Yeah, this is one of the most frustrating things about Google. They're opaque to reduce gaming of the system, but that same quality makes it easy to accidentally choose something Google doesn't like. I'm going to stop short of calling Matt Cutts a liar, but his word isn't always right. Moz seems to tell the truth as they discover it. The only interactions I've had with Google essentially lead to "why don't you spend more money on Adwords!" Well, I won't go into a full rant... hope your traffic recovers.
To paraphrase what b1twise wrote below I would say do not do as Google says when common sense, lots and lots of case studies, and practitioner consensus say otherwise.
Sure...you can move it to a subdomain and if all goes well, great. If not you can cite a Matt Cutts video to say who could have known: https://www.youtube.com/watch?v=_MswMYk05tk
If the organic traffic matters then trust people who are actually affected (http://moz.com/community/q/moz-s-official-stance-on-subdomai...) and figure out another (yes, much more involved) way to deal like reverse proxying if you have other demands to contend with.
My advice based on recent experience: give it time. We made domain changes and rankings tanked, only to recover. We made a major design change and rankings dived, only to recover. Give Google some time to assess.
Yeah, major design changes are painful. Google Webmaster Tools is currently warning me that we have over 100 404s on various URLs, because those pages simply got removed completely.
Actually most of those 100 are the same few pages, but Google indexed it a metric shit-ton of times, and then added each of them to the index, and THEN it penalised us for "spamming."
I came in, removed it completely, and set up a robots.txt so it couldn't happen again. But the penalties and 404s will take up to a year to completely clear out of Google's end.
Funnily enough Bing hasn't penalised the site at all, and we're the number 3 rank for relevant search terms on there (2nd page on Google).
How does having duplicate content on your domain give you a spamming penty?
And for a site thats been a round for a while only having 100 404's is low - most site owners are v bad at tidying up old defunct urls and using script kiddie platforms like joomla and magento dont help.
It sounds like the content got removed completely, so there should not be a URL (either old or new) for that content.
It's true that they should have used 301s or rel=canonical or something in the past, but that also sounds like the least of their past problems, if I'm reading the parent comment right. Right now, I don't think they'd be much better off with 301s to a single URI that 410s instead of just directly serving 404s for everything, even though the 301 to a 410 is more right in some abstract sense.
I just read the article and looked at the image and it's still really unclear. It looks like you lost all your traffic immediately prior to switching to the subdomain and nothing in the article said otherwise.
It's not clear exactly how that graph works; if the indicated bullet point represents "total pages views in July" then it would seem to show a decrease in the period starting immediately after the move.
there are many different types of Google penalties. It doesn't have to be a complete loss of traffic. Sometimes you get bounced back 1 page, other times 3, other times 99 it varies based on the 'offense.'
You are correct in stating this was only a slight drop in impressions. In this case I would say this is Googles fuck up. The algorithm is not mature or smart enough to see that the blog was moved.
At the risk of being slightly off-topic, when looking at the phrase "for better maintainability and performance with GitHub Pages' CDN and future-proofing in case we wanted to switch to a different platform down the road", I do not see a compelling reason for the work expenditure. Unsurprisingly, in solving problems they didn't actually have, they managed to create an entirely new set of problems that they wouldn't have otherwise had. (And the problems they created look to have teeth to boot.) To me, the lesson here is not about SEO, it's about effective management.
As someone who has worked on the tech side of sites powered entirely by organic SEO, I've definitely seen the same penalties occur when cosmetically "cleaning up" URLs or code.
In the author's case, they moved to sub domain for easier maintenance. The article ended with a transparent way of keeping the old URLs and correctly improving the tech behind the scenes.
Be careful of changes, even if endorsed by Google, when things are working. More often than not, there'll be a mistake, you send the wrong signal to Google, and you take a hit. Only do it, IMO, when there's nowhere to go but up :)
> More often than not, there'll be a mistake, you send the wrong signal to Google, and you take a hit.
There's probably a publication bias here. When people change things and traffic goes up, they don't run to tell the world about it.
