The 30-day playbook that took a client from 0 to 47% after-hours capture

A multi-location client was running paid ads seven days a week. The ads worked. The phone rang. But more than half of the leads came in outside of business hours, and the answering service they paid for was capturing about 12% of them as real bookings. The other 88% were a name, a phone number, and a voicemail nobody returned in time.

The owner did not need more leads. They needed the leads they already had to stop falling off the end of the day.

This is the 30-day runbook we shipped for them. It moved after-hours capture from roughly 0% (effectively, because the rare voicemail return was rarely the same day) to 47% of inbound after-hours leads booked into a real appointment, automatically, before the next morning.

It is not a generic chatbot. It is a specific workflow with a specific cutoff. Here is exactly what we did.

Week 1: Audit, not deploy

We did not turn anything on in week one. We audited.

The first job was to figure out where leads were actually coming from and what was happening to them. We pulled four weeks of data from Meta Ads, Google Ads, the website forms, and the answering service logs. We mapped every lead source to what happened to it: contacted, booked, dropped, or unknown. "Unknown" was the biggest bucket.

An owner who is not measuring this is not flying blind. They are flying with a map that was accurate two years ago and has not been updated since. We replaced the map.

The audit produced three facts that shaped the rest of the engagement:

• 62% of all leads came in between 5pm and 10am the next morning, or on weekends
• The answering service contacted 71% of those leads, but only 17% of the contacts became a booking
• The average time-to-first-contact for after-hours leads was 14 hours

Fourteen hours is the entire game. By the time the business called back at 10am the next day, the lead had already searched twice more and found a competitor that responded that night.

Week 2: The cutoff and the script

Once we had the facts, we built one thing: a single SMS-first workflow with a hard cutoff at 60 seconds.

The cutoff matters. Most marketing automation vendors will tell you they have a "fast follow-up" feature and then quietly admit it triggers in 5 to 15 minutes. That is too slow for a cold lead at 11pm who is comparing three businesses on their phone.

We built the workflow natively inside the client's CRM. The flow is simple:

1. Form submits on the landing page
2. Within 60 seconds, an SMS goes out from the business's verified number
3. The SMS opens with the lead's first name and the specific service they requested, not a generic "thanks for your interest"
4. The bot asks three qualification questions, one at a time
5. If the lead engages, the bot offers a real available slot from the business's calendar
6. The booking is confirmed via a second SMS and added to the CRM

The script for the three questions was written by the owner, not by us. That part is important. They knew what a good front desk person would ask on a first call: new or returning client, what is the main concern, when are you available. We just put it in an SMS thread instead of a phone call.

We did not have the bot try to close on price. We did not have it ask for insurance information. Those are downstream conversations that the human team handles in the morning when the lead arrives. Trying to do too much in the bot is how bots feel like bots.

Week 3: The "no answer" branch

The first version of the workflow handled the happy path. Most leads do not give you the happy path.

In week three we built the branches for the leads who do not respond to the first SMS within ten minutes. There are three of them:

• Drip back at 45 minutes with a different angle (often a question about the specific service rather than booking)
• Try once more the next morning at 9am with a soft re-open
• After two no-responses, mark the lead as cold and hand it to the office manager with a one-line summary of what we tried

The two-touch limit was a fight we had with the owner. They wanted to keep messaging. We did not. The data we have from other deployments says the third touch from a bot is where leads start to feel followed instead of helped, and the unsubscribe rate climbs. Two touches plus a human handoff is the right ratio.

The cold leads that the bot hands off are not dead. The morning summary gives the office manager a one-page list of who tried what and when. They make the human call from there. The bot does not pretend to be a person, and the person does not have to reconstruct what already happened.

Week 4: Booking, not just lead capture

The thing that separates this kind of workflow from a generic chatbot is whether the bot actually books the appointment or just collects information and hands it to a human.

A bot that collects information is not solving the after-hours problem. It is moving the problem from "we missed the call" to "we missed the callback." The business still has to call back in the morning, the lead still has to pick up, and the gap is still the gap.

The booking step was the hardest part of the build because it required real, live calendar availability. We connected the workflow to the client's existing scheduling tool through the CRM's calendar integration. The bot reads the real availability and offers two specific slots, not "let us know what works for you." Two slots, pick one, confirmed.

We also added one rule that mattered more than any of the conversational design: the bot never offers a slot less than 18 hours out. Same-day bookings sound good in a demo but in practice they create churn, because the lead is more likely to no-show on something they booked at midnight for the next morning. Eighteen hours is enough gap for the booking to feel intentional, and it leaves the same-day slots for the front desk to fill from the wait list.

What it actually produced

Thirty days after launch, with no other channel changes:

• 47% of after-hours leads booked into a real appointment automatically
• Average time-to-first-contact dropped from 14 hours to 47 seconds
• The office manager spent roughly four hours less per week chasing voicemails
• No-show rate on bot-booked appointments was within a point of front-desk-booked appointments

The 47% number is not a ceiling. It is what the workflow produced in the first 30 days with no further tuning. We have since added a fourth qualification question (insurance verification) and a different script for second-time inbound leads, and the number has moved up. But the headline result, the thing that mattered to the owner, came in the first month from a workflow that took less than three weeks to build.

What we would skip next time

Two things.

First, we spent the first three days of week one trying to instrument the answering service to give us better data. The answering service does not have an API that anyone would call modern. We should have just pulled the data we needed from the call recordings and the CRM and skipped the integration attempt. We lost about two days that way.

Second, we built a long welcome script for the first SMS that explained who the bot was. The owner pushed us to shorten it. They were right. The lead does not care whether they are talking to a bot or a person at 11pm. They care whether they can book an appointment. The current version is two lines. Shorter is better.

When this kind of workflow does and does not work

This works when the business has real, measurable lead volume coming in outside of business hours and a manual process for handling those leads that is leaking. If your after-hours volume is five leads a month, you do not need a workflow. You need a phone.

It does not work when the business does not have a reliable booking system. The bot cannot book what the calendar does not show. Businesses on paper appointment books, or with three different schedulers who do not share state, are not ready for this kind of automation yet. The first job in that case is the calendar, not the bot.

It also does not work as a one-time deploy. The script needs tuning every few weeks based on what leads say back. The owner does not have to do that themselves, but somebody has to. That is part of why we run this as an ongoing engagement and not a one-time build.

If you are running a service business that is leaking after-hours leads, this is the kind of work we do. The whole playbook took 30 days and one technical partner. It did not need five vendors.