Addesu & Maison Flaneur: Blending Unique Homeware with Unique AI-Powered Feeds

The Challenge

Maison Flaneur, a premium homeware marketplace, sought to improve the efficiency of its paid search campaigns across UK and US markets. Their product feed, spanning over 60,000 items from multiple brands, updated daily, posed significant challenges for Google Shopping, where product title quality is critical for matching user intent.

The Strategy

Addesu built a custom GenAI pipeline using ChatGPT to transform product titles at scale. This automated system extracts key product attributes, rewrites titles to meet Google best practices, filters low-quality entries, and manages translations, all in one prompt.

Execution and Innovation

A structured A/B test was run across two six-week periods, allocating 33% of budget to the AI-enhanced group. Monthly, the system processes up to 60,000 products in just a few hours, dramatically improving title quality and targeting precision. The innovation has since been integrated into the Addesu platform for use by other brands.

Results

In the UK, conversion rates rose by 21.36% in test campaigns versus an 11.72% drop in control. In the US, conversions increased 41.18% despite fewer clicks, underscoring dramatically improved efficiency.

Judge's Verdict

"A clever and effective solution to the genuine operational challenge of managing inconsistent multi-brand product feeds at scale."

"Genuine engineering rigour and a sophisticated prompt structure – a clever idea successfully evolved into a scalable product."