25 replies to this topic

[EGOL]

This is a side discussion that I would like to start based upon bloard's thread on the topic of the Rate to Add New Pages. I don't want to highjack that thread.

I want to make a separate post on this topic and state right up front that what I am saying here is speculative but posted for the purpose of getting discussion, feedback and observation sharing from others. It also applies to Google where link power is highly important.

Based upon watching a small number of sites very closely, I believe that there is a balance between inbound link power and the size of a website that determines competetiveness in the SERPs. In other words, I believe that given high quality content on all pages, a ten-page site with 500 links will perform better in the SERPs for it's keyword inventory than a one-thousand-page site with the same identical links. The larger you make a site the more links you will need to hold the same clout in your SERPs.

Each page of unique content that you add increases the keyword inventory of the site and hence the number of queries it can appear for but at a reduced rate of effectiveness. So, in highly competetive SERPs the most effective weapon will be a rather small site that accumulates enough link juice to raise a few pages high enough to appear on the first page of the SERPs. This will do better than a large site with a huge keyword inventory where there is competition out in the long tail SERPs.

The huge site is the weapon of choice where competition is weak and the long tail SERPs are not saturated with other websites.

That's my personal assessment. Does anyone agree? (Again this is a speculation about Google - I believe that things are very different on MSN.)

[Ruud]

I believe that given high quality content on all pages, a ten-page site with 500 links will perform better in the SERPs for it's keyword inventory than a one-thousand-page site with the same identical links

As this is based on your observation I wonder if the following apply:

• are the links to the same spread of pages? or does the 10-pager have all links to the homepage, for example, while the 1000-pager has them to 100 different pages?
• are the links pointing to the (most) relevant pages for the terms you're looking at?
• are the anchors more or less identical in meaning/power?

Each page of unique content that you add increases the keyword inventory of the site and hence the number of queries it can appear for but at a reduced rate of effectiveness.

I don't think the search engines filter the query bubbling down the word pairs on a per site bucket basis. I think that the keyword inventory, to go with that idea, is a giant set per query taken from the larger, total keyword inventory. Any document, not site, can bubble up to the top as the most relevant.

For me this is one of The Keys: every page is a site, every page is a potential landing page.

[A.N.Onym]

EGOL, that comes from a very simple rule we all know.

The more links there are on a page, the less juice they pass to every of the links.

So, if you have most of links to a homepage, you have to link to more pages from it, if you have more pages. And those secondary pages will have more links, passing less weight, to other pages.

So, in essence, the secondary and tertiary level pages will be much weaker on a larger website, even if you get links to the homepage.

Now, if you have 500 links to a small site (10 pages), the chances are there will be about 10-30 links to one of the non-homepage pages. That page will be able to rank fairly well.

On the other hand, if you have 1k of pages, the chances are only few of the pages will be more popular with the others (we keep the amount of incoming links the same, remember?). In reality, though, a larger website with 1k of quality posts over 10, will get that many times links, too.

In essense, what you may have come up with is this: the more concentrated quality on your site is (on the pages), the better off you are. That rings true to me and reminds me of an idea to focus on creating quality articles/posts over average stuff (now Aaron Wall would disagree with this, however).

[Halfdeck]

I prefer a huge site with tens of thousands of IBLs over a 10 pager site with 1000 IBLs.

Internal anchor text works wonders for a huge site - you can basically nail both high traffic keywords and long tails that way.

If I was limited to 500 IBLs, then I'd make the site as big as possible without reaching a point where deep pages started going supplemental.

[Pittbug]

In this scenario where the 500 BLs are the same, the 10 page site is better because PageRank isn't spreadout so thin, however, don't discount a 1k page site either. It may not do very well on competitive keywords, but it could have a huge tail that could bring more traffic than the 10 page site's big keywords.

[bragadocchio]

What are the objectives of the site.

[A.N.Onym]

I'd figure that whatever the objectives are, it'd be easier to achieve them by a more focused website, which is more visible from the search engines.

Then again, there may be exceptions.

Yes, there need to be more identifiers to this problem. And the amount of content/link needs to be more real, too.

[Ruud]

Given the spread of searches I wonder then too if long tail doesn't most of the time outperform laser targeted searches?

[A.N.Onym]

It does - it is long tail (20/80), remember?

Wait. That means that a larger site is more efficient. Hrmm. Depends on the niche, I guess.

