Pick one lead. A real one, from your own system, that came in after 6pm on a weekday or any time on a Saturday. Now trace what happened to it, minute by minute, from the moment it arrived to the moment somebody spoke to that person or gave up.
Do this once and you will find something. Almost everybody does.
Here's the sequence to walk, in order. Write down a timestamp for each step or a blank if it never happened.
The lead arrives. What time exactly, and through what path? Form, call, text, marketplace?
What happened in the next sixty seconds? Anything automatic? A confirmation, a routing rule, a notification to somebody's phone?
Who was notified, and did they see it? Not was it sent. Did a human being look at it that night?
When was the first outbound attempt? Timestamp it.
When was the first two-way conversation? Timestamp that separately. The gap between these two is often the whole story.
What did the seller do in between? Check for inbound activity you didn't respond to. A second call. A reply text. That's a motivated person trying again.
The step that comes back blank most often is number three. Something fired, a notification went somewhere, and nobody looked at it until morning. That's not a broken system. That's a system working exactly as built, and the build assumed somebody would be watching.
Ask who checks the main line's voicemail on a Saturday. In most shops the honest answer is that it depends, which means nobody. A voicemail box with no named owner and no schedule is a place leads go to be found on Monday, and Monday is often too late.
This is the most expensive version because it feels handled. Somebody answered, they were polite, they took a name and a number. But nobody qualified anything, nobody booked anything, and the seller's motivation was highest during that exact call. A message is a record that a conversation was available and you took a rain check on it.
Which is the correct thing for a human being to do at 9pm. This is why on-call rotations quietly stop working around week six. You cannot solve a coverage problem by asking people to be permanently available, because they won't be, and you shouldn't want them to be.
This is the part the trace doesn't show you, so you have to reason about it.
Somebody calling about a property at 8pm has usually spent the day thinking about it. They are not shopping around casually. They sat down after dinner, decided tonight is the night, and picked up the phone. That's as motivated as that person will ever be, and it is the exact moment your coverage is thinnest.
They are also not calling one number. They're working down a list on their phone. The order they get answered in is the order they get to negotiate in, and whoever picks up first sets the frame for everybody after.
By 9:15 the next morning, when your callback list gets worked, that person has slept on it. The urgency that made them call is gone. Sometimes so are they.
Because the lead is still in your system, still open, and still assigned. Nothing about the record says it was mishandled. It looks identical to a lead being actively worked.
Your reporting is built on dispositions and this lead has one: contacted, following up, or something equally neutral. There is no field for the four hours it sat while the person on the other end was ready to talk.
So the failure is invisible at the report level and only visible at the individual lead level, which is why the trace exercise finds things that dashboards don't.
One lead is an anecdote. Pull five after-hours leads from the last quarter and run the same trace on each. Patterns show up fast.
You're looking for the step that comes back blank in the same place every time. That's not five problems. That's one gap, and it has a shape you can do something about.
When you find it, the fix is a question of what you're willing to put in that gap: a person on a shift, a service that qualifies rather than takes messages, or something automatic that answers on the first ring. All three cost something. The one thing that costs the most is the current arrangement, and until you run the trace it doesn't show up as a cost at all.
Run the same trace on a Saturday lead and the gap stretches. A Friday 6pm lead and a Sunday 2pm lead often get the same answer, which is Monday morning, and that's a distance measured in days rather than hours.
Weekends also carry a different kind of caller. Somebody who calls about a property on a Saturday afternoon has time, has probably discussed it with family that morning, and is often ready to have the real conversation right then. That's not a lead to call back. That's a lead to talk to.
If your trace shows a weekday gap of four hours and a weekend gap of two days, those are two different problems wearing the same label, and the weekend one is usually the bigger dollar figure even though it's fewer leads.
When you close the gap, hold whatever fills it to a real standard. Answering is the floor, not the goal.
Somebody or something picks up on the first ring, at any hour, without the caller waiting through a menu.
It finds out why they're selling, what the timeline is, what shape the property is in, and who else is on the deed.
It books a real appointment on a real calendar while the person is still on the phone, rather than promising a callback.
The record lands somewhere your team will see it before the appointment, with the motivation captured, not just a name and number.
Measure whatever you put in the gap against those four. An answering service usually clears the first and fails the rest. An on-call rotation clears all four on the nights the phone gets answered and none of them on the nights it doesn't.
Your leads are fine. Your coverage has a hole in it at exactly the hour your best sellers call, and nothing in your reporting is built to tell you.
Go pull five leads and trace them. It's an hour and it's the cheapest diagnostic in the business.
Five calls you can make in twenty minutes that tell you more than any demo. Run these before an AI agent ever speaks to a seller with your name on it.
Every follow-up process looks fine when the team is full and the volume is normal. Here is how to find the point where yours quietly stops, before a bad week finds it for you.