What is Query Intent Classification?
Query intent classification predicts the purpose behind a query—transactional/product, category/browse, navigational/brand, informational/help, or account/policy—so the engine can render the most useful experience.
How It Works (quick)
- Signals: Tokens/phrases, embeddings, entity/brand matches, session context.
- Models: Rule + ML (transformers) trained on clicks/conversions.
- Routing: Choose SERP type, preload facets, trigger Best Bets/redirects.
- Feedback: Learn from reformulations and downstream actions.
Why It Matters in E-commerce
- Sends “gift card” to the landing page, not a product grid.
- Prefills size/brand facets for product-seeking queries.
- Improves CTR, zero-results rescue, and support deflection.
Best Practices
- Train per locale/category; keep override rules for legal/compliance pages.
- Show subtle chips/badges indicating inferred intent (editable).
- Log errors/backoffs; fail safe to generic SERP.
Summary
Detect intent, then shape the SERP and navigation accordingly—fewer dead ends, faster paths to purchase.
