Half your enquiries arrive when the store is shut

Start with the arithmetic, because it usually ends the argument about whether this is worth attention. A store trading 09:00 to 18:00 Monday to Saturday is open for 54 of the 168 hours in a week. The other 114 hours — 68% of the week — are hours in which people browse listings, submit forms and send messages, and in which your reported response time is being computed from something other than a person answering.

The diagnosis has 2 parts, and most stores get the second one wrong. First, how much genuinely arrives while you are shut, by hour rather than in aggregate. Second, how long those enquiries actually wait for a reply from a human being — which is a different number from the one on your dashboard if an automated acknowledgement is being counted as a response.

The check: an arrival histogram with a response overlay

1. Export 60 days of leads with a creation timestamp accurate to the minute, the source, and the timestamp of the first outbound human message or call.
2. Convert every timestamp to store-local time. Vendor feeds and mail logs frequently arrive in another zone, and a 5-hour offset invents an overnight problem or hides one.
3. Bucket arrivals into 24 hourly buckets, then split by weekday. The pattern on a Sunday is not the pattern on a Tuesday and averaging them produces a number describing no real day.
4. Overlay median minutes to first human reply on the same buckets. You are looking for the shape, not the average: most stores find 3 or 4 hours carrying the great majority of the delay.
5. Mark the buckets that fall outside opening hours. Sum the leads in them and express it as a share of the 60-day total — that is your real after-hours volume, and it is usually different from what the team believes.
6. Separately, count leads that arrived while shut and received no human reply at all before the following evening. That figure is the one worth taking to a manager.

Four causes, and what separates them in the data

Someone was answering, and had nothing to answer with

This is the failure mode that looks fixed on a dashboard. An after-hours service or an automated responder replies in 2 minutes, the response-time metric is excellent, and nothing improves, because the reply could not confirm whether the car exists. A shopper who asks about a specific stock number and receives a friendly message that does not mention that vehicle has learned nothing, and has 4 other tabs open.

A useful overnight reply needs to resolve the enquiry's vehicle reference against live inventory before it goes out, including the 3 ways a stock number fails to resolve — the sequence set out in replying about a car that still exists. That in turn needs an inventory read you control rather than a screen a person logs into during the day. The general argument for holding your own copy of reference data, in a schema you can query at 02:00, is the same one made about parts in loading fitment data into a schema you control.

A fast reply that names no vehicle is not a fast reply. It is a fast way of telling a shopper you were not able to look anything up.

The acknowledgement that is counted as a response

Almost every store has an automatic first message, and almost every store's response-time report is built on whatever populates the first-touch field. When those are the same thing, the metric measures the reliability of a mail server. The tell is a median response time that is implausibly good — seconds, or under a minute — sitting alongside overnight enquiries that everyone knows were answered at 10:00 the next day.

• Split the field. Record acknowledgement time and first human reply as 2 separate columns, and report both. Nobody argues with the number once they can see the gap.
• Exclude system accounts from first-touch calculations, by account rather than by message content, because message templates change and account ids do not.
• Keep the acknowledgement. It is worth sending and it does set expectations — it simply is not the thing being measured.
• Fix the capture path while you are here. An enquiry that sits in a shared inbox until someone forwards it has a first-touch timestamp that means nothing, which is the argument for wiring an enquiry form so nothing lives in an inbox.

The morning queue, worked in the wrong order

At 09:00 someone opens the CRM and starts at the top of the list. Most lists are sorted newest first, so the enquiry that arrived at 21:30 last night is answered before the one that arrived at 22:45, and the one from 19:00 waits until last. That inverts priority precisely: the oldest overnight enquiry is the one with the longest head start on your competitors, and it is the one most likely to have already booked an appointment somewhere else.

1. Sort the overnight queue oldest-first and make that the default view rather than a setting somebody has to choose.
2. Work it before anything else at open, including walk-in preparation, and give the queue a named owner for the first hour of the day.
3. Cap the queue by pre-assigning overnight leads to the people rostered for the morning, so nobody starts the day negotiating who takes what.
4. Measure the drain: minutes from opening until the overnight queue is empty. A store that clears it by 09:30 has a different problem from one that clears it by 13:00, and only one of them needs coverage.

What an after-hours reply can honestly say

The content is more consequential than the speed. A message that promises what you cannot deliver converts an unanswered enquiry into a broken promise, which is worse. Four things can be said truthfully at 23:00 by a system with an inventory read and a rota.

• That the specific vehicle is still listed, named by its make, model, derivative and stock number. This is the only sentence in the message that a shopper is really reading.
• When a person will be in touch, expressed as a time rather than as soon as possible, and derived from the rota rather than from optimism.
• One question worth answering overnight — a part exchange, a preferred contact time — asked so that the morning conversation starts further forward.
• What the reply cannot do, said plainly. Declining to give a figure and saying why is better received than an evasive message, and the pattern for that is in qualifying a trade-in enquiry without quoting a number.

Deliberately out of scope here is the choice between an outsourced answering service and an automated first reply — that comparison has its own page in this cluster, and it is only answerable once you know your overnight volume by hour and whether a useful reply needs live inventory. Once overnight replies do start booking appointments, the next number to watch is how many of those appointments arrive, examined in the appointment was confirmed twice and the customer still didn't come. The rest of this silo sits under lead response and follow-up, within our wider automotive retail and aftermarket work.

• response time
• after hours
• lead handling
• measurement