Promise time: the number the whole service day is measured against

Promise time is the time the customer was told their vehicle would be ready. It is a commitment made to a person, usually by the service advisor at write-up, and it is the number the customer measures the whole visit against. It is not the shop's estimate of when the work will actually finish — that is a separate number, owned by different people, and a system storing one field for both loses the difference precisely when the difference matters.

The two diverge on an ordinary day, not a catastrophic one. A part is a day out. An approval sits unanswered from 11 until 4. A diagnosis takes 2 hours instead of 1. The shop knows by mid-morning; the customer finds out at collection, which is the conversation the whole department then spends its afternoon having.

Who says the number, and what they know when they say it

An advisor sets it at write-up, in the 3 to 6 minutes they have at the lane, holding the least information anyone will ever hold about that job. There is a symptom rather than a diagnosis, a parts list that does not exist yet, and an authorisation limit that may not cover what gets found. Before any of it can be planned, the symptom has to become something the shop can schedule against, which is the mapping layer in op codes behind a customer-facing menu.

Where the booking arrives online, nobody says it at all: the promise time is whatever the calendar implied when it offered a slot. That makes the capacity model behind the calendar the real author of the commitment, and a slot model standing in for a capacity model is exactly why the online calendar keeps selling 8 o'clock on Monday.

Three inputs, and the two that get skipped

• Capacity by skill, not bay count. The binding constraint is usually hours from a technician qualified for that work, and a shop with 8 bays and 1 diagnostic technician can promise 8 jobs and deliver 1 of the ones that matter.
• Parts certainty. On the shelf, on tomorrow's van, or on back order are 3 different promises, and this is the single largest determinant of whether the number holds. A promise time set before anyone looked at stock is a guess with a customer attached to it.
• Approval risk. Any job carrying a second decision — an inspection, a diagnosis that becomes a quote — has a queue nobody schedules. Historic approval latency for that job type belongs in the number, so a job with an inspection attached carries a wider promise than a fixed-scope service.
• Underneath all 3 is what the vehicle actually is. Labour times and parts hang off the specific build, and a decode gives you less than people assume, which is what a VIN decode leaves out.

One field, two numbers, and the slippage nobody sees

Storing one field forces a choice with no good answer. Update it as the job slips and the record of what the customer was told is gone, so nobody can say afterwards whether the promise was missed or simply moved. Leave it alone and the shop has no working forecast anywhere in the system, so the first person to know is whoever is standing at the desk at 5.

What the second field buys on an ordinary Tuesday

• Slippage becomes an event. The moment the estimate crosses the promise, something can fire — hours before the customer would otherwise notice. Whether that message prevents a call or causes one is a question of content, argued in the texts are sending and they still call.
• A revised promise is a new commitment and gets recorded as one. A customer moved from 2 to 4 has been promised something different, and the second promise deserves the same evidence as the first.
• Collection stops being a queue. A vehicle finished at 3 and paid for at 5 occupies a space for 2 hours, which is the case for a hosted payment handoff attached to the repair order — what a pay-by-link service flow is.
• Measurement finally works. Promise-time attainment by advisor and by job type is computable, and so is the more useful number underneath it: how far ahead of the miss the shop knew.

A shop with one date-time field cannot tell you whether it is late. It can only tell you what it currently intends.

The words attached to the number

A bare time is heard as a guarantee. Ready by 3 and we will call you by 3 are different commitments, and only one of them survives a parts delay. Where the job has a decision stage in it, promise the decision rather than the completion: we will call you by 11 with what we have found and a firm time is a promise the shop can keep on almost every day, and it converts one broken promise into 2 kept ones.

Automating that is unglamorous and narrowly scoped: read the job state, fire on 2 events — the estimate crossing the promise, and the decision point being reached — and never invent a time the shop has not committed to. That is the kind of thing we scope as AI agents and automation, with the failure modes it has to be built against in what actually breaks with agents in production. This page sits in service booking and customer communication, part of our automotive retail and aftermarket work.

• service department
• definitions
• customer communication
• scheduling