How should an online store handle filter and facet URLs?

Filters help shoppers but create a crawling problem. Every combination of size, colour, brand, price and sort order can become a separate URL. Google's documentation on faceted navigation says this causes overcrawling and slows the discovery of new, useful pages.

The options Google describes:

• robots.txt: disallow crawling of filter parameters you never want indexed. This saves crawling, but Google can't see canonical tags on blocked URLs.
• URL fragments: implement filters with fragments (the part after #), which Google doesn't crawl.
• rel=canonical: point filtered pages to the unfiltered category. This reduces crawling of those URLs gradually rather than immediately.

If some filtered pages should be indexed, Google recommends keeping filters in a consistent order, using standard separators and returning a 404 status for filter combinations with no results.

The commercial decision matters most. Use keyword research to find the few filter combinations with real demand, turn them into proper landing pages with unique titles, copy and internal links, and keep everything else out of the index.