Though anecdotally, I have a lot of reference-type content that has been bubbling up over the years seemingly solely because it hasn't changed at all in 7 years. The big players like to refresh their design and layout every few years.
In my experience, we've had more bad than good with URL changes.
The interesting part is that often a "URL cleanup" would result in a quick "boost" in traffic/rankings, but would continually slump, often even unrecoverable after reverting.
This is true. If it works dont change it. I had many experiences with google and penalties. Now im afraid even to add a banner to a site because google might not like it.
Thanks for this case study - super important. We use Hubspot and I brought up this issue with them as we want to add a blog to our site using Hubspot, and the only option is to use a subdomain. They pointed me to that Cutts video and said they have the Hubspot blog on their subdomain and it's no problem. But in their example I'm pretty sure the Hubspot blog does well in spite of being on a blog.hubspot.com subdomain and not "no matter where it is". I would trust Rand on this one over Cutts as well.
This is so timely for us. We are thinking of moving ionicframework/blog to blog.ionicframework.com/, but I was concerned about a possible SEO impact. The nginx proxy did cross my mind, though doing the permalink mapping is going to be annoying. Oh well, lesson learned! Thank you for sharing.
This is pretty standard advice. You want your blog on the same domain. Subdomains are generally treated not exactly the same as content on the same domain.
There are a lot of very good reasons to have your blog on the same domain and relatively few to have it on a subdomain. The extra traffic outweighs any benefit to putting it on a subdomain.
I am a complete noob regarding SEO. My naive interpretation of the drop in rating is that it takes time for Google to rebuild the index. It needs not only the content on your website, but links from other websites and people's clicking habit to determine the ranking.
I think he's referring to domain authority (PageRank) and not just indexing. GoogleBot can hit your site several times a day and index things quickly. Building your domain authority on the other hand and connecting all the dots (like your back-link profile) takes a long time.
what's up with all this SEO bullshit on HN in the last few days? and bullshit it is. let's start, i will try to not let this sound too much like a rant, but i can guarantee nothing.
let's do a wittgenstein. do not use the word penalty. a penalty is when you look into google webmaster tools (of all your protocol & subdomains & domains variation and you have a message, that you have a penalty.
go on, look into
https://iwantmyname.com
http://iwantmyname.com
https://www.iwantmyname.
http://www.iwantmyname.com
http://blog.iwantmyname.com
.. other subdomains ..
nothing there? then don't use the world penalty, it has no meaning.
this and your mention in your post of the worst bloodsucking (a.k.a. toolselling) "seo" publication ever leads me to the first issue: you are reading SEO blogs! stop it. do not reed SEO blogs, ever. you will reduce your understanding of google as soon as you start reading SEO blogs. there are 400+ specs and recommendations directly by google of what you have to do to perform well in google, read them first.
the issue at hand is: you got traffic for your blogpost content before the move, then you moved it to another protocol + subdomain combination, you got less traffic, with no positive trend. this was not the desired traffic behaviour.
this was the issue.
your hypothesis is that the traffic drop has to do with the move as the is a strong timely correlation. sound sensemaking.
first of all, be honest in what you did: you
* from a quick view i would say your blog pages are majority of your pages
* you changed the URL of the majority of your pages
* you changed the subdomain of the majority of your pages
* you changed the protocol of the majority of your pages
additionally the minority of the pages has only a very poor interlinking to the majority of the pages (only the blog start page is ever linked)
additionally the blog content seems to be high quality content, while on the subdomain-less site you seem to have lots of very poor content landing pages (duplicate with a hint of text templates) targeting the different domains suffixes https://www.google.com/search?q=site%3Aiwantmyname.com+Easy+...
google now has two statements
* I) a priori: another protocol/subdomain == another domain independet webproperty
* II) google can deal with webproperties split over differnt domains / protocolls if it can determine that they belong together.
the big question is now - did google see the new subdomains as part of the iwantmyname.com webproperty.
why is this, well basically your blog start page link points to the blog start page and has the underlying text "Domain blog"
your titles are not as recommended by google (see: https://support.google.com/webmasters/answer/35624?hl=en#3 section "brand your titles")
the title issue should have been a major warning. we could probably find much more if we would have google webmaster tools access to all the subdomains (protocols) and subdirectories.
