Wednesday, June 17, 2009

Underscores vs Hyphens SEO Conclusive Proof

Once you've read this post, see the new proof about underscores and hyphens in June 2011.

I'm sick of having this argument with people who claim to know SEO and/or programming but demonstrate their ignorance. What really incensed me, was when Matt Cutts, head of anti-spam at Google gave definitive direction here (rare) on what was best practice, hyphens or underscores, and then in person at an interview, he was still misinterpreted by people who wanted to believe what they still wanted to believe!
Come on people! This isn't a religion, why the blindness?
So here's some conclusive proof I have developed in reviewing this issue with my long time friend and colleague James, who does SEO Sydney, Australia.

The test
1. Google: search-engine-optimisation
Results: 44,800,000. This is the big one.
This is clearly the correct search - Google news picks it up, as well as squillions of advertisers. I'm even offered a two word definition of what a 'search engine' is.
2. Google: search_engine_optimisation
Results: 12,500 with a better suggestion
Firstly you note Google offers you a better search, trying to break the single word into separate words, but then shows you your limited results - 12,500 misled minions who think underscores are better than dashes. Notice too that there are no advertisers bidding for this search...!

3. Now finally Google: engine_optimisation
Results: 590 with a better suggestion
This is the clincher. 590 sensless mentions of the word engine_optimisation. And notice again that there are no advertisers bidding for this search because Google isn't counting it as a search for two separate words, but one big ugly word with an underscore in the middle. Instead, I'm being offered a better search with a space before the word optimisation.
Do you get it now?
A space, like a hyphen (or dash) is a word break.
An underscore is not.
That's true grammatically, and programmatically.
Don't be fooled by the keyword highlighting... the bold on a SERP (Search Engine Results Page) is applied after the search is calculated and conducted. So the bottom result shown in the image above (the unfortunate Bond SEO) has the word engine_optimisation in bold, as though Google can see it separately, but this in only because the bold is applied after the search is calculated. We only have 590 results available. If Google were seeing the keyword as separate words, then we'd have millions of results available.
Your honour, I rest my case.
"What should I do?"
First of all, if you really want to evaluate how good an SEO company like the ones listed above are, make this the first question you ask them: do you recommend hyphens (dashes) or underscores in urls? Then get them to explain why. It will help you tell the SEO men from the boys.
What if my massive website already has underscores in the urls and (maybe because of the CMS) in the title tags or meta description?
A good SEO agency with both marketing AND technical programming skills is what you need. This is hard to find an agency that's good at both, that doesn't have two competing halves that don't cooperate or communicate, and that doesn't cost the earth!
We (Pioneer Websites) are a small family agency, solving the issue of competing halves. We probably favor technical over marketing while never neglecting either, and our client's track record speaks for that combination's success. If you're not sure how big a job your site is, send us an email and we'll gladly analyse your site and give you a recommendation.
It could be as easy (for us) as adding programming which corrects SEO underscore vs hyphen errors, picking up any lost SEO value and rechannelling it correctly, without you having to change your whole Content Management System or your content.
P.S. Yes I've spelt it the British way optimisation instead of optimization but the results are the same for either, as you can see, Google's SERP using alternate spelling for my results - but only where I used dashes and Google can see it's a separate word!!!

Bookmark and Share

10 Comments:

Anonymous Vaughn Anderson said...

But who really searches for terms like "search_engine_optimisation"? It's frequently argued that user's have a hard time even finding the underscore, let alone using it en masse.

I think normal people would never ever use an underscore in any search, unless they are looking for something with an underscore in the text (just like a hyphen). So including underscores in a search query as a comparison is really a straw man argument.

More importantly, a dash/hyphen is an real character in hyphenated words, where as an underscore has no inherent meaning, which fits the meaning of a space much better semantically.(grammer?)

Consider this search in both google and bing:
"dashes vs underscores"

Matt Cutt's statements are all fairly old, and don't hold up in real world search results. Consider how well Wikipedia does in google, and it's all underscores.

Recent online SEO rumblings (and plenty of testing) indicates that google doesn't even consider keywords in URLs. Making any argument regarding the content of the URL moot.

Also, as an application programmer I consider: underscore = implied whitespace, hyphen = word character.

Consider these two urls:
www.example.com/the-best-co-pilot.html
www.example.com/the_best_co-pilot.html

With hyphens you are stuck with a less "real world/normal person" search friendly url, simply because the hyphen actually has meaning in English. Where as the underscore has no meaning, and therefore is just a word delimiter.

Ultimately, search engines will change to return the best possible result to fit what a real world/normal person wants. And it appears this is already happening. Search engines already recognize the underscore as a valid (and I postulate, in the future preferred) word separator.

July 6, 2009 at 2:27 PM  
Anonymous Anonymous said...

This is obviously a thinly veiled advertisement for pioneer websites.... a mob that chooses to embed audio in their homepage and thus shouldn't be trusted with anything more technical than a bottle opener.

October 31, 2009 at 12:25 AM  
Blogger Pioneer Websites (Paul) said...

Thank you for your bold comment, 'Anonymous'!

Perhaps you could tender some actual reasoning behind your claim that the article's reasoning is invalid?


P.S. Glad you liked our little audio experiment, you're the first to comment negatively and it's only been a couple of years...

November 2, 2009 at 9:13 AM  
Blogger Pioneer Websites (Paul) said...

@Vaughn

You make a good point:

"But who really searches for terms like 'search_engine_optimisation'? It's frequently argued that user's have a hard time even finding the underscore"

No one does.

But that's exactly why it doesn't assist your SEO to use the underscore in file names, because that's not what people are using in searches. They're not using hyphens either, but at least Google ignores them in searches. It doesn't ignore underscores, and that's what we proved above.

You also made another good point - underscores read more intuitively than hyphens.

