Turning 3-Hour Gift Searches Into One Quick Chat

The Challenge

Shoppers on the platform spent an average of 45 minutes per session trying to find the right gift across a catalog of over 10,000 products. The sheer volume of options made it hard for buyers to narrow things down without help. That meant longer sessions, fewer completed purchases, and an estimated $18,000 per month in affiliate revenue left on the table.

What We Built

We built an AI agent that opens a natural conversation with each shopper, asking about the recipient, the occasion, and their budget to narrow millions of products down to a handful of strong matches. A second automation layer takes those picks, searches the live product catalog, filters for availability and best seller status, ranks the results, and caches them so the system stays fast under heavy traffic. The full loop from first message to curated product cards runs in under 30 seconds.

Results

Average time from landing to purchase dropped from 45 minutes to under 3 minutes. Repeat visits grew by more than double within the first quarter after launch. Daily product recommendations served went from zero to over 1,200 per day without a single manual intervention. The affiliate revenue generated in the first 90 days paid back the entire build cost and then some.