Key results
| Metric | With Yuma |
|---|---|
| Tickets fully automated, end to end | 50% week to week (16% at launch) |
| Including tickets Yuma partially automates | 78% (October 2026) |
| Reply time | <2 min |
| CSAT on tickets Yuma handles | 4.3 to 4.5 |
| Work done by Yuma | 13 full-time agents' worth |
| CX headcount | Flat this year, with no seasonal hiring wave for the holidays |
Full automation: tickets Yuma resolved with no person stepping in, as a share of the support tickets Yuma worked on. Partial automation: Yuma handles part of the ticket, such as collecting the details, before handing it off to an agent.
Introduction
Freebird is a direct-to-consumer grooming brand. It makes FlexSeries shavers for head, face and body, plus a women's line, and most customers stay with Freebird through a blade refill subscription.
Naomi Oriol is Director of Customer Experience at Freebird. She joined in August 2025 and owns the full CX loop: reporting, agent quality, fulfillment escalations and the cross-functional work that touches the customer.
Her team handles about 10,000 tickets a week, and it swings with the season. Subscriptions are the biggest driver, at around 40% of volume: address updates, frequency changes, product swaps and add-ons, and cancellations. After that come order questions, returns, product questions, warranty claims and shipping issues.
“I needed a way to absorb the growth, and the inevitable surprises that come with customer support, without the org chart growing every time our queue did.”
Naomi OriolDirector of Customer Experience, FreebirdThe problem
Before Yuma, every ticket was handled by a person. Freebird ran on Gorgias rules, macros and a few basic automations, plus a team of agents and team leads sorting and cleaning the queue by hand.
The model was simple: when volume went up, Freebird added people. Naomi saw it break in two places.
First, much of the queue was low complexity. Skilled agents spent their days on copy-paste work.
“A lot of the queue is low complexity, so it felt like we were paying really smart and skilled agents to do copy-paste work.”
Naomi OriolDirector of Customer Experience, FreebirdSecond, the peaks were brutal. Black Friday and Cyber Monday volume triples, and the subscription, order and delivery questions all land at once. Hiring and training a team for a spike like that, then carrying the headcount through the rest of the year, didn't add up.
The roadmap made it worse. Every big plan on it meant more volume and more complexity for CX. Hiring to match meant more agents, more team leads, another manager, and more onboarding and ramp time.
What Freebird needed from AI
Naomi had three requirements before she'd trust an AI with Freebird's customers.
Follow Freebird's actual SOPs, in Freebird's voice. Not a generic version of good service. Customers are buying personal grooming products, so tone matters. The AI also had to read the nuance: someone who shaves their head versus someone buying a gift for their daughter, a loyal subscriber versus a first-time buyer.
Own the resolution end to end. Answering the customer isn't enough. The AI had to make every change in the back end, too.
Prove the back-end work is real. This was Naomi's biggest doubt going in. A single address change on a subscription can touch three systems.
“When you update a subscription, you have to change the address maybe in the subscription portal, but maybe you also have to change an order in our e-commerce platform. Maybe you have to put that order on hold before you change the address with our 3PL. So I was like: is it going to be able to do it, or is it just going to tell the customer that it did it but not actually do anything on the back end?”
Naomi OriolDirector of Customer Experience, FreebirdYuma runs those steps in the tools themselves: the subscription portal, the store and the 3PL. Freebird's Handbook says which steps apply and in what order.
A slow rollout, by design
Freebird started with low-complexity, one-touch tickets and put Yuma live on 10% of them. The team QA'd how Yuma resolved each one, then raised the share step by step.
As Yuma learned how Freebird handles things and the Handbook got more complete, the percentage went up. When a process didn't run the way the team expected, they pulled it back, reworked the process and the Handbook, and started again.
“I was totally okay with a slow rollout, because I wanted to get it right. It was important for me that even if customers knew they were talking to a bot, they still felt they were getting the same level of service and the same correctness of resolution they would get if they were talking to one of our agents.”
Naomi OriolDirector of Customer Experience, FreebirdOutcomes
- 01
50% of tickets fully automated, end to end
Freebird started at 16% full automation. Week to week, Yuma now automates 50% of tickets end to end, with every back-end change made and no person stepping in. That's 78% if you count the tickets Yuma partially automates before handing them off to an agent (October 2026).
Even when Yuma hands a ticket to a person, it has usually done the back and forth with the customer and collected what the agent needs.
- 02
Replies in under 2 minutes
Yuma's first reply often lands neck and neck with Freebird's autoresponder, which fires the moment a ticket arrives. Reply time is under 2 minutes, around the clock.
- 03
13 agents' worth of work
Freebird holds its agents to about 90 tickets a day. Yuma does the work of 13 of them, and it works nights, weekends and holidays.