Unfortunately, underscores also serve a very specific purpose in some searches, as Matt Cutts once pointed out, using a specific (but nerdy) example.

We weren't discussing how Google *should* index the use of underscore, rather, we're looking for indications as to how they *do* index the use of underscores.

Your postulation may be fulfilled in the future, however currently the best word separator is a hyphen. If you're correct and they are later given equal rights, then having a site using hyphens is not going to be any disadvantage, yet using underscores before such a speculated time will be.

November 2, 2009 at 9:24 AM  
Blogger PoetryWeb said...

You do know that you misspelled "optimization", yes? In each of your tests, Google tried to correct you with the correct spelling, but you apparently did not pick up on that clue.

No one will search on search_engine_optimisation, spelled your way or the right way. That is not the test you should be conducting. You should search on search engine optimization and see how many results have _ in the links. THAT will give you the true search result set.

Regarding your "supporting evidence". Looks like you read up on Matt Cutts…. In 2005:

http://www.mattcutts.com/blog/dashes-vs-underscores/

You linked to the Guest post (IE: Matt did not write it!): "Vanessa Fox on Organic Site Review session" from APRIL 21, 2006

He denounces both blogs himself, in 2007:

Underscores are now word separators, proclaims Google
http://news.cnet.com/8301-10784_3-9748779-7.html


Please, before you put yourself as an expert, please become an expert.

Thanks,
Jenn

November 10, 2009 at 6:54 AM  
Blogger Pioneer Websites (Paul) said...

@PoetyWeb:

We haven't asked to be knighted as expert - only that you consider the evidence of underscore behaviour on its own merit.

You're only comment on the reasoning presented was that no one searches for the example phrase… which has nothing to do with observing the indexed behaviour of underscores.

Using the inurl: operator proves the exact same point we were making - if you had've taken your own advice you would have noticed that...

Here is an example for you, as you've recommended:

26,700 results for inurl:search_engine_optimisation

731 results for inurl:search_engine_

Google isn't breaking out the characters search_engine_ from within the characters search_engine_optimisation or else there'd be MORE results, not LESS... let alone separating the words 'search' and 'engine'.

If you perform the second search, you'll see most of the results are from URLs that use a valid word break character after the second underscore... namely an opening parentheses, like this:

en.wikipedia.org/wiki/Search_engine_(disambiguation)

Case closed. Again.
Underscores are not treated as word breaks by Google in urls.


BTW, yes, Vanessa Fox wrote one article on Matt's blog and he edited it in two places - we were not being misleading when saying this was information published by him. Are you saying it held incorrect views and he still approved it? On his own mattcutts.com blog? I didn't think so.

Perhaps you could you kindly show us where Matt actually denounces both underscore vs hyphen blog posts as you've claimed, instead of pointing to a summary by someone else of his verbal quote? There was a lot of confusion about Matt's comments at that event, and he has addressed them since.

It doesn't matter, the facts still speak for themselves. Thank you for suggesting we find the extra proof by searching within URLs only.


P.S. Re: spelling - You obviously didn't even read my entire post. I referred to my use of the British English spelling as this is a .com.au blog written in Sydney.

There are other countries outside of the U.S. - if by nothing else, you can tell from the ccTLD in the address bar of your browser. It is suffixed with a ".au" for Australia.

November 11, 2009 at 12:33 AM  
Anonymous Anonymous said...

I did a quick Google search of "Search Engine Optimization" (without the quotes). That's the way a human being would type it in; a human looks for real words.

The very first result is a URL using underscores. There are 3 results with underscores in the URL on page 1. There is only 1 result with dashes in the URL on page 1.

Conclusions:

1) human beings search for real words.
2) Google does not prefer dashes to underscores when returning URIs to fit real-world, real-word searches that human beings type in.

May 18, 2010 at 3:15 AM  
Blogger Pioneer Websites (Paul) said...

Thank you for contributing your conclusions. This is what you have actually proved:

1) Since human beings search for real words, you should separate the real words with a character Google recognises to separate them, not an underscore.

2) URIs are NOT the biggest determining factor in ranking. We agree. Everyone knows that - the biggest factor is inlinks.

Assuming your search was in google.com (given your American spelling of optimisation) you will note the #1 spot with underscores has 42,613 inlinks TO THAT PAGE ALONE according to Yahoo site search.

The #5 spot is the first with hyphens, and it has only 442 inlinks.
That's 1.03% of the #1 spot.
It's only got 37.3% of the inlinks of the #4 spot.

The #6 spot has 6,125 inlinks to the page, but ranks BELOW the #5 with 442 inlinks. (In fact, the hyphenated #5 only has 4,976 links to the ENTIRE SITE!)

Conclusions:

1) Inlinks is the greatest influence on ranking in Google.
2) Hyphenated URIs improve your rank, even if you have less inlinks than your competitor.

May 18, 2010 at 7:16 AM  
Anonymous silk ties australia said...

I think normal people would never ever use an underscore in any search, unless they are looking for something with an underscore in the text (just like a hyphen). So including underscores in a search query as a comparison is really a straw man argument.

December 12, 2012 at 5:33 PM  
Blogger Pioneer Websites (Paul) said...

Dear 'silk ties australia'.

Do you think anyone will trust your opinion, since you spam blogs like this one for nofollow links to your store? You have demonstrated your lack of credibility immediately.

You've also demonstrated your ignorance in your spam comment: there is no argument that common searches use underscores. The demonstrations above point out how Google differentiates between underscores and hyphens.

December 12, 2012 at 6:58 PM  

Post a Comment

Questions and feedback are gladly accepted. We will do whatever we can to answer your question or direct you to the answer.

We personally read and approve every comment. Don't waste your time or ours.

Subscribe to Post Comments [Atom]

<< Home