Enterprise GenAI Product-Metadata Program
April 25 – June 6, 2025
Executive Summary
To boost the performance of Google Shopping campaigns, our retail client's GenAI team piloted an automated Product-Metadata Generation pipeline. By adding richly descriptive, AI-generated tags to sports-bra & legging SKUs, the program produced a 19% lift in clicks, 16% more conversions, and an 8.6% higher click-through-rate (CTR) in just six weeks—while cost-per-click fell 7.2%.
The Challenge
- Sparse feed data: Product titles and descriptions in Merchant Center were too generic ("Women's Sports Bra") to rank competitively.
- Manual enrichment bottleneck: Human copywriters couldn't scale to hundreds of SKUs per season.
- Paid-media pressure: The EMEA Integrated Media team needed quick, measurable wins ahead of FY25 planning.
Objectives
| KPI | Target | Outcome | Status |
|---|---|---|---|
| CTR | +5% | +8.61% | |
| Conversions | +10% | +16.31% | |
| CPC | –5% | –7.23% |
Solution Architecture
- Data Ingestion: Pulled product attributes from the client's PIM API.
- GenAI Tagging Service: Prompt-engineered an internal LLM to output SEO-ready tags per SKU.
- Feed Builder: Merged tags into new titles/descriptions and pushed to Merchant Center.
- Governance Layer: Automated QA for brand-voice compliance.
- A/B Orchestration: Assigned 50% traffic to Control and 50% to Variant feeds.
Results Highlights
p < 0.001
p < 0.001
p = 0.003
p < 0.001
Trend ↓
Business Impact
- Revenue upside: Projected annual incremental revenue $1.2M+ if rolled out across women's apparel in EMEA.
- Creative savings: Eliminated ~120 copy-hours per collection; writers now focus on high-impact storytelling.
- Template for scale: Framework is reusable for other categories.
My Role
Product Lead
Defined hypothesis, KPIs, and experimental design.
Architect & PM
Stood up the GenAI tagging micro-service (Python + Vertex AI) and feed integration in three weeks.
Change Champion
Secured buy-in from Paid Media leadership and trained partners for full hand-off.
"By marrying generative AI with rigorous experimentation, we demonstrated that smarter product metadata directly translates to measurable paid-media gains. The pilot's success paved the way for a global rollout and set a precedent for data-driven GenAI initiatives across the enterprise."
Andrew Hallberg
Senior Program Manager – AI @ Microsoft | Co-Founder & CTO @ HirelyAI
Andrew leads cross-functional AI and digital commerce programs at Microsoft and co-founded HirelyAI, a GenAI-native hiring platform. He specializes in AI program management, product strategy, and ethical AI implementation.