Thank you, David.

Hi there! When the data from an A/B test shows a difference in AOV between the variants, how do you determine if this difference is statistically probable or not? Best Regards, Martin

We do not currently have any sanitation rules in order to maintain the nocrawl param. But that is a good point. 301:ing will be difficult for us but I will definitely add the nocrawl param to the rel canonical of those internal SERPs.

Thank you, Igol. I will definitely look into your first suggestion.

Thank you, Cyrus.

This is what it looks like:

User-agent: *
Disallow: /nocrawl=1

The weird thing is that when testing one of the sample URLs (given by Google as "problematic" in the GWMT message and that contains the nocrawl param) on the GWMT "Blocked URLs" page by entering the contents of our robots.txt and the sample URL, Google says crawling of the URL is disallowed for Googlebot.

On the top of the same page, it says "Never" under the heading "Fetched when" (translated from Swedish..). But when i "Fetch as Google" our robots.txt, Googlebot has no problems fetching it. So i guess the "Never" information is due to a GWMT bug?

I also tested our robots.txt against your recommended service http://www.frobee.com/robots-txt-check. It says all robots has access to the sample URL above, but I gather the tool is not wildcard-savvy.

I will not disclose our domain in this context, please tell me if it is ok to send you a PW.

About the noindex stuff. Basically, the nocrawl param is added to internal links pointing to internal search result pages filtered by more than two params. Although we allow crawling of less complicated internal serps, we disallow indexing of most of them by "meta noindex".

Igal, thank your for replying.

But robots.txt disallowing URLs by matching patterns has been supported by Googlebot for a long time now.

I'll send you a PW, Des.

Hi Mozzers!

We are trying to get Googlebot to steer away from our internal search results pages by adding a parameter "nocrawl=1" to facet/filter links and then robots.txt disallow all URLs containing that parameter.

We implemented this late august and since that, the GWMT message "Googlebot found an extremely high number of URLs on your site", stopped coming.

But today we received yet another. The weird thing is that Google gives many of our nowadays robots.txt disallowed URLs as examples of URLs that may cause us problems.

What could be the reason?

Best regards,

Martin

Thanks a lot, Tom. Time will tell...

Just one last thing:
what damage are you (and Google) thinking of when advising against removing URLs on a large scale through GWMT?

Personally, I think Google says so only because they want to keep as much information possible in their index.

Yes, I have put a conditional meta robots "noindex" on all pages whose URL contains more than 2 GET elements. It is also present on URLs containing parameters of little or no SEO value (e.g. the "price" parameter).

Regarding the nofollow directive, my plan is to not put it in the head but on the individual links pointing to URLs that should not be indexed. If we happen to get a backlink to one of these noindexed pages, I want the link value to get passed on to listed product pages.

My big worrie is what should I do if this de-indexation process takes forever...

Thanks for answering that quickly, Tom!

We cannot robots.txt disallow all URLs. We get quite a lot of organic traffic to these URLs. In july, organic traffic landing on results pages gave us approximately $85 000 in revenue. Also, what is good to know is that pages resulting from searching and browsing share the same URL - the search phrase is treated as just another filtering parameter in the URL.

Keeping the same URL structure is part of my preferred, 2-step solution:

1. Meta Robots "noindex" unwanted results pages (the overwhelming majority)
2. When our Google index has shrunken enough, put rel=nofollow on internal links pointing to those results pages in order to prevent bots from crawling them.

I have actually implemented step 1 (as of yesterday). The solution I was describing in my original post is my last resort solution. I wanted to get a professional opinion on that one in order to know if I should rule it out or not.

Unfortunately, I cannot disclose our company name here (I have a feeling our competitors use Seomoz as well :)). But I'll send you some links in a private message.

Hi all,

We run an e-commerce site with a catalogue of around 5 million products.

Unfortunately, we have let Googlebot crawl and index tens of millions of search URLs, the majority of which are very thin of content or duplicates of other URLs. In short: we are in deep. Our bloated Google-index is hampering our real content to rank; Googlebot does not bother crawling our real content (product pages specifically) and hammers the life out of our servers.

Since having Googlebot crawl and de-index tens of millions of old URLs would probably take years (?), my plan is this:

1. 301 redirect all old SERP URLs to a new SERP URL.
2. If new URL should not be indexed, add meta robots noindex tag on new URL.
3. When it is evident that Google has indexed most "high quality" new URLs, robots.txt disallow crawling of old SERP URLs. Then directory style remove all old SERP URLs in GWT URL Removal Tool

• This would be an example of an old URL:
  www.site.com/cgi-bin/weirdapplicationname.cgi?word=bmw&what=1.2&how=2
• This would be an example of a new URL:
  www.site.com/search?q=bmw&category=cars&color=blue

I have to specific questions:

1. Would Google both de-index the old URL and not index the new URL after 301 redirecting the old URL to the new URL (which is noindexed) as described in point 2 above?
2. What risks are associated with removing tens of millions of URLs directory style in GWT URL Removal Tool? I have done this before but then I removed "only" some useless 50 000 "add to cart"-URLs.Google says themselves that you should not remove duplicate/thin content this way and that using this tool tools this way "may cause problems for your site".

And yes, these tens of millions of SERP URLs is a result of a faceted navigation/search function let loose all to long.
And no, we cannot wait for Googlebot to crawl all these millions of URLs in order to discover the 301. By then we would be out of business.

Best regards,
TalkInThePark