Abandoned-cart recovery that won back 18 percent of carts that would have been lost

The client

A direct-to-consumer fashion label run out of Bangkok, shipping across Southeast Asia. Around fifteen people on the team. Email marketing was being looked after by one person who already had three other jobs.

The problem

Cart abandonment sat at the standard ecommerce seventy percent, but the recovery flow was a single generic email going out twenty-four hours after the abandonment. It was pulling in roughly four percent of those carts. The marketer knew the flow could do a lot more, but writing personalized copy for thousands of weekly variations was not realistic with the time she had.

The approach

The build wired Shopify cart events into n8n and fired three messages at carefully chosen intervals: a soft check-in two hours later, a personalized recommendation twelve hours after, and a single-use discount on the morning of day three. Each message was drafted on the fly by an LLM using the specific items in the cart, the customer’s past purchases, and the local time of day.

The marketer kept editorial control through an approved tone guide and reviewed a sample of generated messages at each weekly check-in. The system shipped roughly twelve thousand personalized messages a month after that without manual intervention.

The result

Cart recovery climbed from four percent to eighteen percent over the first three months. On their order volumes, that came out to roughly forty thousand baht of recovered revenue every week that had been walking out the door. The marketer got her week back for the kind of campaign work that needs a human, rather than the kind that does not.