Context
Halden & Pike sell mid-market furniture online across the UK and Ireland, with roughly 4,000 SKUs and a catalogue that turns over heavily each season. Brand search was healthy. Non-brand had been flat for two years despite steady content spend.
The problem
The faceted navigation was generating an effectively unbounded URL space. Every combination of colour, material, price band and sort order produced a crawlable, indexable URL, and internal links pointed at them freely. Search Console showed 1.2 million discovered URLs against about 6,000 that mattered.
The consequence was not a penalty. It was arithmetic: crawl budget was being spent almost entirely on near-duplicate facet pages, so new season products were taking six to nine weeks to be indexed at all. By the time they ranked, the season had moved on.
What I did
First I established which facet combinations had genuine search demand — colour and material did, price band and sort order did not. That distinction drove everything else.
Valuable facets kept clean, static, indexable URLs with their own copy and internal links. Everything else moved behind parameters excluded in robots.txt, with canonical tags pointing back to the parent category and facet links rendered so they no longer passed internal PageRank.
Then I rebuilt internal linking from the category templates so new products were linked from the category page on publish rather than waiting for a sitemap ping.
The results
Indexation latency for new products fell from six to nine weeks to under five days. Non-brand visibility followed over the next two quarters, and because the recovered traffic landed on commercial category pages rather than blog content, it converted at close to the site average.
What I would do differently
I would have instrumented log file analysis in week one rather than week five. I inferred the crawl waste from Search Console coverage data, which was correct but took longer to prove to the client than a week of server logs would have.