“That's 50% full automation, the end-to-end I was talking about. Even if Yuma can't complete the ticket, it's still able to do a lot of back and forth with the customer to at least collect the information needed, and that saves time for the agent.”
Naomi OriolDirector of Customer Experience, FreebirdA staffing lever you can flex
What surprised Naomi most is how Freebird ended up using Yuma. When volume dropped, the team didn't cut shifts. It turned automation down and banked it for when the queue spikes. Naomi flexes it up and down by channel and contact reason, following how the queue actually moves.
“What surprised me is that Yuma became a staffing lever, not just a deflection tool. When volume dropped, we didn't cut shifts; we were able to turn down our automations. That's how we keep the team stable.”
Naomi OriolDirector of Customer Experience, FreebirdThe other surprise: it worked well enough that Freebird used up its package early. In Naomi's words, a good problem to have.
Peak season is the real test
Freebird's first Black Friday and Cyber Monday with Yuma came early in the rollout, at around 20% automation. It helped, but the real peak came later.
For Freebird, the biggest wave isn't Christmas. It's January, when gift recipients have used their shaver and write in with questions, or decide whether to keep it or return it. That was the first true test for Yuma, and the one where Freebird leaned on it most.
This year, the team has spent months on QA: tightening processes, capturing edge cases in the Handbook and telling Yuma what to do when this or that happens. Naomi expects Yuma to fully automate 50 to 65% of tickets through the holidays. Freebird is going into the season without a wave of seasonal hires.
“I feel good about Yuma handling probably 50 to 65% completely, end to end. And because we're predicting success here, we didn't need to bring on a wave of agents to get us through this peak.”
Naomi OriolDirector of Customer Experience, FreebirdWhat the team does now
Freebird's agents now spend their time on tickets that need judgment: warranty replacements, returns, escalations from very upset customers and anything involving money.
They're also part of Yuma's QA. When an agent spots something Yuma got wrong, they can give it feedback directly, and Yuma captures the new instruction in the Handbook.
And the team has time for work the queue used to crowd out: surprise-and-delight projects, CX side projects and, recently, helping the creative team moderate social media comments.
“We're really playing into their skill set, allowing them to do what they do best: thinking, using their judgment, using their best practices.”
Naomi OriolDirector of Customer Experience, FreebirdCustomers care about speed and a real fix
CSAT on tickets Yuma handles sits between 4.3 and 4.5, close to Freebird's human agents. That's despite Yuma taking many of the toughest contacts. Cancellations go straight to Yuma, and they're the hardest tickets for satisfaction.
Some customers can tell they're talking to AI. Most don't care, as long as the answer is fast and the problem is solved. One customer wrote after a warranty ticket:
“I was worried it would be difficult and time-consuming to deal with Freebird if I couldn't speak to a person. To my surprise, they responded promptly, located our account and sent a replacement razor.”
When a ticket Yuma handled gets a low score, Freebird looks closely. It's never "you're a bot". It's usually a process gap. The team feeds that back to Yuma, and the fix goes into the Handbook.
Freebird's advice for brands weighing AI for support
We asked Naomi what she'd tell another Shopify brand considering AI for support.
Pick a partner that gives it to you straight
“I wouldn't trust anyone who said, "We're going to automate 100% of your whatever, right out of the box." It should be something slow. And you should expect transparency from your partner. There are going to be bumps along the way, but that's just part of the process.”
Naomi OriolDirector of Customer Experience, FreebirdLook for a vendor that helps you get set up and also brings advice from brands like yours.
Close the gaps in your SOPs first
If your SOP has a hole, the AI will have the same hole. Make the documentation clear, including how tickets get tagged and closed, so your reporting stays clean.
QA it like you'd coach an agent
You wouldn't let agents work without QA and coaching, and the same goes for AI. Review low CSAT scores and feed what you learn back in.
“Don't think this is something you just set and forget. I'm going to QA a handful of tickets, I'm going to improve a process by 2%. Because that one change can mean it does 100 tickets better.”
Naomi OriolDirector of Customer Experience, FreebirdConclusion
Freebird went from an all-human queue to Yuma automating 50% of tickets end to end. Replies land in under 2 minutes, CSAT stays between 4.3 and 4.5, and Yuma does the work of 13 full-time agents.
The bigger change is how Freebird plans. Headcount stayed flat this year, automation flexes with the queue, and the team is going into the holidays without a hiring wave.
For subscription brands whose tickets touch the store, the subscription portal and the 3PL at once, Freebird shows what happens when the AI owns the whole resolution. To see how it works, visit Yuma Support AI, or book a demo and we'll show you what your queue could look like by January.







