guide covering the Jobber-to-Xero pipeline specifically.
Four days for a business the size of Spark & Sons (30 people). For a smaller firm, say 5–10 engineers, you can compress it into two days. The shadow day is fixed regardless of business size. The engineer interviews scale linearly. The data analysis is usually half a day once you know what you are looking for.
Automating the wrong things first. The instinct is to start with what feels impressive or technically interesting rather than what generates the fastest payback. The job-to-invoice pipeline is not glamorous. It does not involve AI or machine learning. But it pays back within days and creates immediate cashflow improvement. Start there. Add the sophisticated stuff later once the basics are running reliably.
