Case study

From 89 to 163 booked appointments a month

A franchise consulting firm · GoHighLevel + Retell AI · client anonymized

A franchise consulting firm was paying for leads from ads, website forms and third-party feeds. The leads were coming in, but too many went cold. We rebuilt what happens between the inquiry and the booked meeting.

Consultant network leads, before and after. Other things changed too. This is one firm, not a franchisor. We compare monthly averages, and we can't credit the AI alone.

The results

Monthly average	Before	After	Change
Booked appointments	89	163	+83%
Show rate	32%	41%	+9 points

Before is June to December 2024, the CRM's full history before the AI. After is January 2025 to August 2026.

What the AI did on the phone

From January 2025 to September 2026, nobody on the team dialed these calls.

Calls placed by the AI	5,379
Unique leads worked	1,733
Calls that reached a live person	1,997 (37%)
Calls that reached voicemail	2,169 (40%)
Different people reached live	1,214
Time in live conversation	48.5 hours
Live calls where someone asked for a callback	355
Live calls that ended with a specific appointment time	443

Those last two rows count calls, not people. One person can appear more than once.

What we built

All sources, one CRM

Facebook ads, 18 website forms, email, webhooks and third-party feeds all land in GoHighLevel. Why this has to come first.

A test contact in GoHighLevel with a source tag and an Available Capital field filled in. — A test contact in the CRM. Qualification answers land in fields like this one.

Speed to lead

A new lead triggers an AI voice agent that holds a real conversation, qualifies the candidate and writes the answers back to the CRM. What the speed data really says.

Calling platform analytics showing 1,706 calls, a 42 second average duration, 1,321 millisecond latency and 1,125 successful calls. — AI calling analytics from the voice platform, January to August 2026. One card hidden.

Follow-up system

Callbacks and appointment reminders run on their own, so a busy rep doesn't mean a dropped candidate. How the follow-up works.

A GoHighLevel workflow called AI Follow Up Call, with branches that update contact fields, wait and move the lead on. — The AI follow-up call workflow inside GoHighLevel.

Full-data dashboard

Automated workflows pull leads, appointments, referrals and sources into one dashboard, broken down by consultant. The numbers we track.

Franchise analytics dashboard showing total leads, appointments, conversion rate and confirmed meetings for the last 30 days, with the top rep for each measure. — The live dashboard for one firm, last 30 days. Client name and rep names hidden.

An n8n workflow on a schedule that pulls appointments per consultant from the CRM and writes them to a database. — One of the scheduled workflows behind the dashboard. It counts appointments per consultant. Names hidden.

AI RevOps Engine

The newest piece reads the CRM, calendars, ads and calls, calculates every metric in code and answers plain-English questions. What it finds.

AI RevOps Engine summary titled The revenue you can't see, listing qualified buyers with no meeting, no-show rate, cost by lead source and unread messages. — What the engine pulled from one firm's CRM. Captured on a different date from the other engine screens, so counts differ. Engine output, not an audited figure.

AI RevOps Engine answering a request for everything on one rep, covering appointments, show rate, conversion, pipeline and new leads. — Asking the engine about one rep. Names replaced by the engine's anonymous labels. Engine output, not an audited figure.

What we don't claim

We don't claim every lead got called. The AI worked 1,733 unique leads, while roughly 750 came in each month.
We don't compare lead volume or conversion rate. The 2024 numbers include an initial CRM import, which would flatter the result.
We don't publish an average speed. We haven't finished matching every first dial to the time the lead was created.
We don't credit the AI alone. Other things changed over those 20 months too.
We don't quote deals or earnings. Appointments and show rate are what we measured.

How the numbers were pulled

Call figures come from the live calling platform record, January 6, 2025 to September 26, 2026. Before and after figures come from the CRM, across about 24,000 contacts and every calendar event, as per-month averages.