Link resolvers (e.g. DOI) effect on Google ranking

[swing]

Hello,

I would like to post a mysterious phenomenon of the journal "Forum Qualitative Sozialforschung / Forum: Qualitative Social Research" (FQS) in this forum with the hope to hear about similar experiences, uncover the mystery and find a solution.
The mystery is the Google search results degradation for FQS since its migration to OJS. FQS used static HTML for years and had a great ranking. It migrated then to OJS and the ranking fell to 0. Although there are redirects from the old FQS pages to the new ones, this was probably to expect because everything changed and Google had to realize it. Now, almost 2 years of using OJS, the ranking is better, but still not so good (and definitely not that good as it was earlier).
With the migration to OJS, FQS also changed design, i.e. the HTML structure for articles and started to use Uniform Resource Names (URNs), i.e. there are a lot of redirects from the old URLs to the new URNs that further redirect (HTTP status code 302) to the articles in OJS.
Does somebody know how is it actually with the DOIs/URNs and the link resolvers -- The articles are cited with the DOI/URN (e.g. http://nbn-resolving.de/urn:nbn:de:0114-fqs010146) that further redirect to the real article in OJS. Does this have some effect on ranking?
We would be very happy and thankful if somebody could tell us what happened/is happening, how does this function, what FQS (and other journals) should/could do...

Thanks a lot!
Bozana Bokan

[jerico]

Hi Bozana and all,

as you know we've taken up this problem and are researching it in different directions as we believe that it could have serious impact on many within the PKP community, authors and publishers alike.

We believe that there might exist two mostly unrelated problems with link resolvers, scholarly citation standards and search engine rankings:
1) The 302 redirect (temporary redirect) most popular link resolvers (including DOI and URN) employ might lead search engines to (partially or fully) attribute link credit to the link resolver domain rather than to the target article, it's author and/or publisher.
2) All common citation standards ask authors to link to the original article using it's URL as anchor text and not the title and/or the authors. Search engines use anchor text as one of the most important sources for keywords, though.

The first suspicion is somewhat supported by the fact that we didn't find any backlink from a site that used URNs to FQS (one of the journals Bozana's organization operates). We tried it in both, Google's webmaster tools and Yahoo!'s site explorer. Both, Yahoo! and Google, only find non-redirected links or links that FQS themselves redirect with a 301 response code (permanent redirect). We found several articles on the web that confirm that 302 redirects may leave all or part of the link credit (="link juice" in SEO jargon) with the redirecting page rather than with the target page. According to these sources 301 redirects (permanent redirects) should solve that problem which coincides with our own observation.

The second problem is very obvious. The impact of anchor texts on search results is well known and proven. (The most popular proof being that "Acrobat Reader" ranks very high for the keywords "here" or "click here" in many languages although this word has no semantic connection with Acrobat's software just because so many sites link to the reader with the word "here" in the anchor text.) Unfortunately there's no easy solution to this problem unless citation standards change or search engines learn how to recognize and parse citation standards. Maybe they already do?

Let me quickly summarize the activities that we've initiated so far:
1) We've done preliminary research about the impact of 302 redirects (see results above)
2) We've checked the backlinks for re-directed links in the case of FQS (see results above)
3) We've contacted CrossRef support and asked them about their opinion (no answer yet)
4) We've posted a question on Google's search forum (see http://www.google.com/support/forum/p/W ... 088a&hl=en, no response by the time I write this)
5) Today we also contacted the German National Library who manages the URN standard for Germany and asked them to temporarily switch FQS links to 301 redirects so that we can test our backlink hypothesis.

If any of you observes similar behavior then please tell us. If you use a link resolver scheme like DOI, PURL, etc. please try out whether you can find backlinks in Google's webmaster tools and Yahoo!'s site explorer from pages that contain nothing but a DOI, PURL, etc. link. If you need further assistance, then tell me. I'm happy to help.

Kind regards,
Florian

[jerico]

A quick update: It seems that neither Yahoo! nor Google show cross-domain redirected backlinks even when 301-redirect is being used. So the missing DOI, URN, etc. backlinks are no longer a strong evidence that they are not counted towards the search engine ranking. This does not invalidate the argument, though. Most SEO experts still believe that 302-redirects do not (or not fully) forward link credit while 301 re-directs do. It is just not of much use to check for cross-domain backlinks in webmaster tools or site explorer to check your case.

I also found one PURL-specific Google forum thread: http://www.google.com/support/forum/p/W ... 96fc&hl=en

[jerico]

I just got this answer on the Google help forum which seems to confirm my initial assumption:

"Googlebot does with URLs what the URLs tell Googlebot to do,
if a URL redirects 301 to another URL this tells Googlebot to index
in search results the URL destination of the redirect,
but if a URL redirects 302 to another URL
this does not tell Googlebot
to index the URL destination of redirect.
I think the question to ask to the link resolver sites is if they
would consider changing the redirects 302 to redirects 301,
that would be very easy to implement."

I've moved the Google help thread to the webmaster forum which seems the more appropriate place for our discussion:
http://www.google.com/support/forum/p/W ... d8c1&hl=en

[jerico]

FYI: I now found this small message from 2005 about "Google and DOI". It only speaks about Google Scholar, though: http://www.doi.org/news/DOINewsApr05.html#2