Whether Yoast is suitable for a store with millions of products depends on the workload, software versions, data structure and infrastructure. Product count alone cannot establish a capacity limit. Test metadata output, sitemaps, imports, internal SEO-data processing and editing separately before choosing an architecture.
This is a capacity-assessment guide, not a claimed ten-million-product benchmark. First clarify whether “ten million records” means products, variants, inventory entries, database rows or public URLs. Those are very different workloads.
Separate Yoast's responsibilities from the rest of the store
Yoast handles SEO information and related features, including metadata, structured data, sitemaps and editorial assistance. Product searches, stock synchronization, filtering, carts and orders also involve WooCommerce, the theme, other plugins and infrastructure.
A slow product page does not identify the responsible component. Complex attributes, variant relationships, languages and frequent stock updates can change the load substantially even when the product count stays the same. Cached visitor requests and logged-in editing requests also follow different paths.
| Workload | Observe | Keep the conditions explicit |
|---|---|---|
| Individual product requests | Response time, database work and final SEO output | Cold and warm caches; visitor and authenticated states |
| Lists, searches and filters | Query plans, expensive queries and response tails | Attribute complexity, variant relationships and filter combinations |
| Editing and saving | Browser work, save requests and background processing | Editor settings, other plugins and concurrent editors |
| Imports and updates | Throughput, errors, retries and data consistency | Normal save hooks versus custom writes |
| Sitemap requests | Index and individual shard responses, URL counts and updates | Products, taxonomies, authors and other included resources |
| SEO-data optimization | Processing time, memory and database load | Initial construction versus subsequent incremental work |
Understand Indexables before trying to repair them
Yoast's Indexables architecture stores SEO information in a dedicated data layer so it can be retrieved more efficiently. This is internal plugin data. Running SEO-data optimization does not cause Google to index the website.
After a migration, custom import or relevant upgrade, check whether this data agrees with the actual products. Inspect a representative range of final titles, descriptions, canonical URLs and structured data. A completed progress bar is not a substitute for checking the outputs that matter.
The documented WP-CLI command for the indexation process is:
wp yoast index
The example assumes compatible Yoast and WP-CLI installations and the correct site context. The --reindex option removes existing Indexables before rebuilding them; --network expands the operation to sites in the network. Neither should be appended automatically to every troubleshooting command.
Confirm the recovery route, scheduling and expected load before a substantial operation. No command has been run on your store as part of this guide, and a successful run on a small test dataset does not establish large-store capacity.
Find the particular sitemap request or database query that is slow
Request the sitemap index and representative product, taxonomy and author shards separately. Record timings, status codes, application errors and relevant slow queries. A slow root index does not necessarily mean every product is being traversed in that request.
Yoast's 27.8 technical release explanation describes improvements involving author-sitemap user queries, admin Indexables checks, batched database access and editor rerenders. Its example involving approximately two million users is a vendor test for that particular query, not proof of the same improvement on every million-product store.
Use the actual installed version and query evidence to decide what to change. Do not add database indexes or increase memory merely because a generic recipe says every catalog above a particular size needs them. Extra indexes can affect writes, and higher resource limits do not fix every inefficient query.
Choose sitemap shard sizes by both protocol and observed cost
The Yoast sitemap documentation describes the wpseo_sitemap_entries_per_page filter for adjusting entries per shard. Implement any customization in maintained, version-controlled site code rather than editing plugin files directly.
Smaller shards can reduce the cost of a single request while increasing the number of sitemap files. Check generation cost, response size and cache behavior after changing the setting. Sharding alone will not solve a slow index query or incorrect URL selection.
Google's sitemap requirements limit a file to 50,000 URLs and 50 MB uncompressed; larger sets need multiple files and a sitemap index. These are protocol limits, not a recommendation to fill every WordPress-generated shard to the maximum.
Include appropriate canonical URLs intended for search, with accurate modification information where provided. Inventory rows and variants do not all require independent indexable pages. Verify that the generated URLs exist and their status, canonical and indexing controls agree.
Build a capacity test around the real business
- Record the environment: WordPress, WooCommerce, Yoast, theme, other plugins, PHP, database, cache configuration and hosting resources.
- Use representative data: retain realistic attribute distributions, variants, categories, languages and images rather than filling the database with identical empty products.
- Separate operations: test browsing, search, editing, import, sitemap requests and SEO-data processing independently before combining them.
- Increase load progressively: examine the expected dataset, traffic and update concurrency, including slow-tail responses and failures.
- Isolate causes: compare a controlled change in a test environment to determine which component explains the difference.
- Check consistency and recovery: verify product data and SEO outputs after normal updates, retries and interrupted jobs.
Set acceptance criteria from actual service requirements. CPU and memory averages are not enough: a store can have spare resources while a query or lock causes poor tail latency. A passing test at one scale should not be extrapolated linearly to ten million products.
Record the WordPress environment type too. Yoast's environment documentation says it does not build Indexables when wp_get_environment_type() is not production. Check the actual behavior in the isolated test environment; otherwise a staging benchmark may omit the very work you intended to measure. Document any configuration needed to reproduce the workload and retain the test site's access restrictions.
When to consider a different implementation
If the bottleneck is product filtering, inventory writes or search, address the responsible system. If measurements identify SEO metadata, sitemap generation or an internal indexing operation, evaluate the relevant version, configuration, scheduled processing or a purpose-built implementation.
Separating the storefront, product data, search and editorial content may suit some large projects, but it introduces synchronization and maintenance responsibilities. Replacing an SEO plugin also requires a migration plan for existing fields, redirects, canonical URLs and structured data. Check for duplicate or conflicting output during the transition.
The guide to WordPress SEO and performance tools explains overlapping responsibilities and output checks. Choose a tool or architecture on the basis of a reproducible workload test and an operational plan, rather than an unsupported product-count cutoff.