Discussion
  • eCommerce SEO
  • WordPress SEO

WooCommerce stores past 20k SKUs: have SEO problems ever forced a replatform for you, or were plugin, caching and hosting fixes enough?

Published 1 reply

Discussion starter: an ecommerce SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    The SEO problems that show up on WooCommerce past 20,000 SKUs are mostly implementation problems, not platform limits: crawlable filter, sort and cart URLs, uncached pages that respond slowly to Googlebot, plugin stacks that output conflicting tags, and weak internal links to deep products. Each has a fix on WooCommerce, and none is fixed automatically by moving. Replatforming for SEO alone is hard to justify, because it swaps a known set of problems for migration risk and a new set of constraints.

    The case for moving is real when the driver is operational (catalogue complexity, integrations, multi-market selling, or a plugin stack nobody can maintain) and the SEO symptoms keep returning after they are fixed. A practical test: if the store can hold clean crawl patterns and fast uncached responses through six months of normal plugin updates and merchandising changes, the platform is not your SEO problem.

    On whether SEO problems have forced a replatform: no controlled public comparison of fixing versus moving exists that I can point to, and migration stories are published by the people who ran them, so they lean towards successes. The useful answer is a diagnosis that tells you which kind of problem you have.

    Where WooCommerce strains at this size

    WooCommerce stores products and variations as WordPress posts, with much of the product data in the post meta table. A catalogue of 20,000 parent products with variations can mean hundreds of thousands of rows, and pages that filter or sort on that data get expensive when they are not cached. WooCommerce has added lookup tables to speed up catalogue filtering and sorting, and High-Performance Order Storage (HPOS) moves orders into their own tables. HPOS helps order-heavy admin and checkout; it does not change how category and product pages are queried.

    The crawl traps are specific and predictable:

    • Filter URLs. The classic layered navigation widgets add parameters such as filter_colour=, query_type_colour=, min_price= and max_price=, and filter plugins add their own patterns. Every combination is a crawlable URL if it sits behind a plain link.
    • Sorting. orderby= multiplies every category and filter URL by the number of sort options.
    • Add-to-cart links. Archive “Add to cart” buttons for simple products link to ?add-to-cart= URLs. WooCommerce marks them rel="nofollow", but nofollow is treated as a hint rather than a block, and each of those requests is uncacheable work for the server.
    • Geolocation parameters. With the default customer location set to “Geolocate (with page caching support)”, WooCommerce’s documentation says you may see characters like ?v=xxxxx appended to your URLs, which prevents static caching of prices. That can put a parameter version of your URLs in front of crawlers. Check logs and URL Inspection for ?v= requests before assuming canonical tags take care of it.
    • Internal search. ?s= result pages can be slow on large catalogues, and there is no reason for Googlebot to crawl them.

    Google’s faceted navigation guidance names the two costs: overcrawling, and slower discovery of new, useful URLs because crawling is spent on low-value ones. Server speed feeds into the same problem. Google’s crawl budget guide explains that when a site slows down or returns server errors, Googlebot crawls less.

    Important distinction:

    A crawl trap and a slow server look similar in Search Console (products stuck in “Discovered - currently not indexed”, new products picked up late) but need different fixes. A crawl trap is fixed by stopping the crawl of worthless URLs; a slow server is fixed with caching and capacity. The two compound each other, because uncached filter URLs are expensive requests for the server.

    Diagnose before deciding anything

    1. Server logs: split Googlebot requests by pattern: products, categories, filter and sort parameters, add-to-cart=, ?v=, search, feeds and anything else. If parameter URLs take a large share, you have a crawl trap, and it can be fixed on WooCommerce.
    2. Crawl stats report: average response time, host status and the response mix. Rising response times or 5xx errors during crawl peaks point to caching and hosting.
    3. Page indexing report: read the labels separately. Product URLs in “Discovered - currently not indexed” point to crawl capacity or low crawl priority. “Crawled - currently not indexed” points to quality or duplication, such as manufacturer descriptions shared with many other retailers. “Duplicate, Google chose different canonical than user” points to variants and near-identical products. “Alternate page with proper canonical tag” on parameter URLs is expected when canonicals are set correctly.
    4. Uncached speed: test category and product templates with the cache bypassed. That is closer to what Googlebot gets on parameter URLs than a cached homepage test.
    5. Plugin output: check canonical, robots meta, titles and structured data on each key template after every plugin update. Two plugins writing Product markup, or a setting reset by an update, can undo months of work.
    Symptom Likely cause Fix on WooCommerce Would a new platform fix it?
    Most Googlebot requests hit filter, sort or cart URLs Crawlable parameter URLs robots.txt rules for parameters that should not be crawled; clean, linked URLs for the few filter pages worth indexing Not by itself. Hosted platforms generate filter parameters too, and the same decisions have to be made there.
    High response times, 5xx errors during crawl peaks Uncached PHP and database work Full-page cache for anonymous visitors, object cache, more PHP workers, CDN, search offload Partly. Hosted platforms absorb capacity, at the cost of fees and flexibility.
    Thousands of products “Crawled - currently not indexed” Duplicate manufacturer copy; variants listed as separate thin products Rewrite or enrich priority products; consolidate variants No. The same content moves to a new template.
    Canonical, noindex or markup changes after updates Overlapping plugins, no monitoring Remove overlap, lock settings, automated template checks Partly. Fewer plugins, but apps add their own code.
    Hreflang and currency set-ups fighting each other A multilingual and multi-currency plugin stack Simplify the market structure and retest Possibly, if the target platform’s market features fit your structure.

