URL Re-writing Best-Practice
Introduction
A while back, I wrote this article about how Google recommended NOT using URL re-writing.
The Google article provoked a lot of debate and certainly managed to infuriate a lot of SEO experts.
Google's advice seemed pretty conclusive after that: don't use URL re-writing. However, Google released their updated SEO guide later, and the waters muddied. Page 8 had the following on URLs:
If your URL contains relevant words, this provides users and search engines with more information about the page than an ID or oddly named parameter would
Now this isn't directly saying that Google indexes the words in URLs. The example is showing the URL being displayed in search results for a user to see, so it might be misconstrued. It does however re-open the argument on whether we should add keywords in URLs for SEO, and that inevitably leads to URL re-writing.
I've a number of greenfield projects starting soon, and I've been mulling over whether to use re-writing again. The unresolved issue of Google indexing words in URLs will obviously play an important role in any decision, but the usability argument is possibly more strong. There doesn't seem to be much in the way of research or best-practices, so I've written this article to start a discussion on it.
As I'll explain below, I'd like things to be as evidence-based as possible, with links to relevant authorities. Most of the tips are my own thoughts on the matter, but I've tried to back them up with reasoning.
I welcome comments with evidence or reasoning, whether they back up or argue against what I've written. That way, we can put together a strong set of guidelines which will hopefully benefit everyone.
URL rewriting is something I've been building on for a while now, and your article asks some interesting questions. I've never really considered some of the SEO angles you've raised, its always been about more friendly recognisable urls.
There are certainly a few pointers to take away and experiment with, thanks.
There are various ways hackers can tell what server-side technology is being used, even if you've changed the file extension. You should always keep your server fully updated/patched, so that any known vulnerabilities are dealt with. Relying on just changing the extension will give you a false sense of security.
1. The user knows the domain they want.
2. The user wants the most relevant / meaningful content.
In #1, the user will type the domain (which doesn't require URL rewriting).
In #2, the user should probably just use Google Search as it will return potentially more relevant information.
Anyway, just my thoughts on that one particular aspect of URL usability.
One part in your piece is not right according to SeoMoz SRF's 2011:
http://www.seomoz.org/article/search-ranking-facto...
Rel connical usage is a negative metric to use. My advise is to NOT use canonical URLs at all and work with URL rewrite or 301 redirects in the software code instead.
--
My conclusion: Just use URL rewriting for the usability.. but be sure the lead developer knows how to implement it so it is done the right way.
Thanks Gareth, good piece
http://twitter.com/@isgoed
As they've made the raw data available, I'll see if I can use it to add statistical data to the rules above.
Regarding the canonical urls, I think they may have misunderstood how it works. Their results spreadsheet gives the following description:
"rel=canonical : -1 = present, and points to this url, 0 = absent, 1 = present and points to another page"
From my reading of it, they are giving a negative score when canonical url is working correctly. I've sent them some feedback, so hopefully we'll get clarification.
Thanks for your feedback. It is my experience (and of other people around me) that people always Google for what they want. Mostly the simple web users (but developers like me too) just open up their browser and start typing in the google search field (on their startpage or in the browser's search box). Besides this, I think we can agree people don't like typing. I can' t find scientific evidence this fast for this opinion. I think i can get the evidence myself, by making a custom report in Google Analytics to see how much returning visitors come from search engines and how many type the url directly. "Direct traffic" isn't what it likes.. people can bookmark a page.. it is direct traffic, but still, they are not typing the url.
Regarding the conical urls: Where did you find this result spreadsheet? I didn't notice it myself, so i can't give my opinion about this. If it is like you say, i am curious on their feedback.
It would be interesting to see some data about keyboard usage, although I don't know if google analytics will have those figures.
You can get the results spreadsheet and a sub-set of the raw data in the methodology section here: http://www.seomoz.org/article/search-ranking-facto...
I still need to understand the Spearman Correlations better, but I did some quick analysis of the raw data subset using an online calculator. If you score the correct rel canonical results 1 instead of -1, and set the incorrect canonical results to 0, you get a slightly positive correlation score of 0.0135
These are rough calculations, on a small subset of the data, so it's probably best not to draw any conclusions about rel canonical yet. If seo moz send me the full raw dataset, I'll try and look at things properly.
<a href ="http://www.satvatinfosol.com/online_examination.ph... exam </a>