Search Performance Optimization

Field Weighting Strategies

Field weights determine how much each product field contributes to search relevance. Getting weights right is the single most impactful thing you can do for search quality.

• Title: Highest weight (3.0–5.0) — the most important signal for relevance
• Description: Medium weight (1.0–2.0) — useful for long-tail queries
• Tags / Keywords: Medium-high weight (1.5–2.5) — great for curated metadata
• SKU / Product ID: Very high weight (4.0–5.0) for exact matches — customers searching by SKU expect exact results
• Brand: Medium weight (1.5–2.0) — important for brand-loyal customers

Configure in Dashboard > Search Ranking > Search Weights.

Facet Selection Best Practices

Facets (filters) are powerful but only when applied to the right fields:

• Good facets: category, brand, color, size, price range, rating, availability — bounded values with meaningful groupings
• Bad facets: description, title, SKU — too many unique values, creates an unusable filter list
• Aim for 8–10 facets maximum in the UI to avoid overwhelming users
• Use price range sliders rather than discrete price values
• Show result counts next to each facet value so users know what to expect

Boosting vs Burying Products

Use boost and bury rules to influence ranking beyond pure relevance:

• Boost: High-margin products, new arrivals, in-stock items, featured products
• Bury: Out-of-stock items, discontinued products, low-margin items
• Configure in Dashboard > Merchandising > Boost Fields
• Use Click-Based Ranking to let real user behavior drive ranking automatically over time — this often outperforms manual rules

Index Optimization

• Remove unused fields from the index to reduce size and improve query speed
• Use appropriate field types: text (full-text search), string (exact match), integer/float (range filters)
• Enable variant grouping to avoid duplicate results when a product has many size/color variants — shows one result per product with variant selector
• Schedule regular re-crawls to keep the index fresh; use incremental crawls for daily updates

Crawl Speed Configuration

• Balance crawl speed against your server’s capacity — too fast can overload your site
• Configure in Dashboard > Crawl > Crawl Speed
• For large catalogs (100k+ products), use feed upload (CSV/XML/JSON) instead of crawling — it is 10–100x faster
• Schedule full re-crawls during off-peak hours (e.g., 2–4 AM) to minimize impact on site performance
• Use incremental crawls for daily updates and full crawls weekly

Caching Best Practices

• ExpertRec search results are cached at the edge via Cloudflare — common queries are served in under 10ms
• Cache is automatically invalidated when products are updated via webhook or feed
• For real-time inventory accuracy, use the Search API with the cache-control: no-cache header for inventory-sensitive queries
• Static assets (search widget JS/CSS) are served from a global CDN — no configuration needed

Monitoring Search Performance

Use the Analytics dashboard to track key performance indicators:

• Zero-results rate: Target below 5%. High zero-results indicates missing synonyms, crawl gaps, or poor query understanding. Review zero-result queries weekly and add synonyms or content.
• Click-through rate (CTR): Target above 30%. Low CTR suggests results are not relevant — review top queries and adjust weights or merchandising.
• Search latency: Target under 200ms for autocomplete, under 500ms for full results. Check Dashboard > Analytics > Search Performance for latency trends.
• Top queries: Review weekly to identify new trends, missing products, or opportunities to add synonyms.
• Conversion rate from search: Customers who use search convert at 2–3x the rate of non-searchers — track this to demonstrate search ROI.