Monday, June 6, 2011

New Proof On Underscores vs Hyphens - What You Didn't Think Of

In our previous post supplying proof of the advantage of hyphens over underscores in urls, we supplied screenshots of sample searches in Google showing how at the time, they did indeed index and rank keywords differently depending which word separator you used.

The proof proffered was conclusive and obvious, demonstrating that since search terms separated by underscores are served differently to search terms separated by hyphens Google was indeed seeing the ranking signal in URLs just as differently. Unfortunately, there were many dissenters that couldn't overcome their blind love of underscores to see the truth (I would hate to sit on a jury with some people... urrgh!) but that's another story.

The Slight Change
To be honest I didn't notice when this change occurred, but there is now a partial difference in the way Google handles underscores. And here's some screen shots, which show the difference from our previous tests in 2009. You might not have remembered, but it was March 2009 when in this video Matt Cutts explicitly stated hyphens were treated differently and were to be used in URLs.

Well now in 2011 underscores are hyphens are being treated slightly closer towards equal in the search term box and rankings, but still miles away. They were always treated the same in the syntax-highlighting in the green URLs of the SERP (Search Engine Results Page) because that is calculated after calculating your search results, as our last post demonstrated.

Google still suggests I'm spelling it wrong, and shows me the same results as the hyphenated version. They are in fact, the same results as if I use a space and not a hyphen. No, of course no one searches for engine_optimisation - we're not saying that (thanks to all the drop kicks who commented that on our last post). This does illustrate though that SEO developers need to develop for what users do type in the search box, because that's the way Google ranks results. Searchers never type in engine_optimisation, so why would you put that in your URL???

So Why The Slight Change?
Well it seems that so many webmasters and developers were stuck in their underscore pants that Google actually had to cater to their manner of creating URLs, and that included using underscores. Despite the longstanding evidence to the contrary, there were some big name websites that simply refused to use hyphens, despite all the best SEO advice and proof in the world. (It's likely their developers were simply taking the easy road out because they didn't know what to do with a hyphen when it was presented in the URL and matched with a record in the database... at least, they reasoned, underscores weren't part of common sentences and didn't cause that problem!)

Why Was It Different In The First Place?
There are some 'words' that are joined by an underscore which mean something very specific and very different when not joined by an underscore. The example Matt Cutts used was FTP_BINARY and there are indeed many programming-related 'words' in which underscores feature prominently. Obviously this isn't a big concern for the majority of users, which is likely why the Google programming was finally allowed to change - to court the favor of the masses.

Nonetheless, as you can see in the screen shots, the websites that use hyphenated terms still rank highly even when searching for the term joined by an underscore. Only a few results which actually contain the underscore joined version in the title or use a URL with no space at all between 'engineoptimisation' rank first...

Why All The Confusion In Between?
In July 2007 there was an off-handed comment (notice that's 'off-handed', not 'off_handed') by Matt Cutts at Word Camp 2007 that was misinterpreted and echoed incorrectly around the internet, without any supporting evidence. Talk about your wildfire rumor! SERoundTable didn't publish a balancing post until a month later. Matt Cutts referred to the confusion on his blog and again to the rumors in the March 2009 video linked above. Personally, I had read the transcripts of Word Camp 2007 myself at the time and I can see why there was confusion... but only if you wanted to believe in underscores because that's what you'd been blindly using in hundreds of thousands of URLs until then. The facts were still clear nothing had changed back then... until recently, as shown by our screenshots above.

So Does It Matter Anymore?
Assuming you were brave, logical, and dare I say, empirical enough to see the light previously, you might now be left wondering what to do in a hazy shade of musk. Well apart from the fact that you're still way behind if you want to risk using underscores, you should stop thinking about search engine algorithms and start thinking about users. What is easier to read, underscores or hyphens? I'll withhold my opinion so you can decide for yourself, but I'm sure there will be just as many conflicting views as there ever was!


The fact is in most typefaces hyphens don't touch the adjacent characters, whereas underscores do (point for hyphens) but underscores are smaller space separators than underscores, which are usually wider (point for underscores). The fact that underscores sit lower than the typeface baseline is to some eyes a win (point for underscores), but to others a loss, since they seem to be confused by the reading eye with serifs making the text slower to read and recognise (point for hyphens).

And when you've decided what's easier to read (whichever you decide) forget that and take another look at the screen shots above. Do you want to target your SEO at a misspelt word that Google suggests the correct version of? Sure, if people type in the misspelt word for search. But they don't type in words joined with underscores! Do you want to be the site that isn't on page one of the underscore search page, or the hyphenated site that is still on page one of all three searches??

What Else You Might Not Have Thought Of
There's another SEO benefit to using the right word separator in URLs, and that is that not all links have distinct link text, but rather, the URL as the link text. This means that if correct word separators are used in the URL, they will pass contextual value to the webpage being linked to even if the link merely contains the URL. Our clients have been benefitting from this phenomenon for years now, when visitors copy and paste their URLs into a simple link without utilising anchor text (as many people do not), as their descriptive and accurately separated URLs contained relevant keywords which Google most definitely saw as anchor text.

So the question is, has Google been thorough enough to ensure that underscores are now treated as word separators in anchor text even though they're still not treated as equals in search and ranking? In our experience with Google's thoroughness, the answer would be 'most probably not' - don't count on it. Besides that, why would someone put underscores in anchor text? They don't, and that's why Google wouldn't cater to it. But they do put hyphens in text, and they are already counted as word separators.

But more to the point, is your love of the underscore worth the gamble in rankings?

Bookmark and Share