WooCommerce’s own benchmarks show order lookups that took 200ms on legacy storage often run in 20 to 40ms after HPOS. That does not fix catalog search by itself, but it stops admin slowness from blocking your team during peak season.
what breaks first at scale
Category and filter pages: themes that call wc_get_product() inside loops without caching load full product objects per row. Faceted search plugins that JOIN meta tables multiply cost.
Search: default WordPress search ignores SKU and attributes. Users type SKU; database scans fail; they leave.
Admin orders and exports: legacy order storage in wp_postmeta slows exports and order list screens past roughly 100,000 orders without HPOS.
Cron and sync jobs: inventory sync plugins that write thousands of rows per hour compete with frontend queries on the same database.
| Catalog size | Typical symptom | First infrastructure step |
|---|---|---|
| 5,000 SKUs | Slow filters | Query audit, object cache |
| 10,000+ SKUs | Search timeout | Elasticsearch or Algolia index |
| 50,000+ SKUs | Admin and FE both slow | Dedicated DB resources, HPOS, CDN for assets |
| 100,000+ orders | Order admin unusable | HPOS with sync completed, archive policy |
ugins Run WooCommerce → Settings → Advanced → Features → HPOS compatibility report.
hpos migration without breaking plugins
Run WooCommerce → Settings → Advanced → Features → HPOS compatibility report. Update or replace incompatible plugins before enabling. Use sync mode for two to four weeks. Verify exports, fulfillment plugins, and custom reports against both storage systems.
Catalog over 10,000 SKUs?
Talk to us →redis, elasticsearch, and hosting thresholds
Catalog over 10,000 SKUs?
Talk to us →Redis object caching reduces repeated option and meta queries. It helps most when the same category pages are hit often. Elasticsearch (or managed search) moves full-text and SKU search off MySQL. Plan $500 to $2,000 monthly hosting at 50,000 SKUs with serious traffic: 4+ CPU cores, Redis, staging that mirrors production data volume, and automated backups you have actually restored once.
reading slow query logs
Enable slow query log in MySQL for queries over one second. Reproduce a slow category load, then grep the log for wp_postmeta and wp_term_relationships. Plugins like Query Monitor on staging show duplicate queries per page. A healthy category template should not run hundreds of identical meta queries.
what this looks like when we do it
A large-catalog performance sprint is usually two to four weeks: week one for profiling and HPOS readiness, week two for search index and caching, weeks three to four for theme loop fixes and load testing. Emergency tuning before Black Friday can compress this to one week if scope is limited to search and caching only.
We deliver before/after query timings, a hosting recommendation, and a prioritized backlog. Ongoing work fits website management with monthly performance reviews. Net-new architecture changes go through ecommerce development when you need custom search or ERP integration.
At 10,000+ SKUs, search and database design matter more than another caching plugin. Measure queries, enable HPOS, index search, then tune the theme.