Edited by A.N.Onym, 20 December 2006 - 01:23 AM.

[AbleReach]

The huge site is the weapon of choice where competition is weak and the long tail SERPs are not saturated with other websites.

A large group of laser-targeted pages could fill out a category and feel like a resource to the visitor. If there is a need for a lot of pages in the first place, then a large site could fulfill both highly targeted and category-oriented searches.

[bloard]

First, thanks for giving my original thread enough thought to begin this one. I guess I am going to find out the answer.

What I have is a site where the main topic can be divided into about 4 logical subtopics. I purchased the old site about 6 months ago as an old, forgotten site with some excellent links pointing to it, but that ranked for nothing due to stale links and content. I then reformatted the old content that was already there - which was a directory that hadn't been updated in about 5 years. Thus, the site is currently about 1k pages.

I also threw up one test page on what I would consider a top 50 keyword being targeted. ($15 top bid on overture, 97 million results in google, about 1 million searches/month) I then pointed a couple of off topic links from quallity old sites at it and sat back and watched. 2 months later it ranked in the top 10 for the targeted keyword.

Being afraid to touch it or mess with what was working, I didn't change the content at all, or add a single new link. As expected after 3 months it fell to result 50ish. Have since refreshed the content on the page... gotten a couple of quality on-topic links to point to it again, and it's climbing again, now sitting in the mid 20's. Fully expect it to be page 1 again within the month.

I'm still afraid to fully develop that subtopic of the site. I could add regional pages about the subtopic but am afraid I would steal quality away from the page that is ranking well now. Probably a bad decision on my part.

So... there is another subtopic where regional content is also easy to develop and very popular/competitive. Obviously I won't add scraper or recycled content so as not to harm the overall quality of the site, but I really believe that I have found a balance of unique content that will allow for the creation of about 30k unique pages of content that visitors will actually find helpful. Consider it basically a regional list of local retailers along with current inventory and profiles of the retailers supplied by the retailers themselves... thus totally original, and relevant.

Now the experiment will begin. I am going to thow at least 10k of these new pages at google over the next couple of weeks, thus increasing the size of the site tenfold, without a corresponding increase in inbound links. Will these new pages rank well, given the respect that google seems to give to the site as a whole? Will it harm the page that currently ranks so well on the other subtopic? Who knows, but we are about to find out. I should know the answer within about 4 weeks and will certainly post the results here.

Thanks again for your consideration of my question.

[EGOL]

Thanks to all for the above replies and to bloard for providing a great description of his experiment.

Based upon what I have watched with my sites and my original post, I would predict that the rankings of the original site will decline as lots of new pages are added. My hypothesis is that a certain number of links can support a certain number of pages... if the page count goes up an increase in links is required to retain original rankings.

This is how I think that it works with Google. With MSN I think that the original rankings will hold, maybe increase if lots of links on the new pages point back to the old pages.

Good luck with this experiment. Let us know what happens if you care to share about it.

[bloard]

EGOL - You are scaring the heck out of me. Just about to push the button to send these new pages live... but also looking at the earnings of the site as it exists and wondering whether to bet it all. I'm a gambler at heart, so will probably do it. Plus I have a couple other sites on the same topic I bought and am developing in the same manner with similar characteristics, so hopefully I have something to fall back on.


I honestly never gave your theory much consideration (and I give just about every theory some consideration) so it has me taking a breath and thinking.

Edited by bloard, 21 December 2006 - 08:58 AM.

[luvdavy]

I'm located in a resort area in the real estate field, and have just seen the exact opposite of the results here.
Several programmers have learned how to pull in the MLS listings in a site, and have a script that creates static pages for all the listings. It results in thousands of pages created the first few weeks. Add to the script so that it creates a somewhat different title and description for each page, and these sites will take over the results of every search term they target.

The specific one I'm referring to has been around for several years unoptimized and has almost no incoming links except some satellite sites they created for specific properties. It went from being nowhere in the SERPS to uprooting and passing most of the older, more powerful sites around. Not all of them, but a good number. Now it's in the top 3 for every term it targets...and took only about 6 months to do it.
There are a few other spam techniques going on, but nothing that could have catapulted it to the top except all those extra pages.

So I'd be inclined to say that a super large site would do better than a small one if every page is optimized.
Perhaps the reason the other example given here was not could be that the titles and descriptions...and perhaps internal linking is not good?

