An enquiry comes in. Somebody replies — maybe that day, maybe the next morning once the phone comes out of a pocket covered in plaster dust. Then, usually, nothing. Not because the job died. Because following up is not hard, it is just relentless, and it has no deadline attached to it. A missed callback has a customer ringing back to remind you. A quote gone quiet just sits there, and it is always the easiest thing to leave for tomorrow, because tomorrow never actually punishes you for it — until the customer has already booked with whoever else did chase them.
Where the gap actually opens
Picture a heat pump enquiry that comes in at 9pm on a Tuesday through the website form. It gets a reply the next morning, a proper one, with a price range and a rough idea of timing. The customer says “sounds good, let me check with my partner.” That is the last message in the thread. Nobody decided to drop it. It just moved to the bottom of a list that has a full day of jobs sitting above it, and by Thursday it is not on anyone’s mind at all.
Or a call comes in while a plumber is under a house with wet hands, goes to voicemail, and gets a text back within the hour saying someone will call. Someone does call, later, and the job gets quoted. Then the same thing happens: no second message, no check-in, no reminder three weeks later that the quote is still open. The customer is not being difficult by staying quiet — most of the reasons people go quiet after a quote have nothing to do with the price or the work, as the guide to why customers go quiet sets out. They got busy. They are waiting on someone else. They forgot. None of that means the job is lost. It means nobody reminded them it was still open.
The pattern repeats across trades with small variations. A roofer after a storm gets a burst of enquiries in a single afternoon and can answer the first few properly, but the ones that come in after 4pm get a one-line reply and no second touch. A painter loses jobs not because a competitor’s price was lower but because the competitor followed up and they did not. None of this is about skill on the tools. It is entirely about what happens to a message after the first reply, and that is the part almost every trade business quietly drops.
What AI-assisted follow-up actually does
Strip away the marketing language and the job is narrow. AI-assisted follow-up does four things, and only four:
- Sends the first reply immediately, at any hour. A form filled in at 9pm or a missed call at 6am gets acknowledged straight away instead of waiting for someone to next check a phone.
- Checks back on a quote that has gone quiet. If nobody has replied a few days after a quote was sent, a check-in goes out asking whether it landed and whether there are any questions — the exact message a busy tradie means to send and usually does not.
- Nudges an enquiry that never got a real answer. Sometimes the first reply itself goes unanswered. A short, low-pressure follow-up covers that gap too, rather than letting the enquiry simply die in the inbox.
- Stops the moment the customer replies, or the moment the owner steps in. Once a real conversation starts, or once the tradie picks it up directly, the scheduled messages stop. It is there to fill silence, not to talk over anyone.
That is the whole job. It is not selling the customer anything they were not already asking about, and it is not making decisions about the job. It is doing the one task that a busy trade business reliably fails to do on its own: remembering to check in, on a schedule, without anyone having to hold it in their head.
For a job that has actually reached the quote stage, the mechanics of timing and wording — how soon to check in, how many attempts, what a good message actually says — are covered properly on the quote follow-up page rather than repeated here.
The wording is yours, not the software’s
It is worth being straight about the mechanism, because pretending otherwise would not survive five minutes of a tradie actually using it. Part of the first response and the follow-up messages is handled by software, not typed out fresh by a person each time. There is no point dressing that up as something else — the answering side of that software is covered honestly on its own page, including what it should never be left to do.
What matters is what it is allowed to do with that role. It works from wording the business owner has approved in advance — the actual sentences, the actual tone, the actual boundaries on what gets offered. It does not invent a price on the spot, promise a start date nobody has confirmed, or improvise a commitment the business has not made. If a customer asks something outside that approved wording — a specific price, a change to the job, anything that needs a real answer — it hands the conversation to the owner rather than guessing. The value is not that it sounds like a person. The value is that it reliably does the one repetitive, time-sensitive part of the job, on schedule, without needing anyone to remember it exists.
That distinction — a defined, approved role, with a person still holding pricing and promises — is also what separates this from an unattended call-answering system. The guide comparing an AI receptionist with an answering service goes into that difference in more depth, because the two get talked about as if they are the same thing and they are not.
Where AI should not be making the call
This is the part that actually makes the rest of this page trustworthy, so it deserves real weight rather than a passing mention.
AI follow-up should not be quoting a job. Pricing depends on things a form field cannot capture — access, condition, exactly what the customer means by “leaking,” how much of the roof is actually affected. A number that has not been through a person is a number the business may have to honour anyway, and that is a bad trade for the sake of speed.
It should not be negotiating. A customer asking for a better price, a different scope, or a discount for cash is asking a business question, not a scheduling one. That belongs with the owner.
It should not be handling a complaint. Someone unhappy about a previous job, a delay, or a mistake needs a person who can actually own the problem, not a scheduled message that was written before the complaint existed. A generic follow-up landing in that inbox at the wrong moment reads as tone-deaf at best.
And it should not be anywhere near anything that sounds like an emergency. A burst pipe, a smell of gas, sparking at the switchboard — those need a human decision immediately, not a message queued for tomorrow. A business that lets software run past what it is actually built for damages a customer relationship faster than slow follow-up ever would. The whole point of the system is to catch the enquiries a busy business would otherwise drop, not to replace the judgement a real job needs. Every one of those situations should hand off to a person immediately, and a business setting this up should be blunt with itself about where that line sits before anything goes live, not after it has already gone wrong on a real customer.
How much follow-up is reasonable
There is a genuine question underneath all of this that is worth asking honestly: how many times is reasonable before a follow-up stops being helpful and starts being pestering?
There is no universal number, but the shape of a reasonable approach is fairly obvious once you say it out loud. A short check-in a couple of days after a quote, another a week or so later if there has been no response, and then stopping — that reads as organised. A message every day with no acknowledgement that the customer has gone quiet on purpose reads as harassment, and it will cost the business more goodwill than the job was worth.
Two things matter more than the count. First, the customer has to be able to reply at any point and get a real person, not another scheduled message pretending to be a conversation. Second, there has to be an easy, obvious way to say “not interested” or “stop contacting me” and have that actually be respected straight away. A follow-up system that cannot take no for an answer is not doing its job — it is just being annoying with better timing.
There is also a genuine legal question sitting alongside the tone question — what messaging channels require consent, how that consent needs to be recorded, and what NZ rules apply to text and email follow-up specifically. That is not something to take on the word of a website guide. It is worth getting proper advice from a lawyer or from whoever provides the messaging platform before setting the rules for how a business follows up, rather than assuming general good manners are the same thing as being compliant.
What this actually changes
None of this replaces the tradie’s judgement about pricing, scope, or which jobs to take — that stays exactly where it already is. What it changes is much narrower and much more useful: the enquiries that used to depend on someone remembering to chase them now get a scheduled, consistent check-in instead, and the ones that need a real person still get one, immediately.
That is the whole trade-off, and it is worth being honest about both sides of it. The gap between a first reply and a booked job is where jobs already paid for in enquiries quietly go missing, not because the work was wrong for the customer but because nobody followed up. Closing that gap without dropping the parts that genuinely need a person is the actual goal — not sounding clever, not sounding automatic, just not letting an open quote go cold from neglect.
If it would help to see where your own front desk is leaking enquiries between first contact and a booked job, a free front desk audit looks at exactly that — where replies are slow, where quotes go quiet, and what is realistic to fix first.