Aeion Search for E-Commerce Product Discovery
Site search drives 30-50% of e-commerce revenue. Shoppers who use search convert at 2-3x the rate of browse-only visitors. Algolia + Constructor + Bloomreach + Coveo sprawl costs $50K-500K+/year for the major retailers. Aeion Search delivers hybrid RRF retrieval, semantic intent matching ("dress for summer wedding"), real-time inventory-aware ranking (don't surface out-of-stock SKUs), and personalized re-ranking — all native to your commerce DB. Walkthroughs for DTC storefronts, multi-vendor marketplaces, B2B catalogs.
What E-Commerce Search Demands
`
Walkthrough — DTC Apparel Storefront
`
Walkthrough — Multi-Vendor Marketplace
`
Walkthrough — B2B Industrial Catalog
`
FAQ
Yes. Out-of-stock products are suppressed (or down-ranked) at query time against live inventory — no separate sync needed.
Postgres + pgvector + HNSW scales with read replicas + connection pooling — add replicas ahead of an anticipated traffic spike like Black Friday.
User-history boost via embedding-of-recent-views. Per-user re-ranking is a SQL window function over the top-50.
Per-query overrides (pin, suppress, redirect, boost), managed from an admin UI built for the merchandising team.
Each query records its variant. Conversion attribution ties back to the resulting order. Compare RPV across variants with the BI module.
Postgres ships stemming for 10+ languages out of the box. Per-locale full-text vectors + per-locale embedding column.
Not yet — Aeion Search's embedding pipeline is text-only today. Image-based similarity search is on the roadmap.
Postgres trigram matching plus a per-tenant frequency table. Comparable to Algolia's typo tolerance.
Zero-result queries are flagged automatically. The merchandising team gets a weekly digest of unmet demand — a catalog-gap signal worth real revenue.
Constructor: $50-300K/year. Aeion: $0 platform + ~$100-500/month embeddings. 95%+ savings.