How Healthletic automated 40% of its support with AI
Dietary supplements · Customer case study
Healthletic is a dietary supplement brand that handles thousands of customer conversations each month.
The brand configured Chelsea, its AI support agent in Commslayer, to help customers using information from Shopify and its subscription platform. Chelsea follows Healthletic’s support policies, handles routine requests such as order status and tracking look-ups, shipping and customs questions, and subscription changes, and passes conversations that need human judgment to the team.
In 30 days, Healthletic reported:
-
40%
support automation
(the share of resolved conversations closed by AI with no human reply)
-
56%
AI resolution rate
(the share of the AI agent’s conversations resolved without a handover)
-
3,271
customer conversations resolved by AI
-
67%
lower median resolution time
than in the 30 days before the AI agent went live (5 hours 30 minutes to 1 hour 50 minutes)
Supporting thousands of customers each month
Healthletic receives thousands of customer conversations every month, most of them about order status, delivery and subscriptions.
At that volume, even routine requests represent a substantial amount of work. Healthletic used Commslayer to bring AI into its support operation, helping customers with everyday requests at any hour, including the gaps between its day and night shifts, while keeping its team involved in more complex cases.
“We were running support out of a shared inbox with a growing team across two shifts. We needed one place where agents, order data and subscriptions lived together, and a way to take the repetitive ‘where is my order’ questions off the team’s plate without lowering the quality of the answer.”
Marko Jankovic, Customer Experience Lead, Healthletic
Building support around Healthletic’s policies
Healthletic connected Commslayer to Shopify and its subscription platform and configured Chelsea around the way its team supports customers.
Chelsea can use relevant customer information to answer questions and complete supported requests within the conversation. When a request falls outside those policies or needs a person’s judgment, she hands it over to the support team.
The support team was not reduced. Chelsea absorbed routine volume so agents could focus on the cases that need a person: reships, damaged orders, customs holds and complaints.
Increasing automation and reducing escalations
During the reporting period, Chelsea took part in roughly three in four conversations and resolved 3,271 of them without a handover.
The AI agent resolved 56% of the conversations it handled without a handover. Compared with the preceding 30 days, that resolution rate rose from 53%, and the share of AI conversations handed over to a human fell by more than five points.
Before the AI agent went live, all replies came from the team. In the reporting period, 40% of resolved conversations that received a reply were closed by the AI agent with no human reply, up from 36% in the preceding 30 days.
These improvements meant a larger share of support was handled through automation, with a smaller share of conversations requiring escalation.
Cutting resolution time
Healthletic also recorded improvements in support speed.
In the 30 days before the AI agent went live, Healthletic’s median resolution time was 5 hours 30 minutes. In the reporting period it was 1 hour 50 minutes, a 67% reduction on comparable volume.
The AI agent’s median first reply took 3 minutes 7 seconds.
“The obvious change is speed: a tracking question at 3 a.m. gets a correct answer at 3 a.m. The less obvious one is that the team now spends its time on the cases that need a person — a damaged shipment, a package stuck at customs, a customer who is upset. Those are the conversations that decide whether someone orders again.”
Marko Jankovic, Customer Experience Lead, Healthletic
