A trade day is a sequence of promises about time.
You will be there Tuesday. You will be there between eight and twelve. You are twenty minutes away. You will come back on Thursday to finish. Almost every one of those promises is currently kept or broken by hand, in gaps between jobs, on a phone with one bar of reception and dirty hands.
That is why they get missed. Not because anyone is disorganised, but because the moment a promise needs communicating is exactly the moment you are least able to communicate it.
This is the same constraint that makes the phone a problem for trades, where calls arrive while your hands are full. The difference is that a missed call costs you a job, whereas a broken promise about time costs you a review, and the second one compounds.
The allocation is the hard part, and mostly should stay human
Worth saying up front, because this is where the software marketing concentrates and where the realistic return is smallest.
Deciding who goes to which job in which order looks like an optimisation problem. It is presented as one. In practice it depends on a pile of things that live in your head and have never been written down: which apprentice you would not send to that customer, which suburb is unbearable at four o’clock, which client will keep you an extra hour every single time, which job you want done before the rain arrives.
Systems that promise to solve this tend to produce a run sheet that is theoretically efficient and practically wrong, and then everyone quietly goes back to the whiteboard.
So leave the allocation alone at first. The money is in everything that happens after the run sheet exists.
The messages that run on rules
Once a job is in a calendar with a customer attached, a specific sequence of communications becomes mechanical. Every one of these is a rule, none of them needs AI, and almost none of them happens reliably by hand:
On booking. A confirmation with the date, the window, the address you have on file and what you need from them. Access, parking, pets, someone over eighteen present. Getting this wrong is what causes wasted trips.
The day before. A reminder with an option to reschedule. This is the single cheapest thing you can automate and it cuts no-shows immediately.
On the way. Sent when you close out the previous job or leave a geofence, saying you are en route with a rough arrival time.
If you are running late. The important one, covered below.
On completion. A summary of what was done, the invoice, and anything the customer needs to know.
A few days later. A check that everything is still working, and a review request if it is.
That is six messages per job. Most trade businesses send two of them consistently.
The window problem
Arrival windows exist because travel time is unpredictable, and they are the main source of friction between trades and customers.
The instinct is to give a wide window to protect yourself. Eight to twelve. Sometimes Tuesday. The cost is that the customer writes off half a day, resents it, and mentions it in the review.
Automation does not shrink the window, but it does something better: it lets you narrow it progressively as the day resolves. A four-hour window on booking becomes a two-hour window the evening before, becomes a twenty-minute estimate when you actually leave the previous job. Each of those is a message the customer did not have to chase.
This is worth more than it sounds. Most complaints about tradespeople are about uncertainty, not lateness. Somebody who knows you are twenty minutes away will happily keep waiting. Somebody staring at a four-hour window with no information will ring you twice, which costs you two calls you did not want.
Running late is the message worth building first
If you automate exactly one thing from this post, automate this.
The running-late message is the highest-value communication in the entire chain, and it is the one that is almost never sent, for an obvious reason: nobody wants to make that call. It is uncomfortable, the customer will be annoyed, and it is very easy to tell yourself you will make up the time on the way.
A system has no such reluctance. If the job you are on is running past its expected finish, the next customer gets told, automatically, without you having to decide to be the bearer of it.
The difference in how that lands is large. A customer told at nine that you will now be there at two is mildly inconvenienced. The same customer, finding out at one that you are not coming at eleven, is writing a review. Same delay, entirely different outcome, and the only variable is whether anybody told them.
This is the same principle as the reliability of automated payment chasing. The value is not that the machine is clever. It is that the machine is not embarrassed.
Cancellations and the empty slot
A cancellation costs twice: the revenue from the job, and the hole in a day that is now unrecoverable because you found out at eight in the morning.
The mechanical parts of handling this automate cleanly. Confirm the cancellation, update the calendar, notify whoever is affected, and offer the freed slot to a waiting list. That last step is the one most trades do not have, and it is not complicated: a list of customers who said they were flexible, contacted in order when a gap appears.
What should not be automated is the judgement attached to it. Whether a late cancellation gets a fee, whether a good customer gets it waived, whether this is the third time and the relationship needs a conversation. Those are commercial decisions about a person, and they belong with you. Where that line sits for your business is exactly the sort of thing that should be written into your operating limits.
Travel time is the thing nobody schedules
A small point with a disproportionate effect.
Most trade calendars book jobs, not journeys. The eleven o’clock is entered as eleven, and the forty minutes of driving to get there is assumed to fit somewhere. It does not, and the day slips a little further with each job until the last customer is the one who gets rung at five.
If your job software supports travel buffers, turn them on. If it does not, the connection layer can add them when a job is created, based on the previous job’s address. It is a small piece of work and it is the difference between a schedule that describes the day and one that merely lists it.
What this is worth
Be realistic about which number moves.
This work does not usually win you more jobs. What it does is remove a category of friction: fewer wasted trips, fewer chase-up calls coming in, fewer bad reviews about waiting, less of your evening spent apologising. For a business running five to ten jobs a day, that is a meaningful amount of time and a measurable change in how customers describe you.
If you want it expressed as a number before committing, count two things over a fortnight: wasted trips where nobody was home, and inbound calls that were only somebody asking where you are. Both should fall close to zero, and both are easy to count now.
The general method for putting a number on this kind of work before you build it is in the piece on what AI automation costs a small business, and our rates for the connection work are published.
Where to start
Turn on what you already own first. Most Australian job management platforms send confirmations and reminders and a good proportion of the trades running them have never switched the features on. That costs nothing and covers two of the six messages. If you are not sure which layer of your setup should be doing what, the stack is laid out separately.
Then build the running-late message, because it is the one no product sends by default and the one that changes how customers feel about you.
Then the rest of the chain, in the order the job actually moves through it. This is the same discipline as mapping the process before automating it: follow one real job from the phone call to the payment, write down every point where somebody had to remember something, and automate those in order.
If you want a hand working out which of the six is missing in your setup, tell me what a normal Tuesday looks like and I will tell you where the gaps are. Often the answer is that you already own the software and it just needs configuring, which is a much smaller job than a build.
Frequently asked questions
What parts of trade scheduling can actually be automated?
The communication around the schedule, reliably. Confirmations, day-before reminders, on-the-way messages, running-late notices and follow-ups all run on rules and need no judgement. The allocation itself, deciding who goes where in what order, is harder to automate well because it depends on things that are rarely written down, like who is good with difficult customers.
What is an on-the-way message and why does it matter?
A text sent automatically when you leave the previous job, telling the customer you are on your way and roughly when you will arrive. It matters because most complaints about tradespeople are about uncertainty rather than lateness. Someone who knows you are twenty minutes away will wait happily. Someone staring at a four-hour window will not.
Should I automate telling customers I am running late?
Yes, and it is the highest-value message in the whole chain, precisely because it is the one nobody sends by hand. Nobody wants to make that call. A system has no such reluctance, and a customer told at nine that you will be there at two is far less annoyed than one who finds out at one.
Can automation handle rescheduling and cancellations?
It can handle the mechanics: offering a new time, updating the calendar, notifying whoever needs to know, and offering the freed slot to someone on a waiting list. What it should not do is decide whether a cancellation gets a fee waived. That is a commercial judgement about a relationship, and it should stay with a person.
Do I need new software to do any of this?
Usually not. Most job management platforms used by Australian trades already send some of these messages, and the ones they do not send can be triggered from the same system by a connection layer. The gap is more often that the features are switched off or were never configured than that the software cannot do it.
Wondering what this would look like in your business? A short chat is usually enough to tell.
Let’s chat