Link flow for multiple links to same URL

[doctecs]

Hi there,
my question is as follows:

How does Google handle link flow if two links in a given page point to the same URL? (do they flow link individually or not?)

This seems to be a newbie question, but actually it seems that there is little evidence and even also little consensus in the SEO community about this detail.

• Answers should include source
• Information about the current state of art at Google is preferable
• The question is not about anchor text, general best practises for linking, "PageRank is dead" etc.

We do know that the "historical" PageRank was implemented (a long time ago) without special handling for multiple links, as e.g. last stated by Matt Cutts in this video: http://searchengineland.com/googles-matt-cutts-one-page-two-links-page-counted-first-link-192718

On the other hand, many people from the SEO community say that only the first link counts. But so far I could not find any data to back this up, which is quite surprising.

[rjonesx. 0]

1. The links both get PageRank flow...
2. The link value gets divided, though, so it wouldn't exactly double the value.
3. The link extraction process might choose to only select one link from the page based on certain factors (perhaps ignoring some links not because they are duplicative but based on location, or other qualifiers)

Here is Matt Cutts talking about this very issue. And here again. It is the closest thing we have to an answer.

I think the reason for the "first link counts" is really an extension of an understanding of PageRank. Let's say a page has 1 outbound link. It gets 100% of the value passable by that page. Now, let's say the page adds another link, but it is the exact same link. Now, each link gets 50%. The sum total is 100%. It is as if the 2nd link were never added. But, this calculation changes depending on the other links on the page. Let's say a page has 2 links on it. One to you, one to someone else. 50/50. If you get another, you jump to 67/33. Slightly better. As the page increases in number of links, your additional link approaches a doubling of the first link's value. So on one end of the spectrum it is valueless. On the other end of the spectrum it doubles.

The other question is whether anchor text is counted for all links. Some experimentation indicates that only the 1st anchor text matters. This might also indicate the selection / extraction process mentioned in #2.

That all being said, I think I agree with Matt Cutts on this one. This is such a small issue that you really should focus on bigger picture stuff. It is interesting, yes, but not particularly useful.

I hope that helps!

[doctecs]

I totally agree on the focus thing in general - it's not helpful to act with PageRank in mind when it comes to layout decisions etc.

But: For large websites (e.g. 100,000 pages and up) crawl rate, indexing and rankings of deeper parts of the site depend heavily on the internal link graph. Taking a deeper look at the internal link graph gives us a lot of useful information in these cases, does it?

Now: Think of links sitting in a template that gets used on 50,000 pages. A little change here is likely to cause quite a difference in the internal link graph.

For example I've run PageRank simulations with both models on a smaller website with only 1,500 pages / 100,000 links. For many pages, the little difference ends up with 20-30% more or less internal PageRank - for these individual pages, this could be crucial for crawling, indexation and rankings. Still not useful?

Since moz runs it's own iterative PR like algorithms: How do you guys handle this with mozRank / mozTrust? Which model leads to better correlations with rankings?