Courtesy Auto turns missed calls into booked appointments across seven stores
Case Study
Customer snapshot
Samantha Williams, BDC Director at Courtesy Auto, runs the group's service BDC, answering thousands of calls a day across seven stores.
Before Numa, missed-call recovery was manual: Samantha worked the missed-call queue herself every couple of hours, and a list of roughly 20 missed calls typically converted into only one or two appointments.
Numa now coaches missed callers to expect a callback, so they actually answer when the store calls back, and tags every failed appointment with a reason, giving Samantha a list to work for a win-back later instead of losing the opportunity for good.
After a recent update let callers ask for a human at any point in the conversation, AI-related complaints dropped to what Samantha calls "next to none," a change she says happened almost overnight.
Across 13 recent weeks at Courtesy Chevrolet, the group's GM store, Numa's reporting shows a 79.2% caller rescue rate, a 35.3% booking rate on appointment-agent conversations (663 booked out of 1,879), and a median reply time to customer messages of about 7 minutes.
Samantha Williams wears a lot of hats as BDC Director at Courtesy Auto, but the ones that matter most day to day are monitoring heat cases across all seven stores, tracking handle rates and call volume, coaching agents, and personally stepping in when an appointment stalls out to find a workaround that works for the customer and the store. Her team answers thousands of calls a day.
One or Two Appointments Out of Twenty
The frustrating part, before Numa, wasn't that the process didn't work. It was that it didn't work well enough. Samantha's team already had a high answer rate and a disciplined manual process: she ran the missed-call queue herself every couple of hours so that every miss got a callback.
"We had a pretty high answer rate to begin with, but we weren't really able to capitalize a whole lot on those. We would get maybe one appointment, two, out of a list of maybe 20 missed calls from across the dealerships."
By the time her team called back, customers often hadn't picked up, or had already gone somewhere else. "That was disheartening," she said. "If we could've been quicker, if we could've been available." Missed calls carry a real, measurable cost across the industry, and Samantha's team was already doing everything a manual process could do to recover them. Update calls added to the drag on top of that. With her BDC focused on setting and showing appointments rather than fielding status checks, a repeat callback about a car already in the shop was pure overhead. Her agents' refrain, in her words: "I'm not making any money off of this call. I want to hurry up and get it over with."
"Oh My God, It's Gonna Take My Job"
Samantha is direct about where the hesitation started, including with herself.
"Honestly, I think the biggest one, and it's a very human one that comes right off the bat, is: oh my god, it's gonna take my job. What happens to me if I don't have agents to manage?"
She held what she calls come-to-Jesus meetings with her team to work through it. Her message was simple: the tool was there to help the people already doing the job, not replace them. "This is something that helps the humans that are already there," she told her agents. "Makes us better, makes us more efficient."
Customer Pushback That Mostly Didn't Show Up
Her second worry was that AI on the phones would become its own heat case, "a self-inflicted one," as she put it. Early on, she did get some complaints. When she investigated, she found the phone system working exactly as designed, no dropped calls, no strange behavior. What she'd actually run into was customers venting about AI in general, not anything Numa was doing wrong.
Those complaints thinned out further, and then nearly disappeared, after a recent update let callers ask for a human at any point in the conversation.
"That has made all of those complaints, even about the AI, next to none. That pretty much happened overnight, and I was like: that's it, that's the last tweak we needed."
Coached Callbacks, Labeled Failures, One Place to Look
Three changes reshaped how the BDC runs day to day.
First, missed callers now hear exactly what happens next instead of a generic voicemail prompt.
"Tell me exactly what you want. We are going to be calling you back. Make sure to pick up the phone when we do."
Customers know the callback is coming, so they answer it. "We've been able to save a lot of that."
Second, a feature Samantha wasn't shopping for became one of her favorites. "Oh yeah, and we can add labels," is how it came up in her original Numa demo. "I'm like, what do you mean, add labels? Tell me more."
"Every failed appointment now carries a reason: price, availability, the customer decided to handle it themselves. Then I can pull a list of all of those people and try to win them back at a later date."
Third, tracing a customer interaction stopped requiring a trip through multiple systems. Which number a customer called, which line, who actually sent a given text, all of it used to mean checking several places to piece together the full picture.
"All of those things, they happen in one spot now. That was a nice happy accident that we got out of Numa."
The underlying problem was never really about the phones specifically, it was that a customer's history was scattered across systems that didn't talk to each other. Numa closing that gap by accident is exactly what happens when communication and operations run on one record instead of several.
The Numbers at Courtesy Chevrolet
Samantha's day-to-day account is about the whole seven-store BDC, but the group also has hard numbers behind one store's service operation specifically: Courtesy Chevrolet, the group's GM rooftop.
Across the 13 complete weeks ending in early August 2026, Numa's reporting shows a 79.2% caller rescue rate at Courtesy Chevrolet, meaning roughly four out of five inbound calls ended in a scheduled callback, a text conversation, a live transfer, or a booked appointment rather than a hang-up. The appointment agent engaged in 1,879 booking conversations and converted 663 of them into appointments, a 35.3% booking rate. When customers message the store directly, the median time to a human reply is about 7 minutes.
Group-wide, Samantha describes the same shift in plainer terms: the update calls and repeat callbacks that used to eat her team's day are, in her words, "definitely minimal for my department at this point," freeing the BDC to work outbound campaigns faster instead of constantly reacting to inbound status checks.
"You Can't Say No to That"
Asked how she'd explain Numa to another BDC director, Samantha skips the adjectives and hands over a measurement plan instead: compare your team's conversion rate on its own against the conversion rate with Numa and your people working together, and compare lead handle rate, meaning calls actually answered, against CSI scores, before and after.
"I think I've tried pretty much every manual process in the book to get the numbers up to where they are. It only gets better from there. You can't say no to that. It's impossible to say no to that."