but the biggest warning was the "we did not regain traffic" issue. but - based on your blog post - the thing you did was, you waited? and well, here we are with the bullshit and SEO blogs topic again where they quite often state, that it "can take a while" whereby they mean an undefined timeframe of 3 to 4 months plus. this is evil. if you do a change and you want to see a positive impact it does not take longer than 2 weeks to see a positive or negative trend. anybody who says something else is just lazy. (note: additionally i could not find any sitemap.xml which well indeed slows down the whole domain migration process, this should have been fixed first and is recommended by google again and again for URL changes and domain moves)
so what is the point of this rant:
* do not read SEO blogs
* do not read SEO blogs
* read google specs (in your case: titles, sitemaps, site moves, ..)
* don't use the word penalty
and maybe, just maybe: if you see a traffic drop and you do not regain your traffic, ask a professional SEO before moving your URLs around again and again. even the worst SEO - after selling you a bloodsucking tool - would have fixed the title, interlinking and probably sitemap issue.
or post a thread on the google webmaster forum, actually they are quite helpful there.
This is complete bullshit. Not reading SEO blogs is a TERRIBLE bit of advice. Read everything, and read it with a pinch of salt (Google's info more than most).
well, if you prioritize googles information on top of the other stuff, then there won't be an issue, as people would never find enough time to reach the other stuff, as googles information on the topic is very, very extensive.
sadly a lot of people are not very good at prioritising information intake, that is why i stick with my "do not read SEO blogs" recommendation.
the logic behind this is actually quite simple
let's say: x% of everything written in Z is bullshit (incl. deprecated or no longer valid).
let's agree that if Z is "seo blogs" then x is higher compared to the case where Z is "google recommendation", then it is always a better investment to read an item of x if Z is "google recommendation".
i would place the values of x something like this
if Z is "seo blogs" then x is 80
if Z is "google recommendation" then x is 20
>and read it with a pinch of salt
if we substitute "with a pinch of salt" with "flamethrower" then yes.
Can't agree with that purely for the fact that Google is in the game of providing sly information. You think that HTTPS ranking factor blog didn't have ulterior motives? You think that the disavow tool info isn't just crowdsourcing spam info?
Simple rule for SEO or any online marketing is 'always test for yourself.' be it Google's info you're working off or Search Engine Journals. Blogs can provide a good starting point for non-professional SEOs who don't have the time or resources to do external domain tests. I've been in the game over 7 years now, Moz has always been a good resource.
google has 400+ recommendations (specs, recommendations, posts) about what you have to do to perform well in google. without even counting schema.org or other initiatives.
(google) analytics is an unbelievable powerful tool that has at some companies whole teams dedicated to it
google webmaster tools is a unbelievable powerful tool that tells you 80% of everything that is wrong with your site. you can easily make it your half time job to use this tool to the fullest.
google is part of the triangle: your webproperty, the users, google search
your ressources are better invested into these three areas (analytics, webmaster tools, google recommendations) than in a third party with its own interest i.e.: selling tools
You say that as if Google don't have their own interests: encouraging more people to spend on Adwords, sending more clicks to sites using Adsense, sending more clicks to their own properties, encouraging more people to use Google Analytics.
Google don't want to send traffic to your site, they want to sell ads.
It has. Their software is not bad to be honest, a lot of SEOs think Moz is the be all end all but they aren't. There is a ton of hate for them as well which is unjust considering how much they've contributed to the industry.
Can't argue with this really. We're viewed as the used car salesman of the Internet. In fact the majority of us who work for larger brands and don't engage in black hat stuff are really just multi-disciplined online marketers. SEO is dead, long live SEO.
in the public discourse every poor performance is a "penalty" so this word has lost all it's meaning.
if your site is preforming poor (your desired state of traffic != the traffic you have) you do not - per se - have a penalty. if everything is a penalty, nothing is a penalty.
on SEO blogs everything is a penalty, sometimes with %stupid cute animal name here% prefix. this nomenclature really should not spread over to this otherwise great forum.
