What Happens to the Calls That Come In After You Close

Every service business has an hour when the lights go off. Demand does not observe it. A water heater fails at 9pm. Someone reads three reviews at 11pm and calls before they lose their nerve. A driver gets rear ended at 7:40 on a Friday and starts dialing law firms from the shoulder of the road.

The question is not whether those calls happen. It is what the caller does in the sixty seconds after your voicemail greeting starts playing.

The four exits from an unanswered call

A caller who reaches your voicemail after hours takes one of four paths. They hang up and immediately dial the next search result. They leave a message. They hang up intending to call back tomorrow, and then their Tuesday happens and they never do. Or they book with whoever answered.

Only one of those four leaves a trace inside your business. Voicemail shows up. The other three are invisible, and they are the majority. This is the measurement problem at the center of the whole topic: your CRM is a record of the people who reached you. It is structurally incapable of telling you about the people who did not. The only honest record of what happens after 5 is your phone system's raw call log, which almost nobody reads.

We keep the published figures on call abandonment and missed call behavior on a separate page, our missed call statistics for Las Vegas, so we are not going to restate them here. This post does something different. It builds a model you can run against your own log, using your own numbers, so the output is defensible rather than borrowed.

The model has four inputs

Everything reduces to four numbers, and you can get three of them yourself in about twenty minutes.

The first is after hours call volume: pull ninety days of your call log, filter to calls that arrived outside staffed hours, and divide by thirteen to get a weekly figure. The second is your live answer conversion rate, meaning the share of answered calls that become a booked appointment, a work order, or a signed matter. Most owners know this within a few points. The third is average transaction value on a first visit or a first matter. The fourth, the only one you cannot measure directly, is the recovery rate: the share of missed callers you can still win back with a fast follow up. That one has to be an assumption, and it should be a conservative one.

Three worked scenarios

The table below runs the model for three business types common in the valley. Every figure in it is our own assumption, chosen to be plausible and deliberately modest, not a published statistic. Replace each column with your own numbers and the arithmetic still works.

A few things worth noticing about the shape of these numbers.

The law firm column has the lowest conversion rate by a factor of five and still lands within striking distance of the auto shop, because one signed matter is worth six work orders. In high value, low frequency businesses, a single recovered call per month can pay for the entire system that recovered it. That is not a marketing claim, it is just what happens when you divide $2,500 by anything.

The auto repair column is the largest because volume compounds. Twenty five calls a week is only three or four a night, which feels like nothing when you are standing in the bay. Annualized at a thirty percent close rate, it is a technician's salary.

And the med spa column, the smallest of the three, is still roughly $25,000 of recovered revenue at a recovery rate we chose to be pessimistic. If your real recovery rate is half of ours, the number is $12,500 and the conclusion does not change.

The cost side is not the interesting part

The standard response to a missed call is a text message sent within a minute or two, because a text arrives while the caller is still holding the phone and still in the mindset that made them dial.

Here the arithmetic gets almost comic. Take the med spa: twelve after hours calls a week, two messages each, is 1,248 messages a year. US SMS is priced per message segment and the current rate is published on Twilio's US SMS pricing page. At any per segment rate in the range that page has historically shown, the annual messaging spend for that volume lands in the tens of dollars, against a five figure exposure. Check the live rate before you budget, since carrier fees change, but the ratio is not close enough for the exact number to matter.

The real cost is never the messages. It is the system that decides who gets one, what it says, and what happens when the caller replies at 11:40pm. We broke that down separately in our post on what an AI receptionist actually costs, including the parts vendors leave out of the sticker price.

What to do with this in the next week

• Export ninety days of call detail records from your phone provider, not your CRM.
• Filter to calls that arrived outside staffed hours and divide by thirteen.
• Multiply by your real close rate and your real average ticket to get weekly exposure.
• Cut the result in half twice, once for callers who were never going to buy and once for callers you cannot reach again.
• If the number that survives both cuts is still uncomfortable, you have a problem worth solving. If it is not, stop reading about this and go work on something else.

That fourth step is the one people skip, and it is the one that makes the exercise honest.

Where this model is wrong

It assumes every after hours call is a prospect. In reality a meaningful share are existing customers, vendors, robocalls, and wrong numbers, and none of those carry new revenue. Your call log will tell you the mix if you sample fifty of them by hand.

It assumes the caller who books with a competitor was permanently lost. Some fraction come back later at full value.

It treats a recovered call as equal in value to a live answered one, which is generous. A caller you win back tomorrow morning has had time to compare, and may negotiate harder.

And it says nothing about what a bad automated response costs you. A text that reads like a machine, sent to someone who just had a car accident, is worse than silence. The mechanics of doing this without damaging the relationship are covered on our page about missed call recovery.

Run the model on your own log first. If you want a second set of eyes on the output, our free 15 minute audit will walk your call data with you, and you can read what that covers on the missed call recovery page.