How Adore Me used AI to double its active customers

“We had a one-size-fits-all strategy for reaching out to customers,” said Josselin Petit-Hoang, Adore Me’s marketing manager who has been with the company since 2014. “But as the company grew, and the volume of our customers grew, we wanted to narrow down, as well as reach people outside of email.”

Adore Me — which has raised close to $12 million in venture capital — did $80 million in sales in 2016 and is projected to hit $100 million in 2017, according to internal figures. The online company, which is now sold in some Nordstrom locations, offers a (somewhat controversial) VIP membership model: VIPs get lower prices, free shipping and returns, and other perks, but have to check in once a month to either make or skip a purchase, a part of the deal that has blindsided some customers in the past. Otherwise, a $34.95 charge is credited to their account.

Customers don’t need to be VIP members to shop Adore Me, and this dichotomy dictated the brand’s marketing messages. Petit-Hoang said that, previously, the company had divided its messaging groups into four: members who had recently made a purchase, non-members who were active shoppers, VIP members who hadn’t recently made a purchase and inactive previous customers.

The company mined existing customer information to personalize and automate its messaging process, moving outside of email to text notifications, in-app messages and Facebook ads. Using Optimove’s technology, the company divided customers into 60-plus segments. Amit Bivas, CMO at Optimove, said that Adore Me first segments customers into new, active, and “churn,” meaning they hadn’t made a purchase in a while. From there, customers are broken down into segments like the categories they shopped, what device they shopped on, and how much they tend to spend. He said that a campaign strategy that would take months from start to finish to complete is cut down to a few weeks.

As a result, Adore Me saw a 15 percent increase in monthly revenue from its data-driven campaigns, and a 22-percent increase in average order amount. Its yearly active customers doubled.

“Retailers are still scared to give away access to their data,” said Eric Brassard, a former Saks Fifth Avenue executive and founder of AI recommendation company Propulse Analytics. “It’s like underwear; you don’t share that. But they’re beginning to realize they need better personalization and recommendation, and that they’re not going to become specialists in big data overnight.”

While its overall marketing team is made up of 22 members, just two are working to data-mine and customize messages with Optimove’s technology.

“When you think of the number of notifications you can send in a week, you need to be as relevant as possible” said Petit-Hoang. “This comes down to AI telling us exactly when a message should be sent. And to save resources, you need to automate as much as you can. We are the same team, sending so many personalized messages. Everything is personalized, and everything is automated.”

Petit-Hoang said that the team hasn’t yet explored the full breadth of what AI, machine learning and automation can do for the brand. But elsewhere — and in the lingerie space — brands are already experimenting beyond targeted messages. Cosabella, the Italian lingerie brand, uses AI to inform its product development and to expand into new categories, as well as format its website.

“AI can be used in a subtle way to optimize purchase flow, aid with efficiency and insights, and offer more personalized experiences,” said Mark Jarecke, creative director of agency Four32C. “People often assume AI will just help the consumer. But on the administrative side, they’ll free up staff hours to relocate elsewhere.”