The state of trades admin, 2026.

How to read this

This is a field report, not market research. The evidence base is the businesses I audit through the Tech Leak Audit, the trades websites I assess through my audit tooling, and the two field-service businesses I run myself. That is a deep look at a narrow slice: owner-operated shops, mostly 2 to 20 staff, mostly Victorian. The rule for numbers here is simple: I will not dress pattern observation up as statistics. Where this report gives a figure, it is a raw count from my own audit dataset with the sample size attached (see the numbers section below, n = 49); everything else is stated as what it is, pattern observation.

The headline: admin is the ceiling, not demand

Almost no owner I audit has a demand problem. The phone rings, the referrals come, the suburbs are growing. The ceiling is operational: the owner is personally the booking system, the quoting department, the dispatcher, and the debt collector, in the hours left after the tools. Growth stalls not because work dries up but because the owner's evenings are already fully spent converting the work they have.

The most common business model in the Australian trades is one skilled person, on the tools all day, doing three office jobs at night, for free.

The five leaks, ranked as I find them

1. The front door leaks jobs (and the owner never sees it)

The most damaging leak is the one that leaves no evidence. A missed call while under a floor does not appear in any report; the customer simply books the next number. Websites in this trade are overwhelmingly brochures: they describe the business but cannot take a booking, so every enquiry funnels back to a phone the owner cannot answer. The pattern is worst after hours, exactly when homeowners do their comparing.

2. Quoting eats the evenings

The quote is the most expensive document in the business. It is drafted at night because the details arrived incomplete, priced from memory because the parts list lives in the ute, and sent late because the owner is human. Late quotes lose winnable jobs; rushed quotes win unprofitable ones. Both failures are intake problems, not effort problems, and intake is fixable with structure.

3. The owner is the dispatch system

Jobs live across group texts, two calendars, a whiteboard and the owner's memory. It works, right up until the second crew, the first big commercial client, or the first week the owner takes off. What I flag in audits is not the chaos itself but the key-person risk: in most shops, if the owner's phone died, tomorrow would not happen.

4. Margin is archaeology

Most owners find out whether a job made money when the bookkeeper closes the month, which is weeks after the levers stopped working. Mid-job, the honest answer to "are we still in profit on this one" is a shrug. This is the leak owners feel least and the one that moves the yearly number most, because it silently decides which jobs the business keeps saying yes to.

5. The re-keying tax

The same job details typed into the enquiry email, the quote, the job app, the invoice, and Xero. Ninety minutes every Friday that should be a button. It is the least dramatic leak and the easiest to automate, which is why it is usually the first thing I fix under an Ops Care plan: quick win, visible relief, trust earned for the bigger fixes.

What changed in the last two years

• Customers now expect self-service. The homeowner comparing three sparkies at 9pm books the one with online booking. This expectation arrived from every other industry they buy from, and it does not care that trades run differently.
• The software stack fragmented. The typical shop now runs a job app, Xero, a supplier portal and three spreadsheets. Each tool is fine; the gaps between them are where the admin lives. The industry sold tools; nobody sold the joins.
• Multi-year platform contracts got teeth. Operators are reading renewal letters with compounding increases on terms they signed in a hurry. The appetite I hear for "own the fix instead of renting it" is directly downstream of those letters.
• AI noise arrived before AI usefulness. Owners are being pitched "AI transformation" weekly. The useful version in a trade business is boring: automations that move job details between systems without a human. The transformative version being sold rarely survives contact with a Tuesday.

What actually fixes it

The pattern across every audit that turned into a happier owner is the same, and it is not a platform migration:

• Fix one leak at a time, starting with the one bleeding most. Whole-of-business transformations stall; single fixed-scope fixes ship.
• Keep the tools that work. The job app and Xero stay; the fix goes in the gaps around them.
• Make it someone's job to keep fixing. Systems rot when nobody owns them. That standing ownership is the entire reason Ops Care exists: monitoring, fixes, and one new automation every month, so the leak list keeps shrinking instead of regrowing.

The first numbers (n = 49)

Added July 2026: the first batch of hard numbers from my website audit tooling. The sample is 49 Victorian trades and service business websites audited through my prospecting pipeline. Full disclosure on the sample: these are businesses I sourced for outreach, not a random draw of the industry, so treat the numbers as "what the audit tool finds in the wild", not a census. Every figure is detected from the public website, with the sample size attached, as promised.

• 4 of 49 could take a booking online. The other 45 route every enquiry to a phone that is on the tools all day. This is the front-door leak, measured.
• 22 of 49 had no structured quote or enquiry form at all beyond a phone number or bare email link. The job details arrive incomplete because there is nothing to structure them.
• 24 of 49 had any analytics installed. Half of these businesses have no way of knowing what their website does or loses.
• 9 of 49 showed their reviews on their own site. The proof exists on Google; the website just doesn't use it.
• 13 of 49 had a DMARC email policy. For the rest, the quotes and invoices they send are measurably easier to spoof and more likely to land in spam, and nobody has ever told them.
• 2 of 49 showed any public-facing trace of job-management software (a booking link, a portal, anything a customer touches). Whatever runs in the office, the customer never sees it, which is the joins problem again.

What the dataset cannot yet say: how fast these businesses quote, and whether the owner can state margin mid-job. Those are audit-call questions, not website scans, and they get published the same way when the sample is big enough to mean something.

Take the numbers with you: download the raw counts as CSV or JSON. Every figure is a raw detection from the audit corpus as at July 2026; cite it, argue with it, re-run it.

The scored cut of the same dataset, with the three patterns that kept repeating, is written up separately: I checked 48 Victorian trades websites.

If you want your business in the (anonymised) dataset and a free written finding for yourself in the process, the door is the Tech Leak Audit.