One thing I'd like to know is if all those pages link back to the homepage is really any there any leakage?

Another example would be how a site might have the same set of navigation links in a subsection all linking to each other. It's a good way visitors can see all the pages on a topic but does it hurt in terms of leakage or does it just all even out?

If you look at the useability of the site it seems to me that 60 links on the homepage would be overwhelming to the visitor. So there are other reasons to cut down on the number of links.

I'm bumping this up in hopes that someone can give us some info on how page leakage works.

It won't "leak" by linking to pages on the site that link back, but with 60 links, the amount of PR being given to each of the interior pages is way diluted. Links to pages on the site isn't the same as outbound links going to sites off the site. Which are they?

Using a 'Google PageRank Calculator' is by far the best way to test the effects of links on the PR of the pages on a website (unless you're a clever mathematician). The results can be very surprising and often the opposite of what one would expect. For example, additional outgoing links on a page can actually produce an increase in PR for that page.

Added: there's one in particular that lets you create a grid for a site into which various link strategies can be entered and automatically tested to see the resulting PR effects on all pages. Of course this does not show real PR, but is very useful for comparing one approach against another.

[edited by: Patrick_Taylor at 3:16 pm (utc) on Jan. 30, 2007]

Can you tell us whre to get that test program

>>For example, additional outgoing links on a page can actually produce an increase in PR for that page.

some very clever mathematics must go into that particular PR calculator! seeing as PR is a product of incoming links only, nothing else effects PR.

[of course i do understand the effect of cycling PR through a site and back to the homepage, however the fact remains outgoing links on a page do not effect the PR of that page]

[edited by: topr8 at 5:12 pm (utc) on Jan. 31, 2007]

Can you tell us whre to get that test program

There are several Google PageRank Calculator tools available online. We don't want to play favorites here, fendy, so I suggest you do a Google search and pick one that you are comfortable with.

I wouldn't be particularly concerned about "PR leakage." It's more important to make sure that you aren't engaging in link schemes designed to manipulate your site's ranking.

Many top-quality sites feature quite a few (quality) outgoing links, and are appreciated for that by visitors.

Also I need to ask you what about linking to other sites from home page, will that cause the PR to leak?

If these links are non-reciprocal then there is a leakage of PageRank, i.e. the inner pages as well as the homepage will decrease in PageRank. (However, the effect might be too small to see it on the toolbar.)

An example of the effect can be found here [pagerank.suchmaschinen-doktor.de] - compare example 6 with example 8.

The results can be very surprising and often the opposite of what one would expect. For example, additional outgoing links on a page can actually produce an increase in PR for that page.

Not in case of non-reciprocal links.

...seeing as PR is a product of incoming links only, nothing else effects PR.

Of course, PageRank is a sum of incoming links - however, these additional links affects the PageRank of the page with the incoming links.

Looking at a bigger picture of ranking itself (not just PageRank), we can notice the power of anchor text as an ON-PAGE factor. This alone can more than compensate for the little mathematical anomaly of "PR leakage". Also, have you ever done the exercise of calculating the PR circulation effect of adding a new page or two? That's an eye-opener.

If PR leakage were really something worthy of major concern, then how would a directory site, which is very heavy in OBLs, ever get anywhere?

I feel it's much healthier for your site to have an abudance mentality. The "link love" you flow does come back to you in many ways, because the web is a vast system, and what goes around DOES come around.

If you hold on to what you've got too tightly, you just might squeeze the life out of it. Instead of putting your time and energy into controlling PR leakage, write some more good content instead. That will give you a much better ROI, and it can keep on growing in its effect, year after year.

All that said, I think you do want to take care with your domain's root page. Scattering many OBLs around on the homepage doesn't make a lot of sense. It's like inviting people to a party at your house, and then telling them to get lost when they show up.

People tend to worry way too much about their OBLs leaking PR. You are much better off spending time working on your internal navigation structure.

If you think directories are all about outbound links, you probably haven't done a lot of link counting on directory pages. They are certainly heavy on their OBLs, but they are also heavy on internal links.

Looking at a bigger picture of ranking itself (not just PageRank), we can notice the power of anchor text as an ON-PAGE factor. This alone can more than compensate for the little mathematical anomaly of "PR leakage". Also, have you ever done the exercise of calculating the PR circulation effect of adding a new page or two?

Of course, PageRank leakage is nothing to worry about. And indeed ranking might be improved by adding additional link to a page due to the benefit from on-page optimisation. Nevertheless, the leakage effect exits.

The effect of adding a new page is increasing the (real) PageRank of (almost) one (strictly spoken it depends on the linking structure). In case that that you're having a page without outgoing link the effect is much better. However, while the effect of adding a new page for the original algorithm can be calculated one has to say that Google changed the algorithm significally some year ago. Have you ever tried to add billions on pages and 'produces' PageRank? It won't work. Instead of 'producing' PageRank you nowadays need PageRank to get the pages spidered.

The biggest single PR error that sites make is in linking to www.domain.com/index.html when Google lists www.domain.com/ as the root.

Related to that is having both www and non-www URLs in your links. All links should go to one version, and none to the other.

In any case, one (usually non-www) should return a 301 redirect, and the other a "200 OK" with content.

Worry about those factors first. They will have the most impact.

The biggest single PR error that sites make is in linking to www.domain.com/index.html when Google lists www.domain.com/ as the root.

Of course, one should avoid that the same page is indexed with two different URLs. On the other hand, Google is clever in merging the same page (with different URLs) espacially in the case descriped above. If the pages are merged both show the same PageRank as well as the same backlinks. To verify if the pages are merged try a google search for 'info:http://www.domain.de' and 'info:http://www.domain.de/index.html'. In case that both URLs are merged just one is shown as result (probaply that one without 'index.html'). If the pages are mereged there is no negative effect compared to just linking to one page, i.e. backlinks and PageRank are added.

>> On the other hand, Google is clever in merging the same page (with different URLs) espacially in the case descriped above. <<

I can show hundreds of examples where the results for / and for /index.html or .php are not merged.

They used to run a "merge" process over their database several times per year, but I don't think that has happened for perhaps several years now.

If you have results that have been merged then I would put that down to luck more than anything.

I said "Of course, one should avoid that the same page is indexed with two different URLs.". I didn't say that these pages are always merged. However, in all cases I know (static pages, more than a dozen examples) these pages are merged. I don't know what the criteria to merge pages - at least it has nothing to do with luck. Perhaps the probability of merging pages is better for static than for dynamic pages. (Therefore, I said "espacially in the case descriped above", i.e. "/" and "index.html".)

If Google already merged pages there is no need to change anything.

doc_z,

I think you might be a bit out of date on Google doing a merge these days.

ISTR a shift back in 2004 around canonicalization in its various forms. I could be wrong, but I don't think so.

[edited by: theBear at 4:15 pm (utc) on Feb. 8, 2007]

Maybe I'm out of date, but these examples are still valid and some pages were created less than a year ago.

Anyway, I never said you should expect that Google is doing this. Moreover one should avoid this situation. However, if Google already had done this (as in the cases I'm referring to) there is no need to make changes.

I've seen some merges that still happen, but it is only on static sites, that are not updated for a very long time, and pages that are obviously the same such as / and /index.html. It happens, but you can't count on it.

The biggest single PR error that sites make is in linking to www.domain.com/index.html when Google lists www.domain.com/ as the root.

g1smd, thanks for that info.

What is the beat way for me to link to the homepage from internal pages.

1) Anchor link "Home"
2) Anchor link "word1 word2" (word1word2.tld is my domain name)
3) An image w/link
4) An image with alt tag w/link