Pagination and filter pages generate thousands of addresses

Aug 29
3 min read

Updated: 4 days ago

Introduction

A shop has two hundred products across eight categories. Each category can be sorted four ways, filtered by three attributes, and split across several pages of results, and every combination produces a distinct address.

The catalogue is small and the number of addresses is enormous. Most of them show the same products in a different order, and left alone they consume crawling effort, compete with each other and bury the pages that were meant to be found. The decisions here are few, and they are worth making deliberately rather than inheriting whatever the platform does by default.

1. Pagination and filter pages multiply faster than anybody expects

See the scale of it before deciding anything.

Every sort option multiplies the total, every filter multiplies it again, and combinations multiply it further still. A catalogue that a person would describe as small can present tens of thousands of distinct addresses to anything that crawls it.

2. Decide which of these pages should be found at all

The decision everything else follows from.

A filtered view that matches a real search — a size, a colour, a material people genuinely look for — may deserve to be indexed and even to have its own written introduction. A sort by price descending never does, and nor does the ninth page of a listing.

3. Keep the main category pages as the priority

They are what the effort should protect.

Category pages are usually the strongest commercial pages an online shop has. Everything else in this area should be arranged so those pages get the attention rather than competing with a hundred filtered variations of themselves.

4. Use canonical tags on the variations

The standard treatment.

Filtered and sorted views can name the main category as their canonical, which consolidates the signals while leaving the pages working normally for anybody using them. This is precisely the case the tag exists for.

5. Let paginated pages stand on their own

The convention has changed here.

Later pages of a listing should generally reference themselves rather than pointing back at the first page, so their products remain discoverable. Pointing every page at page one can hide most of a catalogue from search entirely.

6. Make sure paginated links can be followed

Discovery depends on it.

If moving between pages requires clicking a control that loads results by script, a crawler may never reach anything past the first screen. Ordinary links between pages, present in the page itself, keep the whole catalogue reachable.

7. Block the combinations nobody should crawl

Where crawl budget is genuinely at stake.

Parameter-heavy addresses generated by multiple simultaneous filters can be excluded so that crawling concentrates on real pages. On a large catalogue this matters; on a site of thirty products it is not worth the configuration risk.

8. Watch what internal search generates

An overlooked source of addresses.

Every search somebody types on your site may produce an indexable results page, and those pages accumulate without limit. They should generally not be indexed at all, and most platforms offer a setting for exactly this.

9. Check what is actually indexed before changing anything

Evidence rather than assumption.

Your search console will show how many addresses are known and how many are indexed, and a large gap in this area is the symptom to act on. Restructuring a catalogue that has no problem is a substantial risk taken for nothing.

Add real content to the filtered pages you decide to keep. A filtered view with a written introduction, describing what that selection suits and who buys it, stops being a near-duplicate and becomes a page that can compete on its own terms.

Conclusion

Understand that sorting, filtering and pagination multiply a small catalogue into an enormous number of addresses.

Decide deliberately which of those views deserve to be found, protect the main category pages as the priority, apply canonical tags to sorted and filtered variations, let paginated pages reference themselves so deeper products stay discoverable, make sure the links between pages can actually be followed, block only the parameter combinations where crawl budget genuinely matters, keep internal search results out of the index, check what is actually indexed before restructuring anything, and write real content for any filtered page you decide to keep.