Welcome to the Q&A Forum

Hi Ryan,

thanks for your reply! First off, I am aware I won't be able to compute the PageRank, rather my little own approximation of it. Basically OSE does the same (I guess they also tweaked the algorithm quite heavily).

Purpose being to optimize internal linking / link juice distribution. With Page Authority I can't preview the effects of changing internal links.

So you'd say the best approximation of a page's link power is page authority?

Hi,

we're running a project here where we try to optimize our internal link structure by computing the PageRank ourselves given your existing link graph. Right now we weight all pages the same, which is kind of incorrect given that quite a number of our pages have external links, hence have more LinkJuice.

We'd now like to include a weight that respects the external link power to a page. Anyone got suggestions OSE metric would be good? I was thinking of Page Authority, but I assume this is computed by taking both external and internal links into consideration.

Alternatively I could use sum all PageAuthority values of inlinks to a certain page, but as Page Authority seems exponential, this seems to be the wrong thing.

Anyone out there you who tried the same before?

Cheers,
schuon

Hi Dan,

it's generally a good idea if you identify a (unique) primary keyword for each of your pages. Am I understanding you right, that you have a bunch of producer pages, which you want to have ranking for the name of the producer?

Then I'd say the primary keyword is the name of the producer and I would try to link to those pages with the primary keyword. This helps Google in understanding what the page is about. Slight variations of the keyword are appreciated. Also you could have some "click here" anchor texts (which is totally natural in your anchor text distribution).

Just avoid the following thing: assume you'd be selling cars and you link to /Porsche sometimes with the anchor "Porsche", but other times with "Ferrari". This throws Google overboard and it can't tell if it should rank this page for Porsche or Ferrari. If you'd link with "see more cars" Google will figure this is off-topic.

HTH,
schuon

Hey guys,

I was just wondering what the whole schema.org markup means for people that run search engines (i.e. for a niche, certain products) or price comparison engines in general.

The intend behind schema.org was to help the engines better understand the pages content. Well, I guess such services don't necessarily want Google to understand that they're just another search engine (and thus might get thrown out of the index for polluting it with search result pages).

I see two possible scenarios: either not implement them or implement them in a way that makes the site not look like an aggregator, i.e. by only marking up certain products with unique text.

Any thoughts? Does the SEOmoz team has any advice on that?

Best,
schuon

Well, we just want to show less links to Google than to the user (but the links for Google are still a subset of the links shown to users). The links we'd do as JS links are those to less often applied search filters, which we don't index in order not to spam the search index.

Fortunately, if Google is smart enough in decrypting the links it wouldn't do any harm.

Thanks for our ideas tough! Especially the site: thing I considered myself, it really takes ages until something is de-indexed (for us, using robots.txt did speed it up by a magnitude).

Hi,

in an effort to prepare our page for the Panda we dramatically reduced the number of pages that can be indexed (from 100k down to 4k). All the remaining pages are being equipped with unique and valuable content.

We still have the other pages around, since they represent searches with filter combination which we deem are less interesting to the majority of users (hence they are not indexed). So I am wondering if we should mask links to these non-indexed pages with JS, such that Link-Juice doesn't get lost to those. Currently the targeted pages are non-index via "noindex, follow"  - we might de-index them with robots.txt though, if the "site:" query doesn't show improvements.

Thanks,

Sebastian