Google Says "Let a Trillion Subdomains Bloom"

> You follow best practices. You get torched for it. You are deciding how many employees to fire & if you should simply file bankruptcy and be done with it.

Anytime I read any SEO complaints against Google, I notice this huge sense of entitlement that the authors and commenters have, as if Google owes them something with respect to search engine rankings just because they might be paying for AdWords or some other Google service. You have the full right to complain if your ad campaigns do not work as advertised. But it's your own fault if you have to fire employees because Google changed their primary search algorithm which tanked your traffic. Sorry. Google does not owe you anything even if their statements are contradictory.

> And the truth is, this sort of shift is common, because as soon as Google openly recommends something people take it to the Nth degree & find ways to exploit it, which forces Google to change.

The author hits the nail on the head right here and even suggests the proper method:

> The only ways to get clarity from Google on issues of importance are to: * ignore what Google suggests & test what actually works

Don't try to read too much into how Google works from a comment by an employee here or an official blog post there. Just do what your site needs to function properly and most importantly offer something that people want.

Matt Cutts isn't claiming here that subdomains are better than subdirectories, just that they are treated differently (which should come as no great surprise -- "different things are treated differently" isn't exactly a shock.)

Suggesting to someone that they try something different when their first approach doesn't seem to be working out is hardly revolutionary or headline-worthy.

But somehow this short quote manages to get a whole blog post hung on it, with the eventual conclusion that "the average webmaster is told to "sit and spin"".

The author took the lazy way out, appealing to emotion and outrage, rather than actually doing the experiments to see how google differentiates between subdomains and subdirectories.

Also: All we know is that it has been close to a half-year since Panda has been implemented, and in spite of massive capital investments virtually nobody has recovered.

Well, of course.

Panda was aiming to lower the rankings of certain kinds of content-suppliers. These sites weren't supposed to be able to just make a few changes and recover; they were supposed to be permanently affected.

Panda wasn't an assault course that Google built just so that a few webmasters could test their skillz.

Furthermore, this is now a few SEO articles I've seen hit the frontpage today with very few people in the comments seeming to like the article. Considering HN has a very respectable pagerank of 6 and doesn't use nofollow tags, could it be that we're getting spammed for backlink juice?

If this is a matter of certain SEO-reliant websites being sent to the back of the line and not being able to figure out how to cut in front of everyone else again, well, good.

And the Google index is still plenty spammy, as we discussed just recently on HN. There was also an article about lead generation sites crowding out genuine local businesses for searches like "locksmith emergency portland". Google needs to be trying more changes, not less.

EDIT: oh, read some replies down the thread and feel the love for the SEO guys. I get it - heavy SEO and outright search engine spam at this point is more annoying than email spam ever was.

Here is a message for the SEOers: "Nobody wants to hear you complain. Nobody cares. Nobody sympathizes." I say we google-bomb that and make it the first result for "SEO". [That was a joke.]

A lot of the commenters seem to feel that Google doesn't owe webmasters anything. I disagree. Every time a SERPs page shows up with my site in the results, Google is selling advertising on top of my content. I gave them permission to use snippets of my content in return for sending whatever traffic they think should be sent my way. It's a mutually beneficial relationship (though they definitely hold most of the power), but it's certainly not a gift. We can both pull out of the deal at any time, but there's no reason to since we're both profiting from it.

I run sites to help college students find jobs and internships. One is called One Day, One Job and the other is One Day, One Internship. Every day I write about a new company. It's usually 400-600 words and I talk about what the company does, who they are, why they're interesting, and what kind of entry level jobs or internships they have. I get thank you notes from readers every day. I've helped put dozens of people into jobs they never would have found without my help. I get a ton of praise from college career services offices. There's certainly a consensus that my content is extremely high quality, useful, and something that Google would want searchers to find.

