Approve quotes from the van: a WhatsApp workflow for trades

Most trades businesses don't lose work because the phone went unanswered. They lose it because the quote sat in a notebook until 9pm, and by then someone else had already sent theirs. This post walks through a quote approval workflow you can run between jobs: the AI drafts it from the call, it waits in your approvals inbox, WhatsApp nudges you, you tap approve, and the customer gets their confirmation. You still make every decision.

The bottleneck is after the call, not during it

Missed calls get all the attention, and fairly — a call that rings out is a job that rings the next name on Google. We've written the arithmetic on that in what missed calls really cost a UK trades business.

But plenty of trades answer the phone perfectly well and still lose the job. The call happens at 11:20am while you're halfway through a bathroom. You scribble a name and a rough job on a bit of plasterboard offcut. Then it's the next job, then the merchants, then home. The quote gets written at 9pm if you've got anything left, and at the weekend if you haven't.

So the enquiry-to-quote gap is a day, sometimes three. The customer has usually rung two or three people. Whoever gets back first with a clear price and a date is very often the one who gets the work — not because they're cheaper, but because they made it easy to stop shopping.

The fix isn't working later. It's shrinking the admin between the call and the customer's phone buzzing, until it fits in the two minutes you spend sitting in the van with the engine running.

What the workflow actually looks like

Here's the flow, step by step, with nothing dressed up.

1. The call comes in and you can't take it. One dial code forwards only the calls you miss. When you can answer, you answer — business as usual. Your number stays the number on the van.
2. The AI receptionist answers and takes the details. Who's calling, what they need, when they want it. It runs 24 hours a day, so the 9:40pm boiler call is handled the same as the 9:40am one.
3. It drafts a quote or booking from that conversation. Using your standard prices and your availability — for example: Boiler service — £95, Thursday 2pm.
4. The draft lands in your approvals inbox and you get a WhatsApp update. You get a message after every call, so you know what happened whether or not there's something to approve.
5. You read it and choose: Approve, Edit or Reject. Approve if it's right. Edit if the price or the slot needs changing. Reject if it's not a job you want.
6. Only then does the customer hear anything back. Nothing reaches them without your yes. Once you've said yes, the confirmation goes out with the price and the time on it.

The whole review, for a standard job, is a ten-second read and a tap. You can do it standing at the back of the van with a brew. The bit that used to take twenty minutes at the kitchen table — remembering the call, finding the notes, writing it up properly, sending it — has already been done by the time you look at it.

Do the maths on your evenings

Why the approval step is the point, not the friction

It would be technically easy to let an AI confirm jobs by itself. We don't, and you shouldn't want it to.

Prices in the trades are rarely fixed. A boiler service is £95 until the customer mentions the airing cupboard is boarded in. A callout is £80 unless it's the third one this year for the same landlord. You know a hundred things about your prices, your patch and your diary that no system knows, and the moment a machine commits your business to a date you can't make, you own the mess.

So the split is deliberate: the AI does the work, you make the decisions. It handles the conversation, the note-taking and the drafting. You handle the yes. If a draft is wrong, Edit is right there — change the price, move the slot — and the corrected version is what the customer sees.

There's a secondary benefit that owners tend to notice about week two. Because every job passes under your eye as a one-line summary, you end up with a much clearer picture of what's actually coming in. Calls are recorded and transcribed too, so if a customer later says "I was told £95 all in", you can check rather than argue.

Setting it up so approvals are genuinely one tap

The workflow only stays fast if the drafts are usually right. That's on you as much as the AI. A few things that help:

• Write down your standard prices. Service, callout, first hour, out-of-hours uplift. Anything with a fixed number should have a fixed number the AI can use, so those quotes come through ready to approve.
• Be clear about what you won't quote on the phone. Full rewires, refits, anything that needs eyes on it. For those, the right outcome is a booked survey, not a price.
• Set your real availability, not your optimistic availability. If you never take a job before 9am, don't let Thursday 7:30am appear in a draft.
• Decide your out-of-hours rule. Do you want emergency calls at 2am pushing a notification, or do you want them queued until 7am? Both are defensible; pick one and stick to it.

Handle approvals in batches

Two or three sweeps a day beats checking constantly. Between jobs, after lunch, and before you leave the last site. Anything urgent will be sitting at the top. This keeps the customer's wait to a couple of hours at worst, which is still faster than nearly anyone else quoting the same job.

Reject is a real option

Some enquiries are not for you: the wrong side of the county, a landlord you've had trouble with, a job that's beneath your minimum. Rejecting is a tap, and it takes the guilt out of it — nobody's been promised anything.

Where this doesn't fit

Some honest limits, because an approvals workflow is not a fit for everyone.

There's also a fair objection about tone. Some customers want to talk to the boss, and no summary card replaces the five minutes of chat where you work out what they actually want. If your business runs on that, treat the AI as cover for the calls you'd otherwise miss, not as a replacement for the ones you'd enjoy taking. We've set out that trade-off in more detail in an honest comparison of AI receptionists, answering services and voicemail.

What it costs to run

Plans start at £29 a month with 100 call minutes, £69 for 300 and £115 for 600, with extra minutes at £0.25 each. Every feature is on every plan — approvals aren't an upgrade. There's a 7-day free trial with 30 call minutes and no card needed, and no contract. The full breakdown, including how the overage adds up on a busy month, is in what you'll actually pay.

Setup is deliberately small: give it your website, and it's live in about a minute. You keep your existing number, so there's nothing to reprint and nothing to tell your customers. If you want the caller's side of it — what they hear, how the job gets logged — that's covered in what actually happens when an AI receptionist answers.

The test of whether this workflow is worth it is simple, and you can run it this week without buying anything. Write down the time of every enquiry you take, and the time your quote actually went out. If the gap is under an hour, you don't need this. If it's routinely a day or more, that gap is where your lost jobs live — and it closes from the van, not the kitchen table.