Thrasio saves $260K with auto QA and CSATai

Thrasio is the next-generation consumer goods company reimagining how the world’s most-loved online marketplace products become accessible to everyone. Since 2018, it has grown more than 200 brands and estimates that one in six U.S. households has purchased a Thrasio product.

Thrasio has invested heavily in its customer experience. The customer obsession team provides 24/7 support and resolves over 70,000 tickets per month, earning Thrasio a CSAT score in the top 10% of all industries.

Problem

With thousands of phone calls and other customer interactions coming in every week, Thrasio’s team of CX analysts was only able to manually audit about 3% of interactions–and they estimated it would have taken a team of 528 analysts to manually review 100%. 

In addition to being time-consuming, the manual QA process was subjective. Analysts completed a QA checklist for every interaction they reviewed, but different analysts could interpret the same phrases or behaviors in different ways. This made it difficult to provide unbiased coaching to its customer obsession representatives.

Solution

Thrasio is using Capacity to uncover product issues and insights based on customer service phone calls, chat conversations, emails, Amazon reviews, and negative customer experience data sources.

With Capacity, Thrasio has configured 1,000 categories–collections of phrases that describe a specific concept or behavior–to track issues across all tickets and brands, allowing team members assigned to each brand to identify and address new issues quickly.

According to Erika Tagami-Manahan, Global Services, Customer Experience Lead at Thrasio, it was a lengthy manual process to identify a new product issue before Capacity. “Now, we are able to streamline our process by auditing only around 200-300 tickets to identify issues,” she says. “And then, once we build new categories to track those issues in Capacity, quantifying them becomes easier and more efficient.”

Thanks to Capacity, the Global Services team can efficiently uncover and address product issues without incurring unnecessary, labor-intense processes–ultimately ensuring they are delivering positive brand experiences to their customers.

Free Resource

Your Post-Interaction QA
Checklist Is Here

Score every customer interaction consistently with this ready-to-use QA checklist — and see how Capacity's Auto QA can do it automatically, at scale.

Get the Free Checklist

Results

Since implementing Capacity, Thrasio has:

• Saved $260K in annual cost-savings with higher QA efficiency
• Raised QA rates from 3% to 100% of all customer interactions
• Achieved a 97% CSAT score

Being able to audit 100% of interactions and surface insights from them allowed Thrasio’s services team to improve the agent coaching experience by eliminating bias–agents receive data-backed feedback based on all their interactions. 

Thrasio has also dramatically decreased the time its CX analyst team spends manually auditing interactions. “Instead of spending time listening to calls, reading emails, or evaluating each interaction, we’re able to use our CX dashboard,” says Joyce Gimeno, CX Analyst at Thrasio. “It makes our analysis more efficient and allows us to focus on improving the most important areas of our customer experience.”

The CX analyst team is now saving approximately $260,000 per year, thanks to its increased efficiency with Capacity.

Increasing customer satisfaction with predictive insights

Thrasio is also taking an innovative approach to using CSATai, Capacity’s proprietary predictive CSAT model. CSATai analyzes the language used in every customer ticket and provides a CSAT score based on the likelihood the customer would say they are satisfied or dissatisfied. Thrasio has developed an automated process for acting on CSATai scores:

1. Thrasio uses Cal, its generative AI application, to create personalized responses to customer service tickets. Cal also provides a confidence level rating (0 to 100%) for each response.    

2. Human customer service agents review the tickets when the confidence rating is below a specific threshold and modify, enhance, or rewrite the responses as needed.

3. Once the agent or Cal sends the response to the customer, the ticket is marked as solved in Zendesk.

4. Capacity automatically evaluates the full history of the solved tickets by analyzing the words and phrases the agent (human or Cal) and customer used. Using an AI model trained to predict customer satisfaction scores for customer service conversations, Capacity determines the probability of a positive (CSAT) or a negative (DSAT) customer survey reply and sentiment score.  

5. Capacity appends the predicted CSAT or DSAT score to the customer service ticket in Zendesk.  

6. Tickets with CSAT probabilities remain closed. If there is a probability of a negative response (DSAT), the ticket will be reopened, and a human agent will be notified.  

7. Tickets with DSAT probabilities are either returned to the human agent who handled the ticket or the next available human agent, if Cal replied. The human agent determines if they need to reach out to the customer directly to address the issue. This improves the customer’s experience and reduces the likelihood that they will submit a negative survey response.

“We call this [process] ‘checking your homework,'” says Gershwin Exeter, Vice President of Global Services. “It allows an associate to look at a ticket and say, ‘Maybe I should respond to this differently.’ Having that predictive score and an extra layer of review has helped us increase our CSAT from 89 to 97%.”

See Capacity in Action

Ready to See What
Capacity Can Do?

Get a personalized demo and see how Capacity's AI platform can streamline your support, reduce costs, and delight your customers.

Request a Demo