Insights from the AACSearch team about search relevance, ranking, and building a search platform.
Why facet counts get slow as a catalogue grows, the difference between an exact and an estimated count, and how to keep faceted navigation fast without hiding filters shoppers need.
How AACSearch signs outbound webhooks, what happens when your endpoint is briefly down, and how an Idempotency-Key keeps a retried write from landing twice.
Collection namespacing on the engine proxy, full versus search key scopes, short-lived browser keys, per-organization quotas in a Durable Object and organization-scoped events — what a platform operator gets when embedding search for many customers.
A search box answers one query. A DIY engine gives you a retrieval primitive. A Search OS is the layered system around it — query understanding, indexing, synonyms, ranking rules, curation, results and analytics — that a tenant actually controls. Here's how AACSearch implements each layer.
Click-through rate tells you whether a ranking change helped in production. A golden query set tells you whether it's correct before you ship it. Here's how AACSearch's quality lab computes nDCG, MRR and MAP.
How to write pins, boosts and exclusions that keep working after your product feed changes, and why ID-based rules quietly break.
How AACSearch combines lexical and vector search results into one ranked list, and why reciprocal rank fusion beats naive score blending.
Why zero-result search rates spike after a catalogue import, and the layout, transliteration and broadening ladder AACSearch runs before giving up.