[bloard]

Ok.. here is some more data to throw on the topic... As I stated, this site had traditionally been just an old directory with very short descriptions for the various sites. So when I got it, I used the same database, but a newer directory script that created a separate page for each site listed in the directory. All URL's were new when I took it over because I converted it from CF to PHP. I didn't care at all about the directory pages as I was just trying to get my test money page to rank off of the age of the site.

From June when I redesigned the site through October, google had indexed about 1k of the pages. The directory portion of the site was getting about 300 uniques per day. Then in early Nov., I decided to finally create a sitemap for google so it could find about 3,000 of the directory pages it was missing. Thus, in mid November the size of the site as far as google was concerned tripled. Not caring about the directory pages, I never added anything to make the individual pages have any real value. For a week or two as google found these previously undiscovered pages, traffic went up a fair amount, but over the past month has stabalized back around the 300 uniques per day... even though the site is now 3x's the size it was before as far as google is concerned. Looking into it a little more, the reason for this is that the vast majority of the pages of the site, including the new pages it found after doing the sitemap are supplemental.... and frankly should be supplemental.

But, regardless, the size of the overall site increased 3 fold in November as far as google knows (although the number of pages in the google "active" index really didn't increase much). Yet traffic to those pages, has essentially remained unchanged. None of this should be news to anyone, and can be explained by the poor quality of the pages that were found and tossed into supplemental.

But remember that "money page" I keep referring to? I have seen no effect (either positive or negative) on that page due to either of the following: 1. The size of the site increasing 3 fold overnight. 2. Google discovering 3 thousand "new" pages it considered "bad" and tossed to supplemental. Personally, I would think that this might have had a negative effect on the money page and site as a whole because my ratio of non-quality to quality pages increased substantially (supplemental to regualar) Whereas before, the site might have been 500 pages regular index, 500 supplemental... it is now about 3,500 supplemental, 500 regular.

Again, I have seen no impact on the money page that I would attribute to this issue. While that page languishes now in the 20-30 range for its big keyword, it had already fallen to 50ish at the time the sitemap was added. It seems to rise and fall on its own merit... age of content, quality and continued new links.

But... enough of an issue has been raised here that I think I am going to roll out the new pages about 100 at a time at first to make sure the content is good enough to stay out of supplemental. I'll start a separate thread asking what it takes to get those 2,500 pages out of supplemental, but I imagine it has something to do with "good content".


Sorry so long.

[mugshot]

While I have always agreed that links do matter...but this is slightly boggling.

I propose this then. If I have a site that has 5 pages with 100 links ranking better than a site with 80 pages with 100 links, then why don't I just create 10 5-page sites and dominate? I know this is extreme, but sure sends that message.

I think it might be more accurate and less misleading if we went on a %-based linkage? But then, we have to take into account the quality of the link, relevancy, etc...a bunch of stuff...

sigh...why am I doing SEO

Good post starter EGOL!

[JohnMu]

Have you ever done a pagerank series for a large site? I made a tool (for myself :-)) to do a full pagerank calculation for sites, where the value feed and the outlinks can be specified per page. It's interesting to play with site sizes and structures and to see how they effect pagerank distribution.

One thing that will make a difference is if the site is fully indexed or not. From what I have seen there is a threshold (I assume in terms of pagerank, at least for Google) which Google requires in order to crawl+index a page. If you have a large site but only enough "value" to get 10% of the pages indexed, your pagerank will only be spread on those pages, leaving more value per page than if the whole site were indexed.

In general, this means that the value of the pages will be slightly above the minimal value (threshold) - except for the pages with value feeds. Say you feed 10% more value to the site (more links) - that value will be used to get more pages above the threshold, getting them indexed as well (with minimal / threshold value). Adding more value = getting more pages indexed with minimal value (again, except for the pages where the value is fed and of course depending on the site structure; I'll assume we have an optimal stucture, but it doesn't really matter that much).

Now if the site is fully indexed, then all additional value will go towards raising the value of all pages (according to the pagerank distribution of the site-structure).

Of course if we take a fully indexed site from say 1000 pages and reduce it to 10, then we will have a higher value per page, probably pushing the pages themselves higher in ranking (depending on whether or not PR is the limiting factor for the pages / keywords).

