“Automation will save you 20 hours a week” is the sort of claim that makes anyone who has run a crew switch off. So let us do this the other way around: work out where your admin hours actually go, decide which of them a machine can plausibly take, and see what is left.
The method below takes about twenty minutes with a notepad. It is the same one we use before quoting an automation project, and it is deliberately conservative — if the numbers do not stack up, we would rather find that out before anyone spends money.
Step 1: Count the tasks, not the hours
Nobody can accurately tell you how many hours a week they spend on admin. Everyone can tell you how many quotes they send. So count events, then attach a time to each.
For a typical small trades business, the recurring admin looks like this:
| Task | Typical frequency | Minutes each |
|---|---|---|
| Taking an enquiry and writing down the details | Every enquiry | 5–10 |
| Turning a site visit into a written quote | Every quote | 20–45 |
| Following up on quotes that went quiet | Every quote, 1–3 times | 5 |
| Scheduling a job and telling the crew | Every job | 10–15 |
| Rescheduling when something moves | Often | 10–20 |
| Chasing job notes and photos off the crew | Every job | 5–15 |
| Raising the invoice from the job details | Every job | 10–20 |
| Chasing the invoice | Every overdue invoice | 5–10 |
Fill in your own frequencies. A two-crew business doing forty jobs and sixty quotes a month is usually somewhere between 40 and 70 hours of admin a month once you total it — often spread across an owner’s evenings rather than a bookkeeper’s timesheet, which is exactly why it feels invisible.
Step 2: Sort each task into one of three buckets
Not all admin is automatable, and pretending otherwise is how automation projects fail.
Bucket A — mechanical. The same inputs produce the same outputs every time, with no judgement. Copying job details into an invoice. Sending the “just checking you got our quote” follow-up on day three. Turning a voicemail into a text summary in the team chat. Assembling a job pack from the enquiry. These are near-completely automatable, and realistically you recover most of the time they consume.
Bucket B — judgement with structure. A person decides, but a machine can do everything either side of the decision. Pricing a quote is the classic one: the measure-up and the margin call are yours, but pulling the customer details, laying out the line items from your rate card, generating the PDF, sending it and logging it are not. Here you typically recover half to two-thirds of the time — the quote that took 40 minutes takes 15.
Bucket C — human. Standing in someone’s kitchen working out what they actually want. Telling a customer the job will run over. Deciding which crew handles a difficult client. Automate none of it, and be suspicious of anyone who offers to.
Step 3: Do the arithmetic
Multiply frequency × minutes × recovery rate for each row, and total it. For the two-crew business above, the sum usually lands somewhere in the range of 8 to 15 hours a month recovered from a first automation project touching quoting and job scheduling, rising as more of the workflow gets connected.
That is a smaller number than the marketing headline and a much more useful one, because it is a number you can check against reality in month one.
Then convert it. Reclaimed admin hours are worth either:
- the hourly cost of whoever is doing them — if it is an admin salary, this is a real cash saving; or
- the value of what the owner does instead — if the owner reclaims six evenings a month and spends two of them quoting work that would otherwise have waited a week, the value is the jobs won, not the hours saved.
The second is almost always the bigger number in a trades business, and it is the one that decides whether a project is worth doing.
Where the biggest wins usually are
Across the trades businesses we talk to, the same three bottlenecks come up:
1. Quote turnaround. The business that gets a written quote back the same day wins work off the business that takes four days, at the same price. If your quote lives as a photo of a notepad until the weekend, automating the quote assembly is worth more in won work than it is in saved minutes.
2. Missed calls during the day. A crew on the tools cannot answer the phone, and a homeowner ringing three trades will book the one who answers. This is where an AI receptionist or even a simple missed-call-to-text response earns its keep — not by saving admin time, but by stopping the leak.
3. The job-notes-to-invoice gap. Every business has some version of a job that finished three weeks ago and still has not been invoiced because the notes are in someone’s head. Automating the capture at the point of work — a structured form or a voice note the crew records before they leave site — shortens the cash cycle far more than chasing does.
What a first project should look like
Pick one workflow. Instrument it so you can see the before and after. Ship it in weeks, not quarters. Then decide whether the second one is worth doing based on what the first one actually returned.
A first automation project that touches every part of the business at once is a project that takes six months and gets abandoned. One that takes quoting from 40 minutes to 15 is one the crew will actually adopt, and it pays for the next one.
Want the numbers for your business?
Run the table above for your own job and quote volumes and send us the total — or just tell us roughly how many quotes and jobs you do a month and where the pain is loudest. We will tell you honestly which of it is Bucket A, which is Bucket B, and whether the maths justifies a project at all. Get in touch.