Fixing Category Architecture for an E-commerce Catalog

The challenge

The catalog had grown organically for years — new filters and sort options added whenever a developer had time, with no one tracking what that did to the URL count. A Search Console crawl-stats check showed Googlebot spending the overwhelming majority of its crawl budget on filtered, faceted-navigation URLs (color, size, sort-by combinations) that were functionally duplicate content, while genuinely important category pages were being crawled infrequently and, in some cases, barely ranked at all — cannibalised by their own filtered variants.

What made it harder

• Faceted navigation generated an estimated tens of thousands of indexable URL combinations, most differing only by sort order or a single filter value
• Several important category pages were competing directly against their own filtered variants in search results, splitting ranking signal instead of consolidating it
• Product-level schema was inconsistent — some templates had it, others didn't, and none of it matched cleanly to a defined entity structure
• Engineering resources were limited, so any fix needed to work within existing platform constraints rather than requiring a full replatform

The approach

• Ran a full crawl-budget audit via server logs to quantify exactly how much Googlebot activity was going to low-value faceted URLs versus real pages
• Defined a canonicalisation and indexation policy: which filter combinations should be indexable (genuine search-demand ones), which should canonicalise to the parent category, and which should be noindexed or blocked entirely
• Restructured the category hierarchy so navigation and internal linking reinforced the pages meant to rank, instead of splitting equity across dozens of near-duplicate variants
• Rebuilt product and category schema on a consistent entity model — Product, Offer, and AggregateRating markup applied uniformly across every template, not just some
• Consolidated the worst cannibalising URL clusters with 301s and canonical tags, and cleaned up internal links that had been pointing at the wrong variant
• Verified the fix in Search Console's crawl-stats and index-coverage reports over subsequent weeks to confirm crawl budget was actually reallocating, not just theoretically fixed

The outcome

• Crawl budget shifted measurably away from low-value faceted URLs and toward the category and product pages that actually mattered
• Previously-cannibalised category pages stopped competing with their own filtered variants and began ranking as the single, consolidated authority for their terms
• Consistent schema implementation improved rich-result eligibility across the catalog, not just on hand-picked flagship products
• The indexation policy was documented and handed off so future filter/facet additions wouldn't quietly recreate the same bloat