How Glamnetic Eliminates the Hidden Cost of Manual Order Modifications
TL;DR
- Glamnetic's most costly tickets were the simple ones with a deadline: change my address, cancel my order, fix what's in my order.
- They gave their AI agent, Gina, the ability to act on orders rather than just reply, integrating with Shopify and their 3PL, ShipMonk.
- Every Action that touches an order requires the customer to confirm before Gina modifies anything in Shopify. This keeps Glamnetic in control.
- On AI Agent 3.0, with no change to the actions or the policies, the successful completion rate roughly doubled across all three intents.
- AI Agent actions didn’t hurt satisfaction. CSAT on these intents holds at 4.8 to 5.0.
About the brand
Since 2019, Glamnetic has been changing the beauty game with innovative, easy-to-apply magnetic lashes, and has grown into North America's number one premium press-on nail brand.
Its CX team has evolved from customer service into a 360-degree customer experience organization, led by Mia Chapa with Jackie Briones. They call themselves a tiny but mighty team, and they run an order volume that would normally require several times the headcount.
The Challenge: Resolving time-sensitive order modification requests before they're fulfilled
Glamnetic faced a large volume of urgent tickets related to order modifications on email and chat: shipping address changes, item edits, or cancellations. Every ecommerce team deals with these. Someone moved. Someone shipped to their office and then took the week off. Someone autofilled an address they last used two years ago. Someone placed an order at midnight and regretted it by morning.
On paper, these are the easiest tickets in the queue. Operationally, they are the most expensive, because the fix has a deadline that has nothing to do with the support team's response time. Once the fulfillment request reaches the 3PL, like ShipMonk, the address is locked, and the outcome is fixed. The parcel goes to the wrong place, and Glamnetic pays to send a second one.
Peak season made it worse. During Black Friday and Cyber Monday, address changes arrived faster than a human team could work them.
"During Black Friday Cyber Monday, order edit requests really impacted our business. Customers were urgently requesting to update their shipping address. We had a limited window to do this and a high volume to get through.”
—Jackie Briones, Customer Experience Manager at Glamnetic
Additionally, order edits used to be a leading source of low CSAT scores because customers would be worried that the order would ship before the team could fix it. Customer confidence wavered if Glamnetic took a long time to respond.
“If it's an address change, we would have to reship a package to the correct address because it's not the customer's fault that we were unavailable,” Mia explains. “Now, with Actions, it's instantaneous.”
The Solution: Implementing Actions to handle shipping address changes, cancellations, and order edits as soon as the request comes in
Prior to AI Agent Actions, if customers reached out about a mistake they made with their order (like ordering the wrong item, entering the wrong address, or even accidentally placing the order), the team might not have caught it before the fulfillment request was sent to the warehouse. This meant the warehouse had already shipped the order, which meant the business had to pay for a return or re-ship the item. It was a costly problem.
Reducing customer anxiety with instant address changes
One of the first AI Actions Glamentic introduced was editing shipping addresses. When a customer realizes they’ve entered an incorrect address, they reach out with a sense of urgency and panic. Responding to them and resolving the issue with the same level of urgency is key.
This Action involves multiple steps: Gina, Glamnetic’s AI Agent, collects the required fields before acting, detects merged orders, checks whether the order is already fulfilled, confirms the new address with the customer, and only then writes the change into Shopify.
Across email and chat, order modifications made up 11% of Glamnetic’s post-purchase support tickets. Because of the actions Glamnetic has set up, 41% of these requests are auto-resolved without reaching a human. Three months ago, only 19% were getting auto-resolved with the older model of Gorgias AI. Now AI is seamlessly executing address changes. But Gina isn’t acting without any guardrails; these address changes need to be validated by the customer which adds a level of control.
Canceling orders before it’s too late
Cancellations are one of the most costly requests to miss in real time. An unwanted parcel ships, comes back as a return, and gets refunded anyway, meaning Glamnetic has to incur shipping and return costs.
