The process, as it actually ran
Every enquiry arrived in one inbox, mixed in with supplier mail, invoices, and the customer who has already booked and wants to move to Thursday. The owner read all of it himself, between calls, usually twice: once on the phone to see whether it mattered, and again at a desk when he had time to answer properly.
Nothing was written down about which messages deserved an answer. He knew. That knowledge was the whole system, and it lived in one head that was also driving to jobs.
The failure was never that a message got lost. It was that a quote sat half-written for four hours while the customer, who was ready to book, kept shopping.
The numbers
We timed the loop against his own mail over one period. Reading a message and deciding what it was took about four minutes. Writing a reply that was actually useful, against the thread and what the customer had bought before, took about eight. Sending it and getting the work into the calendar took about six.
Over that period: 239 messages read, 55 worth a real reply, 13 that became booked work.
That is 956 minutes of reading, 440 of writing, and 78 of booking. Just under 25 hours, to convert 13 jobs.
Read the funnel rather than the total and it gets sharper. Eighteen minutes of his attention per enquiry that converted, plus the four minutes each spent on the 184 messages that went nowhere. He was not slow. There were simply more messages than there were hours, and the sorting was the expensive part.
What changed
Classify on arrival, not on reading. Every message gets tagged the moment it lands: a new enquiry, a reply on a live quote, or something with money in it that a person needs to see. The four minutes of deciding what a message is were the cheapest minutes to give away, and there were more of them than anything else.
Draft against the history, not from scratch. The reply is written against the thread and what that customer has bought before, in his voice, because it was built from a year of his own replies. He edits and sends.
He talks to it in WhatsApp. Not a dashboard, not a login: the whole system sits behind the app already on his phone, and he texts it the way he would text someone who works for him. Where is that job. Send the quote. What is still open. It answers in the same thread, and the crew carry on exactly as they were, which is the part that makes it stick.
The reports write themselves and arrive the same way. The evening spent reconciling the day against a pile of texts is now a message that turns up with the day already reconciled: what came in, what went out, what is still waiting on somebody.
A person still sends. Anything that commits him to a customer waits for a yes from him first. That is not caution about the technology; it is that a quote is a promise, and the promise should come from the person who has to keep it.
What the board says now
His system keeps its own count of what it has run and what that gave back in hours, and the count moves every week. It lives on the case page rather than in this sentence, because a figure printed in a newsletter is only true on the morning it goes out.
Read whatever it says honestly, because the shape of the number matters more than its size. It is hours he did not spend, priced at his own rate. That is not revenue anyone measured. The hours are the real figure; the money is what those hours are worth if he does something else with them, and whether he does is up to him.
The number that matters most is not on the board at all. It is the quote that now goes out in the hour rather than in the evening.
If you are doing this yourself
You do not need us for the first part, and it is the part that pays. Spend one week noting, for every enquiry, when it arrived and when you answered. Not how long you took to write it. The gap between arrival and answer is the number that loses jobs.
Then sort your own inbox by hand for a week into those same three buckets. If you find, as most do, that more than half of what you read needed no reply from you at all, you have found the hours before you have spent a penny.
The rule underneath all of it: automate the sorting before you automate the writing. Sorting is where the volume is, it is where judgment matters least, and it is the part that can go wrong without anyone getting an email they should not have.
