Removed HTTPS via GWMT, suddenly ZERO results in SERP

[chewy]

Learned I had an ecommerce site that had a ton of https product pages in the Google index.

Came to realize this needed cleaning up, and fast.

Site has about 200 products. Google has been showing anywhere between 6,000 pages (at the start - there was a ton of infinitely repeating non-canonical junk, we got that cleaned up), and it was then down to 1,900 pages - much of this due to the "S" pages and other nonsense that is in the process of cleaning up through 301 redirections, etc. What a mess!

Any https is handled by the shopping cart, that is on an entirely different domain, so https should never have entered the picture.

Site is running on Shopify - their advice was to block the pages with a couple of lines in the header.

{% capture protocol %}{{ 'reset.css' | asset_url | truncate: 5, '' }}{% endcapture %}
{% if protocol =='https' %}<meta name="robots" content="noindex, follow" />{% endif %}

So we did that.

Then this smart guy (me!) learned that I could create an https instance over on GWMT and ask it to remove all the urls.

So I did it.

Within minutes, the whole site has disappeared from Google. And I mean REALLY disappeared.

1900 pages down to 3 really wrong pages should be redirected but that's another story - they are an artifact of the cart being set up incorrectly years ago.

Been trying to isolate these problems and solve them one at a time.

Organic Traffic has crashed as you might imagine.

Should I get busy and phone up the Google geeks? Did I miscalculate that GWMT would help?

Or is this just the Google Dance and all I need to do is wait? How long?

I'm about 2 days into this and wondering if I need to get busy or what...

[levo]

I did this exactly last week. Go to WMT > Removed URLs, and select Removed from top right drop down menu, click the reinclude button. Should recover in 1-2 hours.

[chewy]

trying to read between the lines here - but perhaps some clarification is necessary?

I have 2 GWMT accounts for this domain.

One for http, one for https.

Are you suggesting I re-include the https (secure and duplicate content) pages that I am trying to get OUT of the index?

Of course, what I want is for the HTTP pages to get back into the index, and GWMT there shows everything is hunky dory.

[lucy24]

I have 2 GWMT accounts for this domain.

Whoa. Hang on there. Two entirely separate accounts? Or a single account encompassing all four forms of the domain name?

[chewy]

Not sure where I read this - probably MOZ, but I believe that's how they suggested removing https pages from the index.

Yep... (not sure if the new system allows for posting of URL's but hey, i'll give it a shot to see what happens!)

[moz.com...]

Yes, indeed, you can create another GWMT incidence within the same account. It doesn't quite seem to work right, but hey it was worth a try.

This domain has no www so i don't know why I would have FOUR incidences.

On a hunch, I followed Levo's suggesion, resubmitted the root (on the https side) and boom - back in business in a couple of hours.

Whew. Just wish we had a bit more guidance from the big G but with moz and WebmasterWorld, I seem to do OK.

>>Actually, I don't know what I'd do without WebmasterWorld!

[levo]

Don’t use the URL tool to remove the URLs you don’t want in search results. It won't keep your favorite version of a page; instead, it'll remove all versions (http/https and www/non-www) of a URL.

[support.google.com...]

[Robert Charlton]

This thread, while not precisely on the topic, gets into aspects of cleaning up canonical issues wrt http and https, and has some very helpful posts....

Cross-site canonical meta tag questions
http://www.webmasterworld.com/google/4539413.htm [webmasterworld.com]

In my experience, mod_rewrite is necessary to clean this up. Robots.txt, meta noindex, or using Google's remove page feature won't do it, and can get you into trouble.

There are good reasons post why the canonical link element may not work. I would use it only if it is manually set, and I'd use it in addition to proper 301 canonicalization, but I would not depend on it.

One question... do you currently have relative urls in your navigation code? If so, that's a strong clue about how this happened.

[edited by: Robert_Charlton at 8:56 pm (utc) on Apr 11, 2015]

[phranque]

redirect those non-canonical HTTPS requests to the canonical http URL.

the reason you are tracking all the versions of your site in GWT is so that you can verify that Google is reporting what you would expect for each version.

[lucy24]

This domain has no www so i don't know why I would have FOUR inciden[t]s.

You need to add it to the wmt account in order to-- paradoxically-- tell google that you don't use it.

[chewy]

meanwhile, all seems back to normal (with the https back in the SERP).

Need to come back to this thread maybe after a bit more reading - please is there a reasonably recent, reasonably accurate written up guideline of how to use this tool somewhere (other than at Google?)?

thanks all - will return shortly!