    The fix stack, roughly in order of cost

    1. Crawl control. Decide which filter pages deserve indexing (brand and category combinations with real demand, for example) and give them clean, linked URLs with unique titles. Disallow the rest of the parameter space, sort orders, add-to-cart and internal search in robots.txt. Google’s guidance recommends robots.txt disallow, or URL fragments, for facets you do not need indexed, and says canonical and nofollow are generally less effective in the long term. If you do want some facet URLs crawled, it suggests the standard & separator, a consistent filter order and a 404 for combinations with no results.
    2. Caching. Full-page caching for anonymous visitors on product and category pages, and an object cache such as Redis for database queries. Check that cached pages still show the right stock and price.
    3. Capacity. Enough PHP workers and database resources for crawl peaks, not just average traffic. Watch Crawl stats response times after each change.
    4. Database hygiene. Autoloaded options, expired transients, orphaned post meta and revisions. Turn on HPOS if orders are part of the load.
    5. Search offload. Move product search to a dedicated search engine and keep result pages out of the index.
    6. Plugin audit and monitoring. One SEO plugin, one source of Product markup, and automated checks on key templates after each update.
    7. Links to deep products. Crawlable category pagination with plain links (not only a JavaScript “load more”), breadcrumbs, and related-product links that reach beyond the bestsellers.

    Watch for this:

    A robots.txt block stops Google seeing any noindex or canonical on the blocked URLs, so parameter URLs that are already indexed can sit in the Page indexing report as “Indexed, though blocked by robots.txt” for a while. If index clean-up matters more than immediate crawl relief, let Google see a noindex first and block afterwards. If the server is struggling, block first and accept the lag.

    Catalogue churn adds a merchandising decision at this size. WooCommerce’s setting to hide out-of-stock items removes them from category listings, and with them the internal links, while the product URLs stay live. Products that will return are worth keeping listed. For permanently discontinued products, redirect to a clear replacement where one exists. If the page stays, mark availability as https://schema.org/Discontinued and keep the price in the Offer, because merchant listings need a price on every Offer.

    Hypothetical example:

    A homewares store has 24,000 products and 130,000 variations. Logs show most Googlebot requests going to filter, sort and add-to-cart URLs, Crawl stats shows average response times above a second, and 8,000 product URLs sit in “Discovered - currently not indexed”. The owner has a replatforming quote. Before signing, the team blocks the parameter space, gives 150 brand-and-category filter pages clean URLs, adds full-page and object caching and upgrades PHP workers. Success after six to eight weeks would look like a much smaller parameter share of Googlebot requests, lower response times and a shrinking discovered-but-not-indexed backlog. If those move, the replatform case has to stand on non-SEO grounds. If they do not, it gets stronger.

    The six-to-eight-week window is deliberate. Changes to robots.txt are picked up within about a day, so crawl patterns should shift in the first two weeks. Working through a backlog of undiscovered products takes longer, because it runs at whatever crawl rate the faster server earns.

    When replatforming is the right call

    • Operational requirements WooCommerce can meet only with heavy custom work: complex B2B pricing, multi-warehouse stock, frequent ERP or PIM syncs across the full catalogue, or many markets with localised catalogues.
    • No developer capacity to maintain the plugin stack, so SEO regressions return after updates.
    • Total cost of ownership (hosting, development, licences, incident time) above what a hosted or enterprise platform would cost, including the migration.
    • A performance ceiling that remains after crawl control, caching and capacity work: uncached templates still slow, and Crawl stats still showing high response times or server errors.

    In each case, SEO is a requirement and a risk line in the decision rather than the reason for it. The migration itself is a major SEO risk: 20,000+ product URLs, categories, images and reviews all need mapping and redirects.

    Questions to put to any target platform

    • Can we control the URL structure, or are there fixed prefixes? Shopify, for example, fixes /products/ and /collections/, so expect product and category URLs to change, with a redirect map covering the whole catalogue.
    • How are filter URLs generated, and can we block or index them selectively?
    • Can we edit robots.txt, canonicals and robots meta per template?
    • What are the current limits on variants and options per product, and does our catalogue fit them?
    • How are redirects managed at 20,000+ URLs: bulk import, limits and performance?
    • Can we control Product structured data and XML sitemaps?
    • What do the apps or extensions we need add to page weight?
    • Which AI shopping and agentic checkout integrations are supported? Google’s Universal Commerce Protocol lets shoppers check out with Google Pay inside AI Mode and Gemini with participating merchants, including Shopify merchants. As of September 2026, Google’s help page describes UCP-powered checkout as available to select merchants for products eligible in the United States, Canada and Australia, and WooCommerce’s feature request for native UCP support, posted in January 2026, is still open. Weigh it only if your market is covered and the channel matters to your buyers.

    A practical position

    If this were my store, I would treat a replatforming quote as the trigger for a six-to-eight-week fix sprint, not as the fix. Block the parameter space, cache and resource the templates Googlebot needs, remove the plugin overlap, and measure crawl share, response times and the indexing backlog. If those numbers recover and hold through a few plugin updates, the platform is not the SEO problem. If they do not, the replatform case is stronger, and SEO’s job becomes writing the platform requirements and the redirect map before anyone signs.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.