It has been suggested that growing a site too fast makes G think your new pages must be SPAMy and may penalised the new pages and/or the whole site.
Its all to do with pr flow within the site. With 30 pages you were probably feeding lots of pr back to the home page. Now each page pr is being shared amoungst more links on each page to your new pages. In other words, where you had only a few links per page (each taking a proportion of the pr off the page) now you have more links and thus each link is carrying less pr.
Have you added a lot more links off your home page? If so, it is no surprise that they now make the next level only pr3. The home page may have reduced in pr because now it has pr3 pages pointing back to it rather than pr4.
If you had linked all the new pages from one of your original pr 4 internal pages and left the old pages as they were, I suspect your old pr4 and pr5 home page would have stayed the same.
You should consider how much pr each page needs in order to rank well for its target phrase. The new pages may be targeted at non competitive phrases so they may only need pr1 or 2. If that is the case, you need to redesign the pr flow so that the index page returns to pr5, your next competitve pages pr4 and the non competitve pages can drop to pr1 or2.
(I have a horrible feeling you are going to reply that you have not touched your old pages.... sigh :)
|(I have a horrible feeling you are going to reply that you have not touched your old pages.... sigh :) |
And you'd be right!
You are also correct about adding links to the home page. I probably went from the 30 to about 130, as I added 100 categories worth of products.
I guess the lesson here is that you can only promote your most important pages. I think I have become too generalized by offering so many categories, that not only have I diluted my pagerank, I've diluted my site theme as well.
So the solution is - restructure to a pyramid and pick my top 20-30 pages to receive the PR from the home page, or get lots more incoming links to add PR to my home page, huh?
an alternative is that you have added 1 and 1 and made 3.
i'm not at all sure that what you have done will have caused your home page to loose pr, infact as long as all your pages link back to the home page which i'm sure they do, then this is not the cause of your problem.
> restructure to a pyramid and pick my top 20-30 pages to receive the PR from the home page
Thats what I would do, and try and get a logical theme running down the links. You could cross link at lower levels to create themes.
>or get lots more incoming links to add PR to my home page, huh?
Or on theme links going deeper. Any page can be the status of a home page, if it has enough links in. You could make a third level page be the one ranking for a topic and make it like a specialist home page, with sites linking directly to it. The ultimate is too have twenty home pages in a site. Many people still see a site as a kind of 'brochure', but with any page ranking in google on its own merit, this concept should be abandoned.
>infact as long as all your pages link back to the home page which i'm sure they do, then this is not the cause of your problem.
I think it is. I reckon there is a formula for pr flow which dictates that the more links off a page, the less proportionally they give out. This is too stop massive cross linking within a big site having an advantage (it may effect anchor text value) and directory links giving significant pr to listed sites.
A pr5 page has 10 links out giving 0.5pr to each link.
A pr5 page with 50 links out gives not 0.1 but maybe 0.05pr per link. In otherwords they are devalued.
If that is the case, it is difficult to feed enough pr back to maintain the homepage pr and thus it loses some.
Just a theory.....
yes of course less pr is passed to each page when there are more outgoing links ... but how does this decrease the pr of the home page, the sub pages, yes i agree but the home page not he way i see it?
something else must have decreased the pr of the home page.
yes there is a dampening factor in the original published formula, but imo it is not 50% as you suggest in your example (i realise you probably just used that figure to easily demonstrate the concept - but i feel the actual dampening value is negligible).
anyway, i've increased the size of a site by thousands of pages and the ranking of the home page hasn't gone down, so i imagine the pr has remained stable - i don't know if the pr has gone up or down as i don't know how to get the value of it.
If Pagerank were calculated according to the first original formula, adding more pages could even increase main page PR. But only under condition, that linking structure were concentrating PR on main page instead of spreading it among the site.
In original formula, each new page had additionally 0.85 of real PR (real, I mean not the logarithmic one). So adding more pages theoretically increased site total PR.
Of course, PR formula probably changed, but some basic rules remained. If you have too many links from high level pages to low level pages and crosslinking between low level pages, your PR spreads between pages. In the opposite, many links from low level pages to high level pages concentrates PR on main pages.
|something else must have decreased the pr of the home page. |
I've used the "PageRank spreadsheet" to do a simulation of PageRank dispersal.
If you start out with a homepage PR5 and then link to 10 pages, your homepage PR is reduced to 4.25 or 4.1 or 3.5, etc depending on how you link your pages.