I've made SEO an important part of my strategy. I've optimized my pages, reworked my site architecture, and built a ton of high quality links. I've also done a lot of keyword research to try and better understand the psychology of a job seeker whose main tool for finding jobs is the Internet. I've tried a lot of other marketing techniques. I keep coming back to SEO. It's the absolute best way to reach job seekers. I've always followed Google's rules as closely as possible. I keep up with what Matt Cutts is saying and what Google is posting on their Webmaster Central blog. I'm not trying to game the search engines, I'm just trying to make the most of the opportunity that they offer to expand the reach of my content.

One Day, One Job got hit by Panda on February 24th. One Day, One Internship was never hit. Because the sites are nearly identical (including a lot of posts that are 90-95% shared content), it seemed weird that one would get hit and the other wouldn't. I figured out quickly that it had to be some sort of duplicate content issue. For a long time Google has said that duplicate content was not something that would be penalized--they'd just filter out anything that looks to be a duplicate. I found that One Day, One Job would rank for job related queries, and One Day, One Internship would rank for internship related queries. Rarely would they both show up on the first page for a given search term.

The reason that I built two separate sites was that I thought it was the best choice for my users. Job seekers and internship seekers are looking for slightly different things, so why not build separate sites and e-mail lists for them. I'm not trying to manipulate anything. I'm just trying to serve two mostly separate audiences with very similar content. Google assured me this wasn't a problem--and if it was they'd just show the user the most relevant result.

Google messed up with Panda. Even though One Day, One Job always gets the content first, has more and stronger incoming links, and has been around longer than One Day, One Internship, it got the penalty. One Day, One Internship is not the canonical source, but Google thinks it is.

So far I've been unable to fix this. I tried linking all duplicated posts on One Day, One Internship to the corresponding post on One Day, One Job to tell Google that it was the original source. I gave this over a month to set in, yet saw no results. Now I've used rel="canonical" to give Google a strong signal that One Day, One Job posts should be seen as canonical. It killed my One Day, One Internship traffic, and One Day, One Job has not seen any benefit yet. I'm waiting to see if it will eventually start working, but I'm not very hopeful.

My next step may be combining the two sites. I'd prefer not to, but it might make the most sense. I probably should have built the two as one site from the start, but I liked the idea of having two separate, but connected brands. That decisions shouldn't matter as much as it does now.

I strongly believe that Google's search results are worse without my content ranking as well as it used to. Obviously I'm biased, but I think most of my site's users would agree. My traffic on One Day, One Job is down 30-40% from where it was, and I've generated thousands of dollars less in revenue than I would have without the Panda penalty.

I don't doubt that Panda had an overall positive effect on the search ecosystem, but it really sucks for those who have worked hard to play by Google's rules. The repeated contradictions have caused me to follow paths that I may not have otherwise chosen.

Traffic from Google is extremely valuable to me. I'm willing to bend over backward to get it, but how do I do that when they are completely unclear about what they want (or even worse when they're very clear about what they want and then they change their mind)? I've been through their "guidance on building high quality sites" (see here: http://googlewebmastercentral.blogspot.com/2011/05/more-guid...). I know that I've built an extremely high quality site. Panda was a huge setback for my business. 5 months later I still haven't recovered, though I've tried a number of things that one would think would work. What do I do? I can't give up on Google. It's too important.

From Matt's quote, I just got the impression that sub-domains are an option and that GBot treats them as logically cohesive groupings of information, I didn't see anywhere that he said subdirectories were suddenly bad.

Then the author went on about best practices and getting burned for them -- he'd know better than I would if that has been true over the longer term -- I just didn't interpret Matt's comment to imply all that.

Some people must have benefited from Panda... I still find plenty of junk in my search results, but it makes me sick to think that if they hadn't rolled it out how impossible it would be to find good content at all.

I wonder what the logical conclusion of search engines/SEO and all this jazz will be in 10 years. If history is any indicator, we'll all swing back to some AOL-esque, walled garden, ad-driven existence where "Search" means nothing more than "find applicable ads"... I suppose that is somewhat true now.

Sympathy? Not really. I hope he stubs his toe so hard they have to amputate his foot.

Maybe the nextgen frameworks will incorporate an easy way to create subdomains via a script?