Product Exclusion Rule Engine

Challenge

Running product ads across large catalogues means a long tail of items that quietly drain budget — low return on ad spend, poor conversion, wrong commission tier. Finding and excluding those products by hand, across multiple marketing networks and thousands of SKUs, is impractical, so wasted spend accumulates. The client needed a way to define exclusion and labelling rules once and have them applied automatically, with the ability to review candidates and undo mistakes.

Solution

We built a rule-driven exclusion engine for Shareight that identifies low-performing products and removes or relabels them across marketing networks (eBay and Affpro). Operators define rules — based on performance metrics, commission tiers and product patterns — in a clean interface, including drag-and-drop ordering of how rules run. The engine evaluates products against those rules and uses AI (Google Gemini) to catch low-value products that simple filters would miss. Matching exclusions are applied to Google Ads automatically, and every action is recorded with full history so changes can be audited and undone. The rules run on a schedule so the catalogue stays optimised without manual effort.

Project Results

Wasted ad spend on poor-performing products is now pruned automatically by rules rather than manual review, across multiple networks at once. Operators can see exactly which products a rule will affect, reorder rule priority visually, and undo any exclusion thanks to full history tracking. AI pattern matching extends exclusions to products that metric thresholds alone would miss.

Features and Benefits

• Rule-Driven Exclusions: Define performance and commission-tier rules once and apply them automatically.
• AI-Assisted Matching: AI catches low-value products that simple thresholds would miss.
• Visual Rule Ordering: Drag-and-drop control over the order in which rules run.
• Google Ads Integration: Exclusions are applied to live Google Ads campaigns automatically.
• Full History & Undo: Every action is audited and reversible, so changes are safe to make.

Technologies Used

• Go: Rule engine, product evaluation and scheduled processing.
• PostgreSQL & MySQL: Rules, candidates, campaigns and network data.
• Elasticsearch: Fast product lookup supporting rule evaluation.
• Google Gemini: AI-assisted detection of low-value products.
• Google Ads API: Applying exclusions to live campaigns.
• Next.js & Material-UI: Rule builder, candidate review and history interface.

Development Process

We modelled the work around rules and candidates: operators define rules, the engine produces candidate products, and exclusions are applied and recorded. We combined performance metrics with AI assistance so the engine could catch low-value products that thresholds alone would miss. The interface was built for control and confidence — visual rule ordering, clear candidate previews, and full history with undo. The rules run on a schedule, while the audit trail makes automated exclusions safe to trust.