You might have seen our SEO challenge last month, where three of Silktide’s intrepid developers battled to get their website the highest in Google’s rankings for the obscure term: ‘frictionless owls’

Here’s the order they came when we finished the challenge:

1. frictionlessowls.com – David’s site had content rich with relevant keywords.
2. www.frictionlessowls.co.uk – Ali’s site had a few pages of unique content.
3. frictionless-owls.com – Oliver’s site with machine-generated nonsensical content.

There are so many factors that affect the order of rankings, so we decided to continue testing.

Was it the auto-generated content that caused Oliver to fail?

Originally we were very quick to assume that Oliver’s site (frictionless-owls.com) came last because it contained pages and pages of machine-generated content which didn’t make sense to any human. We assumed that Google is clever enough to pick out nonsense from real text which is why it didn’t score highly, but we later found out that might not be true.

As you can see, Oliver’s domain name was identical to David’s apart from the fact that it was hyphenated.

The negative effect of the hyphen

SEO experts SEOmoz recommend not to use hyphens in URLs, saying they “detract from credibility and can act as a spam indicator”. You might think this is unfair, but imagine if someone registered face-book.com. It instantly looks like it’s trying to be a Facebook rip-off and you’d be immediately suspicious.

So that got us wondering, could it be simply the hyphen in Oliver’s domain name that meant he failed? If so, it means he was doomed right from the start.

We decided to do a test, just for fun, and to see if we could learn anything else from our frictionless owls websites. So we swapped the content around between David and Oliver’s sites.

With David’s content on Oliver’s domain name, and Oliver’s content on David’s domain, we eagerly waited to see what Google would do. If they had a problem with Oliver’s spammy content, the site that’s currently in number one spot should plummet, while the bottom site should bounce to the top.

Remember, we still hadn’t linked to these sites, so there wasn’t any incoming links to affect the testing.

Playing the waiting game

We left these sites alone for a few weeks without an update, so when we swapped the content around it took the Googlebot a few weeks to get around to revisiting the sites.

When it did, we checked the rankings constantly, making sure that we searched using a server, and not from someone’s computer, to avoid the bias of Google’s personalised search.

Nothing changed

After 4 weeks, nothing changed in the rankings The domains were in exactly the same order. But, some other things had altered.

Google usually shows a few subpages, and bundles them all together in one search result with “show more results from…” etc. It was when these pages changed that we knew Google had indexed our swapped content correctly.

Google was also taking content from the page and showing it in the snippet, displaying some of Oliver’s nonsensical gibberish for the top result.

This was when I facepalmed.

Oliver’s machine-generated content was beating both mine and Ali’s content. Now is the time we need John Connor – the machines have won!

So it is obviously possible to fool Google with fake content. However, in our test the fake content was only up against two other websites. If you were to put Oliver’s machine-generated nonsense into a real SEO arena, it would probably rank poorly. In our SEO challenge though, we proved that Oliver was beaten by his hyphen.

The problem with hyphens

So it looks like hyphens in domain names do mean lower positing in Google’s SERP. If Oliver had the non-hyphenated domain name from the start, it’s likely that he’d have won our SEO challenge).

I’ve heard people advising not using hyphens in domain names for years, but always for the reason that it’s hard to understand when you read it aloud, or more difficult to remember. Not for SEO reasons. But now I’ve got proof that Google will favour non-hyphenated domains.

We know that our test isn’t entirely scientific, and it’s based on one isolated incident. But what I hope this shows is that choosing the right domain name is crucial to success. If you’re setting up a new site, here’s some classic advice for choosing the right domain name from the SEOmoz blog.