The appointment was confirmed twice and the customer still didn't come

There are 2 different failures inside a no-show number and they need opposite responses. Some appointments were never real: set to end an awkward conversation, agreed vaguely over the phone, or booked by someone whose performance is measured on appointments set rather than customers who arrived. The rest were real and lost to friction — no vehicle held, nobody named to ask for, a journey nobody discussed, a confirmation nobody replied to. Reconfirming harder fixes the second population and simply annoys the first.

Which is why the first move is not a new confirmation sequence. It is a breakdown that tells you which of the 2 you have, and a store that keeps 3 extra columns on the appointment record can produce it in a morning.

The check: set-to-show, split three ways

1. Export 90 days of sales appointments: appointment id, set-at timestamp, scheduled-for timestamp, setter, setting channel, vehicle stock number, and arrival record.
2. Compute set-to-show per appointment as a binary arrived or not, using a single arrival definition applied to every row.
3. Group by setter. You are looking for the person with a high set count and a materially lower show rate than everyone else, which is an incentive problem rather than a process problem.
4. Group by channel: phone, text, online scheduler, in-person, third-party. Each produces a different quality of commitment and they should never be reported as one figure.
5. Group by lead time in bands — same day, next day, 2 to 6 days, 7 days or more. Show rate falls as lead time grows, and a store selling mostly 7-day appointments has a scheduling problem, not a customer problem.
6. Finally, cross-tabulate setter against lead-time band. The rows with high volume in the longest band and the lowest show rate are where the loss is concentrated.

Four causes, and the evidence that separates them

The appointments that were never real

This population is created, not encountered. A shopper who says maybe Saturday to end a phone call has not made an appointment, and recording one converts a soft signal into a hard number that then gets reported upward. Where a business development desk is measured on appointments set, the incentive does the rest, and the store ends up with a healthy-looking set count and a floor that has learned to ignore the diary.

The correction is a measurement one before it is a behavioural one. Report set-to-show by setter rather than sets alone, so that an appointment which never arrives costs the setter something in the same report that credits them for setting it. Then require the appointment record to carry a specific vehicle and a named colleague before it can be saved at all — a system that cannot store a vague appointment cannot report one.

An appointment with no vehicle and no named person is a note about somebody's intentions. Counting it as a booking is how a store lies to itself for a quarter.

Nothing was held, and nobody was named

A shopper agrees to come in on Thursday to see a specific used car. Nothing marks that car as spoken for, so on Wednesday it is sold, demonstrated to somebody else, or moved to another rooftop. Nobody tells the customer. They arrive, or more often they do not, having seen the listing disappear online the night before.

The parts counter solved this problem years ago with a written set of checks before a special order is raised — verify the part, verify the customer's commitment, record who authorised it — the discipline set out in checks before a special order is raised. A sales appointment deserves the same treatment, and the mechanics are not difficult.

• Apply a soft hold to the stock number for the appointment window, visible to everyone on the floor, with an expiry rather than a manual release.
• Name a colleague on the record, and make that name appear in the confirmation message. Ask for Priya at 14:00 is a materially different instruction from come in on Thursday.
• Alert on the hold breaking. If the vehicle is sold or moved, the appointment owner gets a notification the same hour, and the customer gets an honest call rather than a wasted trip.
• Offer a second vehicle at the point of breakage rather than at the door, using the availability check described in replying about a car that still exists.

Nobody discussed the journey or what happens when they get there

Two facts predict arrival and are almost never captured: how far the customer is travelling, and whether the visit fits into their day. A 40-minute drive on a Saturday afternoon with 3 other dealers closer to home is a fragile commitment, and it becomes more fragile with every day of lead time. Neither fact is hard to ask for, and both belong on the appointment record rather than in somebody's memory.

Expectation is the other half of it. A shopper coming in mainly to hear a part-exchange figure will not travel if they suspect no figure is coming, and will leave immediately if they arrive and one is still not available. Set that expectation before the visit, in the language described in qualifying a trade-in enquiry without quoting a number, and capture the appraisal inputs beforehand so the visit produces the number rather than starting the process of producing it.

A confirmation that asks nothing tells you nothing

Most reconfirmation sequences send 2 or 3 outbound messages restating the time. They are sent, they are delivered, and they produce no information at all, because a message with no question has no reply. Silence then gets read as agreement, which is precisely the wrong inference.

1. Ask a closed question that requires an answer: still good for 14:00, or would 16:00 suit better? Either reply is useful and either confirms a live human on the other end.
2. Send the first ask when the appointment is set, not the day before, so that a soft appointment collapses early enough for the slot to be reused.
3. Send the second ask the evening before, carrying the vehicle, the named colleague and the parking or arrival instruction. Practical detail raises arrival more reliably than enthusiasm does.
4. Treat no reply as a downgrade, not a confirmation. An unanswered appointment goes back into the working queue and someone calls it.
5. Stop after the second unanswered ask. A third message is noise, and the general design of stages, owners and stop conditions belongs to a follow-up cadence that survives a busy Saturday.

Fix in this order

1. Write down the definitions of set and show, and add setter, channel and lead-time band to the appointment record. Nothing else is measurable until this exists.
2. Make a vehicle and a named colleague mandatory before an appointment can be saved. This removes the never-real population at source rather than arguing about it.
3. Change the confirmation to ask a closed question, and start reporting reply rate alongside show rate.
4. Add the soft hold and the breakage alert on the vehicle. This is a small build against your inventory data, of the kind covered in our internal tools and operations work.
5. Only then revisit incentives. A setter target is defensible once set-to-show is visible per setter, and indefensible while it is not.

Two boundaries. Service-department no-shows are a genuinely different problem — different customer intent, different capacity constraints, different reminder economics — and folding them into this analysis produces a blended number that describes neither. And an appointment record is only as good as the enquiry it grew from, which is the argument for capturing enquiries durably in the first place, in wiring an enquiry form so nothing lives in an inbox. The rest of this silo sits under lead response and follow-up, within our wider automotive retail and aftermarket work.

• appointments
• no-shows
• sales process
• measurement