Glamnetic maintains control by once again ensuring customer confirmation is required before Gina cancels any order in Shopify. Cancellation is configured as a multi-step action: Gina proposes it, stops, and waits for the shopper to confirm before anything is touched.
"Sometimes orders are placed by mistake, or there's that remorse. When they come in, they ask for a cancellation, boom, Gina will take care of that for you."
—Mia Chapa, VP of CX and Strategic Initiatives at Glamnetic
AI Agent 3.0 and Skills: the new framework that makes this possible
Glamnetic spent more than a year building the raw material. Knowledge for Gina, Actions wired into Shopify and ShipMonk, policies for what she should and should not do.
Before Skills, a customer wrote in, and AI Agent searched across everything it knew, picked what looked relevant, and made its best guess. For an order edit, a best guess isn’t good enough. Gina had to know which fields she must collect first, which order states are still eligible, and which Action to fire.
Skills changed the shape of that. Gorgias ships 55 intent categories and classifies each incoming message into one at 95% accuracy. A Skill is tied to an intent, which means that there is a specialized instruction for every topic. This means each one is handled with precision and expertise. When the message is classified, that Skill is injected straight into AI Agent’s context instead of being retrieved through a general knowledge search, and it outranks everything else the agent knows.
Glamnetic's address-change Skill shows exactly what Gina must do before acting:
- The exact fields Gina must collect before she is allowed to act
- Merged-order detection, so two orders combined into one shipment are handled correctly
- The boundaries: which order states are still editable, and what happens when they are not
- The routing through ShipMonk, their 3PL
- The Action itself, linked inside the Skill
"We have specific required fields that we want to ensure are followed before Gina acts. Along with general requirements and boundaries," explains Jackie Briones. "Our actions are actually linked within our Skills to ensure the same process gets followed every time."
This made the biggest difference for Gina’s success rate with order edits. The same Actions, integrations, and confirmation steps now follow more specific, intent-based instructions.
How to replicate Glamnetic’s strategy. Download their 6-step framework with templates, AI prompts, and timelines.
The Results: Doubling successful order modifications without dropping CSAT
By optimizing their intent-based Skills, Guidance, and multi-step Actions, Glamnetic saw significant improvements in some of the most potentially costly ticket requests: order edits, shopping address changes, and cancellations.
In the past 3 months on AI Agent 3.0, Glamnetic achieved:
- 2.2x more shipping addresses edited in time by AI Agent
- 1.8x more orders canceled before reaching fulfillment
- 5.3x more order item modifications successfully executed, reducing unnecessary returns
Throughout this, Glamnetic maintained industry-leading CSAT scores on these intents:
- 4.88/5 on address changes
- 4.80/5 on cancellations
- 4.98/5 on order edits, resulting in 0 detractors year to date
By enabling Gina to act, not just respond, Glamnetic cut the number of order modification conversations that need a human by 44%. Before AI Agent 3.0, nearly 7 in 10 of these conversations ended with a human stepping in. Today it is under 4 in 10.
- 52% fewer escalations on order cancellations
- 47% fewer on shipping address changes
- 26% fewer on order item modifications
Over the same period, the share of these tickets Gina resolved and closed on her own nearly doubled, from 27% to 48%.
Next steps: Empowering Gina to be a personal shopper
Glamnetic is constantly experimenting with new tools, strategies, and features to push the boundary on what’s possible with its customer experience. The team recently ran an A/B test with a 50/50 traffic split on their product pages, embedding common AI FAQs on product pages to spark more conversations. Automation and take rate climbed sharply and handover fell.
Past that, Mia and Jackie want the same thing: a unified customer profile that turns Gina from a support agent into a personal shopper who already knows what you bought last time.
Looking to take order edits, shipping address changes, and cancellations off your team’s plate? Talk to our team today.
.png)
.png)
