Going from great to top 1% with AI automation

Following the recent launch of AI-powered support resolution automations for Assist, we sat down with Gershwin Exeter, Chief Experience Officer at Thrasio, to uncover what it took for his team to successfully implement advanced AI. Spoiler alert: it’s a whole lot more human than you might expect!

If you can measure it, you can improve it

Like most CX success stories, this one starts with data. Before even thinking about implementing AI, Gershwin and his team zeroed in on the aspects of Thrasio’s support operations that needed measuring and improving:

• CSAT
• First reply time
• Full resolution time
• Employee satisfaction
• Total cost of ownership for support ops software

Having clear insights into key support metrics let the team know exactly how they were performing and where they could make the most high-impact improvements. Agents and their team leads focused on boosting CSAT, reducing first reply time, and speeding up full resolution time. Meanwhile, Gershwin and his leadership team prioritized maintaining high employee satisfaction and reducing the total cost of ownership for their tech stack.

With this baseline view into support ops performance established, the team was in a great position to add AI to the mix. And the results have been remarkable! After integrating Assist automations into their support processes, Thrasio saw their first reply time drop from 69 minutes to one minute and full resolution time speed up from hours to minutes.

AI so good you might want to buy it a beer

Thanks to the power of generative AI, automated responses by Assist achieved a quality so high that they’re indistinguishable from human responses. Automated responses can handle tricky situations like returns and warranties by responding to the tone and content of customer communications. Automations can even offer discounts, just like a human agent. In fact, Assist’s responses were so helpful and accurate that one delighted Thrasio customer offered to buy the bot a beer.

A six-step recipe for AI success

Gershwin maintains that technology is only one-sixth of the solution. Here’s his six-step recipe for successfully implementing AI:

1. Strategy: Have a clearly articulated strategy to drive your decision-making.
2. Organization: Ensure the team is along for the ride during your AI transformation.
3. Processes: Be willing to rethink and rework your processes to fit with your updated approach.
4. Insights: Use data to measure, review, and adjust to what you learn.
5. Technology: Choose technology provided by companies that share your values and working style.
6. Culture: Align your actions with company values during the AI transformation

Where customer satisfaction meets employee satisfaction

One reason Assembled Assist has been the perfect match for Thrasio is Gershwin’s belief in engaging — not deflecting — customers. Many AI-powered tools focus on deflecting customers to self-service resources, which may or may not actually resolve their issues. Instead of preventing customers from reaching a person who can solve their problem, Assist provides customers with resolutions. At the end of the day, great support is about providing the right answer at the right time — whether it’s from a human agent or an automation.

For Gershwin and his team, onboarding Assist took just 45 days from start to full automation. Behind the scenes, he reoriented the team to ensure they were set up for success. This meant providing 26 hours of training so everyone was fully aware of how AI would change their function, as well as rewriting 337 processes. It was a full-on AI transformation, made possible by the strength of Thrasio’s existing support operations foundation.

Through it all, the team thrived. Employee satisfaction actually increased from 71% to 79% following the implementation of AI. These results speak for themselves: a strategic, human-centered approach to rolling out AI has delighted Thrasio’s customers and employees.

Thinking about AI? Just do it

When it comes to advice for support leaders undergoing their own AI transformation, Gershwin’s advice is clear:

1. Create a strategy and circulate it. “Take the time to write it down,” Gershwin said. “This will make sure people know what’s going on, even when you’re not there to explain it.”
2. Do your research.  As Gershwin put it: “Learning about AI is so much more than a 10-minute conversation. That would be like going back to the 90s and asking someone to explain the internet to you! There’s enough free content out there for learning. Plus, there are so many tools you can play with and improve your practice.”
3. Just do it. “It’s ok to be imperfect,” Gershwin said. “It’ll be better than what you have right now, and then you can iterate to improve over time.”

Ready to start your AI transformation? Book a demo today.

‍