Search basics: URL structure vs Information ArchitectureWill Critchlow
I’ve recently noticed some confusion around the industry on the differences between URL structures and Information Architecture (IA). I thought it was worth clarifying a few points and giving you all some language that is useful when talking about the differences.
Pre-requisites – if you aren’t familiar with the following elements, it’s worth reading these primers before you dive in deeper here:
- What is a URL?
- What are the SEO considerations for URLs?
- Google’s guidelines for URLs
- What is Information Architecture?
- That article focuses specifically on IA considerations for SEO, for a broader overview of IA more generally, this is a great resource
The specific thing I want to clarify is the differences between decisions about the path in your URLs and decisions about your IA as this is where I often see a ton of confusion.
Decisions about URL structures and decisions about the IA of your website both involve questions about grouping and hierarchies of pages. For example:
- URL: should the path of an individual product be:
- IA: how should we group our product pages and link between them:
- Should there be a link “up” to the parent category?
- How many “levels” of sub-category page types should there be?
- How do we link between sibling products in the same (sub-)category?
- How many products can we reach in (e.g.) 3 clicks from the homepage?
- How should we handle facets?
The fact that both concern groupings and hierarchies has led too often to people misinterpreting IA questions as URL questions.
From an SEO perspective, most of the grouping and hierarchy questions we care about are questions about which pages should exist (e.g. should there be an indexable page for “red men’s shoes above a size 11”) and how should our pages be linked together (both from a crawling perspective – thinking about considerations like click depth, and from a ranking perspective – thinking about considerations like internal link equity).
Unfortunately, I’ve too often seen these IA questions expressed as URL considerations, and this can lead to advice that is less effective than it should be. For many of these IA questions, you can come down on either side of the IA decision with either URL structure:
- You can choose to have a (sub-)category page type without necessarily having the (sub-)category appear in the URL as a keyword or as a folder (and indeed, there are times that this is a good idea if products can be in multiple categories or if they often move category)
- You can choose to link “up” the hierarchy or “across” to sibling products with or without those link targets sharing elements of their paths (e.g. a product page at /product-slug can link to a parent page at /category even if it doesn’t have a URL of /category/product-slug)
In general, the IA considerations are more important than the URL considerations, and you should focus on the Information Architecture with higher priority. It’s IA that governs the flow of internal link equity (PageRank) and also that governs crawlability and discoverability of different pages and page types. In general when we talk about pages being “higher in the IA” or “closer to the homepage” we mean in click-depth rather than folder structure. You can’t fix IA issues with URL changes alone. For whatever IA decisions you make, you can then make decisions around how to structure the paths for your pages’ URLs to make the best trade-off you can between the constant tensions:
- It can be good to have appropriate keywords in the path (for users and search engines(*))
- Human-readable paths are helpful (and structure can help with reporting)
- BUT shorter paths are generally better than longer
(*) see “the importance of keywords in URLs” below
Probably the only real constraint that paths create for IA is that if you go down the path of having nested folders, that will generally imply the existence of pages at each level of folder. In other words, a page at /level1/level2/level3 implies that /level1/level2 will also exist as a real page (as much for users as anything else).
Summary of IA vs. URLs
Information Architecture decisions for search performance focus on:
- What (kinds of) pages should exist on my site?
- How should our pages and page types link to one another?
You may choose to group pages of the same page type together by, for example, placing them in a folder, but this is an independent decision about URL structure. In general, URL structure decisions are less important than IA decisions.
How important are keywords in URLs?
At MnSummit, my colleague Rob heard Google representative John Mueller say that there was no SEO need to translate URLs for foreign language sites. This surprised me, because (unless Google is already translating all inputs and outputs) this implies that keywords in the path make no difference in search either. I would have thought that all else being equal a page called /shoes/red would outperform /products/12512 for a whole variety of reasons.
So: I’m inclined to add this to the list of things that Google says are true that may be technically true, if read narrowly enough, but are unhelpful in the real world. My most charitable reading is that John is saying something like “Google does not have a specific element of the algorithm that checks language in page paths”.
So, although the primary focus of this post has been IA considerations I do think that it’s sensible to have some element of descriptive keyword in your URLs because although we can’t be certain it’s an explicit ranking factor:
- Above everything else, it’s good for usability
- Google says this explicitly in their guidelines
- It is a keyword-relevance signal of some kind (however weak)
- Google’s guidelines only argument is the usability one above, but that whole guideline section is explicitly about performance in Google, with primarily technical advice, and it seems reasonable to me to believe that they are saying we prefer (and rank better) pages like this because users prefer them
- We do know explicitly that URL, path, and filename are explicit keyword signals for some file types
- It is more likely to result in relevant anchor text in external links
- A point made even in Google’s own SEO guide (in a section entitled “Simple URLs convey content information” which supports the arguments above as well)
- I would expect a better click-through rate from the search results when it does rank
And for sure try to have URLs in the correct target language. Regardless of whether you agree with me or John Mueller about the SEO benefit, I think we both agree that your users would prefer URLs in their own language.
A note on changing URLs and moving content
There are always risks to moving content – even with well-implemented redirects and no mistakes – and so you should only undertake URL changes with care. In general, IA changes are more reversible as things are more likely to go back to how they were if you undo the change while the nature of a 301 (“permanent”) redirect is that it should signal that things are not going to change back.
For that reason, while we would often recommend moving from dynamic URLs with a bunch of parameters in them to cleaner URLs, and may recommend moving from impenetrable URLs to more readable ones, it will generally be hard to justify a move from reasonably-good URLs to arguably-better. Do your own risk assessment, and proceed with caution!