Good Rel.ations – A Beginners Guide to Rel Attributes
Estimated reading time: 8 minutes, 39 seconds
In recent times we’ve covered a lot of stories about HTML tags that include, canonical, author, publisher and many more for reasons that they are (relatively) new; topical, useful and beneficial for webmasters seeking to get more (and more relevant) traffic to more and more appropriately indexed pages.
We bandy about terms like “attributes”, “tags” assuming we’re talking to an initiated audience which is often the case. However not all our readers are search professionals, web developers or HTML proficient. Many are webmasters and small business owners who wear many hats. We thought it might be useful to collectively examine the more useful of these tags in their employ as rel attributes; particularly the ones that might directly boost our efficacy or solve a problem from an organic search perspective.
What are rel attributes?
Rel attributes pertain to links and describe the relationship of the linked target to the current. Links are a very strong contributing component to all major web search engine algorithms and within that there are many data properties and qualities pertaining to the links that point to a document, that are under consideration. In that way rel attributes assist search engines in getting additional meaning from those links attributed. Let’s examine which ones you need to know about; which ones solve common webmaster problems and which are more valuable in the meaning they give to search engines.
Need to know rel attributes
First out of the starter gates is rel=nofollow, which is actually a bit of a misnomer as it doesn’t always mean a search engine will not follow a link but is more of a concern from an SEO perspective as this attribute prevents the flow of equity/authority from the linking page to that linked target; such equity being PageRank which is a Google algorithm. If you’re unfamiliar with PageRank then this video by Google head of webspam is a good place to start.
It’s perfectly reasonable to wonder why we might need such an attribute as doesn’t this seem counter-intuitive to the spirit of the web and link based ranking? Perhaps so, but like most rel attributes rel=nofollow was introduced to solve a problem or at least disincentive a practise that was becoming problematic; that being the issue of comment spam.
You can use rel=nofollow when linking to sites that you may not trust, or perhaps if you have comment functionality on your blog you could consider setting this as a preference to discourage comment spam and because you can’t really legislate for the quality and validity of the sites’ of commenters. Indeed some blog platforms may have this as a default setting when comment facilities are on. It is your choice as a webmaster though if you choose to use this!
One situation that you would be wise to be aware of and consider the use of rel=nofollow is in commercial content that contains a link to any of your documents. According to Google guidelines paid links should use the rel=nofollow attribute to prevent the flow of PageRank. Whilst it’s not my place to police the internet I’d advise anyone to be aware of the major search engines’ stance on paid links and how the nofollow attribute should be used.
Problem Solving Rel Attributes
In 2009 the big three search engines announced support for this particular tag which solves a common problem for many webmasters and for search engines trying to “treat” this problem accordingly and that problem being duplicate or substantially similar content.
Duplicate or substantially similar content is a problem for search engines in that:
- It’s hard to work out the intended “original” or authority source
- Can lead to unnecessary pages (and use of resource) in the index
Its a problem for webmasters in that:
- It can lead to equity dilution if other sites link to duplicated pages
- It can mean that your pages are not getting indexed if a search engine thinks they are the same/not offering enough differentiation to warrant indexing
- It can be a huge waste of crawl budget say if for example your CMS generates query URLs
- It can cause ranking issues if a query or otherwise duplicate but peripheral URL (e.g. paginated URL) usurps the preferred target page
However duplicate or substantially similar URLs can occur for all kinds of common reasons such as:
- Legacy build, migration of otherwise CMS issues can often lead to you having multiple versions of a home URL e.g. http://anysite.com versus http://anysite.com/html
- Query URLs or URLs that contain appended parameters that might track (for example) how a visitor has arrived at a site
- Ecommerce sites may carry many, many versions of products that differ just a tiny bit from page to page e.g. 10mm wood screws, versus 12 mm wood screws
The canonical tag allows you to tell the search engines which of the set is the intended; the preferred above all others, however it can cause a lot of harm if implemented incorrectly or used to treat symptoms when perhaps another solution would treat cause. If you think that some of the issues described above are in play for your site and you want to dig into implementing and understanding rel=canonical then I’d recommend starting here, checking here (point 3) and also here.
In something of a follow-on from the previous; rel=prev (or rel=next) are perfect for paginated component URLs and help solve the duplicate (or substantially similar) content issues that can arise from pagination issues. I don’t want to re-invent the wheel here as we already have a fantastic post on State of Search which rounds up some of the leading experts speaking about this issue, usage of rel=prev and next and how this interplays with rel=canonical, so I’m simply going to tell you that you must read this piece.
Can be used to link to an alternate version such as a printer friendly page, however more topical and useful for international SEO is using rel=alternate in conjunction with hreflang.
Using this markup you can tell Google that you know there are multiple versions of what is mostly the same page BUT are targeting different countries. Either because they are in different languages or the same language but with slight colloquial differences e.g. American English or that there are functional differences such as currency and payment method. Essentially the pages may look the same or substantially similar but there is a good reason both should be indexed and shown to their intended and geographic market.
We’ve done a lot of work and research recently at my company on implementing this and some of this was quite a trial by fire, so do save some time by reading our case study there and using the tool we created to help generate hreflang sitemaps.
Hierarchical or Social Rel Attributes
I’m going to group the next set of useful attributes together as these aren’t so much structural or problem solving but are more useful in the additional context, meaning or hierarchy in the relationship they describe generally between the people behind the linking page and linked target. These attributes contribute to what is understood to be “the social graph” i.e. the sub-section of the link graph that describe social connections between the people referenced in about or as authors of content.
Rather obviously using rel=author names the author of a page, however not just nominally as a signature or credit, but more as a labelled entity, a “known quantity” and is verified using the Google Authorship program.
It stands to reason that credible and prolific authors should carry some validity and authority to their content and to do so currently requires an author to have a Google+ profile and set up a few things there.
This is interesting and useful on a number of levels but most pressing from an SEO perspective are the traffic and authority benefits. In terms of traffic when an author is marked-up as such and has their Google+ profile set-up correctly, the authors picture and name and picture will appear as a rich snippet in the search results.
Now this makes a result more attractive as it is visually distinct from surrounding results that don’t have the author markup, plus in addition this lends credibility and reassurance that the content is written by someone who is perfectly happy to have their professional reputation behind it. In addition if an author is particularly prolific in a sector then they may be known to a community and instantly recognised thus their content trusted (or perhaps not for that reason), but either way this helps get more, more relevant traffic.
Secondly there is much evidence based speculation that were just around the corner from an AuthorRank announcement from Google. Many industry professionals, myself included and also this great post on SEOMoz cite patents, clues and general movements towards adding (if not already) the authors “rank” as a contribution to a document authority. In fact tool providers such as Linkdex already include features that help you identify the most influential authors in your sector, who could be fantastic to approach for business and content partnerships that could benefit your organic search reputation. In fact our own Editor reviewed this very feature on Tuesday.
The other social rel attributes include rel=me, rel=contact and myriad interpersonal XFN rel attributes that describe the relationships between us all as web citizens. Social sites such as Twitter, Quora and Linkedin use these attributes as the linking attribute between pages that connect you to me and me to those other guys. In fact until very recently we used to be able to see exactly who Google knew us to be connected to in our “social circle”. Until they took it away to spoil our fun.
From an SEO perspective however it still makes sense to use these attributes when appropriate given the way they describe interconnectivity, add validity to individuals as entities or agents (if you read the SEOMoz post linked to above).
Now; all of the above aside we have still only scratched the surface on rel=attributes and even then only those most impacting to us as webmasters. If your interest is piqued there’s some learning resource here and here.