In other words, it looks like your PR is summed together over your entire site, and adding more pages spreads it out like butter. (Or jam, if you prefer)
Umm, in case you weren't watching, even Google does not care about Page Rank anymore. While it is nice to try and guess what is going on, the real answer is "who cares?" Google hasn't updated the toolbar in something like 15 weeks now. If it were that important to them, don't you think they would be updating it more frequently?
Personally, I find the announcing of updates by Yahoo refreshing. It sends a signal to webmasters - Thanks for your help, here is some information that you can use.
Google's attitude is one of "don't give away our secrets, keep 'em guessing." Think about these two positions. Is Yahoo that much better than Google at keeping the spammers away? Or is Google that much more insecure about the ability of spammers to manipulate the results?
Google uses these forums to their advantage, but what have they given back to this community lately? Google Guy's participation is near zero lately. Each time he has given email addresses to report problems, I have sent solid examples over to them. What I get back are cut and paste responses.
My sandboxed site has nearly a half million words on it right now. Today I got one visitor from Google (and I was the only site with this five word term on it). They have over 800 pages of mine in their index and I rank #245 for my own unique site name. Google is broken.
>Result: Homepage dropped to PR4, orginal 30 pages dropped to PR3, other pages to PR2
Was this 15 weeks ago? You may have lost a good link in from another site?
I've added about 11,000 pages to my original 50 page site over the last few months and Google has indexed, and is returning every page high in its SERPs. The homepage is PR6, the older original pages are PR5 and PR4 and every new page I've added is PR0 - it would appear that Google doesn't care about updating the PR figure any more because many of the PR0 pages are out-ranking sites with a much higher PR.
Its true that the toolbar is unreliable and often very out of date, but that does not mean that pr flow and the pr system is now unimportant. In fact the reverse could be true. By being careful which pages link to others in a site construction, the relevance as well as value of pr could be greater, according to the nature of a link rather than just pr. Thus a low pr page with logical and relevant links in can do well if in combination with other factors. A high pr page could rank poorly if the links in do not reflect the search term in some way.
Everything and nothing is important. Make things relevant and logical, then move on.
Assuming that there are no dead ends (orphan pages) on the site and no outgoing links:
according to the original papers:
sum of raw PR over all pages of the site = constant + number of pages
sum of raw PR over all pages of the site = constant
This means that you can just influence the distribution but not the total PR, i.e. total PR is independent from the linking structure.
|A pr5 page has 10 links out giving 0.5pr to each link. |
Toolbar PR has a logarithmic scale.
if your site was not a strong pr5 then it is likely that it got devalued anyway. if your site's (page acturally) pr feed is growing slower than the average growth rate on the net, your site is loosing pr.
pr is not an absolute value but a relative one. it is distributed amongst all pages on the net. so you only can devalue the homepage pr with your linking structure.
however, what i was seeing is that new pages pass pr (toolbar pr) like this:
- new page added
- google toolbar pr update -> new pages gets pr, but cannot pass it
- next google toolbar pr update -> pr gets passed
above happened a lot to my pages, and although it does not make sense if you consider the orignal pr formula it seems to happen. so you have to wait 6-12 months to see the impact of those pages to your pr. however, to measure that does not make sense at all, as to many thing will change over that period (pages with backlinks to your site change in pr, overall pr distribution changes, changes on important sites like dmoz happen).
what does this mean to you? just built the site so it fits best for your visitors and get good backlinks and outlinks. and keep the site fresh.
awesome thread you guys/gals.
I think i got my site right.
Homepage PR2 (first PR rating on new site)
all other pages PR1
all pages have a global navigation bar w 6 links (main pages)
each main page has several pages below (2 or 3 levels deep) within each main section there is navigation for that section (plus 6 global links on every page)
subsections have global navigation links and subsection links.
Total pages 200. (with lots to add)
Is there a potential problem with this type of link structure?
I wasn't really thinking about PR when i did this. I was thinking about making sense of a very diverse range of topics which are related in a few different ways.
I do have a related question. I posted it in the HTML forum but not getting any response (only blank stares).
It is really an SEO/security question regarding canonical vs relative linking. I'm interested wether this choice could also affect PR distribution.
So not wanting to throw this thread off topic but wanting some input from people interested in SEO/security issues I will post a link to my thread and propose the PR distribution question for this thread.
does relative linking affect PR distribution differently than absolute canonical linking or does it really matter? I would think not so I don't really want to start a new thread just to ask that. but since we are learning about PR distribution it seems like a valid question. Got me wondering anyway.
|pr is not an absolute value but a relative one. |
(Raw) PR is an absolute value just toolbar PR is a relative one.
|does relative linking affect PR distribution differently than absolute canonical linking |