However, you can't compress a site from 1000 pages to 10. 1000 pages will attract more possible links than 10 pages could :-). If you were to reduce the wikipedia to 100 pages, would it still be as valuable as it is now? Would it still attract as many links as it does now? In reality, you can't compare a site with 1000 pages (and say 500 links) to one with 10 pages (and 500 links) -- it's more like 1000 pages/500 links vs 10 pages/5 links.

If we take a fully indexed site and add more pages, it will reduce the value of each individual page, slowly going down to a minimal/threshold value (and then crossing into the supplemental index). However, adding more pages could also mean attracting more links. Diversity (a large site) can be a good thing - it helps you to gain links which otherwise would not be possible.

John

[bloard]

Ok John - I fully understand and agree in general as it is similar to the image below of the champagne pyramid below:


Pour it in at the top and it will flow down to pages below. Unless you have enough "flow" going in, it won't "spill over" to those pages below it. Now I also understand that optimal internal linking isn't a pyramid, but rather linking back up the pyramid as well... but lets use the pyramid example for discussion sake...

Assume, rather than pouring all of the champagne into the top glass, you choose one on the side a few levels down. Obviously that is going to fill up the glasses below it (assume that to be one section of a site), but won't flow to other parts of the pyramid not directly below that section.... but does it matter how many empty glasses remain in the overall pyramid that aren't getting any "feed" as it relates to getting those in the targeted section to rank?

In other words... if I have a 10k page site... but only 50 pages on topic A. If I pour enough "feed" as you call it into the part of the pyramid on topic A so that those 50 glasses are full, is topic A still going to rank well... all else being equal... even though there are an additional 9,950 empty glasses in the pyramid?

Hope this makes some sense...

[AbleReach]

If we take a fully indexed site and add more pages, it will reduce the value of each individual page, slowly going down to a minimal/threshold value (and then crossing into the supplemental index). However, adding more pages could also mean attracting more links. Diversity (a large site) can be a good thing - it helps you to gain links which otherwise would not be possible.

Let's consider one of the Mack Daddys of big sites - Amazon.com. Right now Yahoo Site Explorer is showing 148 million pages with 33 million inlinks. LOL. Be still my heart. (Amazon)Baby got back(links)

OK, here goes again...

Though Amazon is HUGE, if you look at it as having a 148:33 page to inlink ratio instead of simply showing monster pages and linkage, the mind is less boggled.
As I recall, it has, in its way, grown gradually over the years.
Some of those 148 million pages will get a little fresh content boost whenever a review is added.
It's got content broad-ish (books under $25 or whatever) and targeted (one exact book) arrayed with lovely buckets and hubs.
Some inlinks are affiliates, and some are wish lists (what a cherry idea that is for link-building!) and some are recommendations of products from a site that has become an easy resource, like a dictionary of commercially available books - search for the book you want someone to see and as often as not they'll have it if it came from a major publisher, even sometimes if it's from a smaller house.

My intuition, if I were an aspiring Mack Daddy, would tempt me to add new pages a chunk at a time and with fanfare, to make use of any new content boost while link building. If I didn't have new linkworthy or fanfare worthy content and I was already meeting existing viewer public service need I'd probably want to hold off on growth until growing both content and a public perception of need and needs met. An Amazon example is any of their services that encourage user involvement - wish lists and reviews, and sell-here-too stuff like seller search and affiliate sales, effectively making allies of potential customers of who knows how many drop shippers. Very clever.

Some of this - like the authority of being a place to link to almost any commercial book - is the stuff of already having achieved a Mack Daddy reputation. Some ideas, like featuring Bill Mayer doing art and culture "fishbowl" segments that happen to feature products, these could be translated into useful features for small to mid-sized sites on up.

Instinct and ideas that feel linked to audience I got. Someone who plays data like a fine violin may have an instinct for wiring in some balanced formula for whatever makes sense in terms of how much content to add at once -- I think. IMHO, looking at either the stickiness or the volume without the other won't lead to satisfaction.

[JohnMu]

If you're interested in pagerank and want some examples, you might want to take a look at Bob Mutch's Page Rank Explained page and Phil Craven's Page Rank Calculator -- always good to get the creative ideas flowing :-).

Bob played with different site structures with the identical inbound link value. I'm not certain if that's realistic, but then again, making a large site does not mean that you'll automatically attract more (or better) links .

John