How Lovepop’s AI-led customer service strategy is driving growth

Streamlining your customer support apparatus these days means leaning into the fast-growing ecosystem of AI platforms: In a perfect world, they offer maximum efficiency on the front end, and a treasure trove of data on the backend. While there’s still a need for the human touch in customer service, many companies are using AI tools for that crucial first contact.

One company taking this approach is Boston-based 3D greeting card brand Lovepop. Co-founded by CEO Wombi Rose in 2014, Lovepop incorporated AI into its customer service team—which Lovepop calls its “happiness team”— about three years ago. It has since seen these tools improve the customer experience and bring in data points that inform overall growth.

“The most important area that brands are trying to lean into AI is really around keeping up with those customer expectations,” said Danielle Harvey, VP industries, partnerships and emerging products, at digital analytics platform Quantum Metric. “There is a focus on being more efficient within a lot of these companies, so the two of those areas can overlap and AI can certainly help work faster, better, and more efficiently.”

Growth in numbers

According to AI customer service platform Crescendo, Lovepop utilized its omnichannel AI assistant—nicknaming it “Joy”—across voice and email to help with customer conversations and product recommendations. The voice assistant was then integrated with Lovepop’s other software, including Zendesk (CRM) and Gorgias (a help desk platform).

Rose said Lovepop adopted AI tools as a way to meet customer demand during seasonal spikes—like Valentine’s Day and Mother’s Day. The aim was to “alleviate some of that training load” and to “scale [Lovepop’s] trained people into those seasons a little bit better.”

According to Rose, AI tools have increased conversion rates by 25% and resulted in an average higher order value, while its net promoter score and customer satisfaction score stayed pretty much the same.

“We’re really trying to help people just send more cards,” Rose said. “We have consistently seen that anyone who’s interacting with our happiness team is just a stronger long-term customer.”

“Using a single AI chat for both customer questions and product discovery led to stronger shopper conversion and larger baskets for those customers who chatted with us,” said Emilio Latorre, SVP of operations at Lovepop, via a Crescendo press release.

Rose said that even though the AI tools have been beneficial for Lovepop’s customer service strategy, he still thinks it’s important to know when to transfer over to a company representative to solve an issue.

“AI is just a tool, just like a calculator—like you wouldn’t expect a calculator to do your math homework,” he said. “It’s really our people who are doing the work and making the decisions.”

Finding opportunities

Internally, Lovepop’s AI tools have helped the company find opportunities for new products, identify gaps in their offering, and understand customers better. In short, it’s an opportunity to leverage data.

Rose gave the example of the brand’s No. 1 Valentine’s Day “possum card”—featuring a pop-up trash can, the text, “Life without you is trash,” and a slide-out note with a message that can be printed on demand; it has since become one of the top-selling Mother’s Day cards. “The tool that we use to merchandise within that taxonomy is just like one piece of the equation, but then also the tools that we’re using to see what people are sending more of. Then that shows us gaps in our portfolio,” Rose said.

Lovepop also offers many general greeting cards, and the AI recommendation tool helps it see how those products are used. Rose said the company gets “about 100,000 design requests every year,” and the AI tools help because “eight people can’t parse 100,000 design requests.”

Rose explained that the AI tools fall under Lovepop’s four values of “engineering, hunger, magic, and connection,” specifically the first two.

“We’ve had use cases where for six months, we were just banging our heads against the wall and we weren’t able to figure out how to get the tools to do it,” he said. “Then we tried different approaches and all of a sudden we can get it to work.”