Directory Structure

[Visit Thailand]

I have noticed that similar pages on one of our sites have different PR.

So I would like to ask is there any difference between a page in the main root directory and a sub (one level down) directory ?

Should I always place a index.html in each directory especially if it contains important .html pages ?

[ciml]

I wouldn't worry about directory structure from a PageRank point of view, it's the link structure that matters if most of your incoming links go to the home page..

For a lot of sites, the URL structure and link structure are similar, which is why it _looks_ as though directory structure makes a difference.

[edited by: ciml at 2:42 pm (utc) on Aug. 1, 2002]

[Nick_W]

Yes, and won't google 'guess' a PR on pages it hasn't indexed, usually one point lower per directory level further adding to this myth.

I don't speak with authority here, it's only recently I really learn't this ;)

Nick

[electro]

As far as I can see, the real directory structure of a site DOES have an impact on PR, EXCEPT when a page is heavily linked to. For example, the classic sructure for a basic site would be:

index.htm (PR 5, for example)

- a-folder -
.......page 2 (PR4)
.......Page 3 (PR4)

.......- another-folder -
..........page 4.htm (PR3)

Providing that most incoming links goto index.htm. Going 'flat' (i.e. no folders, which is what I do), will usualy give you:

index.htm (PR 5, for example)
page 2 (PR5)
Page 3 (PR5)
page 4.htm (PR5)

Or sometimes:

index.htm (PR 5, for example)
page 2 (PR4)
Page 3 (PR4)
page 4.htm (PR4)

BUT... if you have folders and an internal page (lets say page 4.htm) with loads of inbound links, youll get:

index.htm (PR 5, for example)

- a-folder -
.......page 2 (PR4)
.......Page 3 (PR4)

.......- another-folder -
..........page 4.htm (PR5)

How this would pan out on a flat structure is hard to say

I just keep all pages in the same root directory, naming them like this:

atopic_index.htm
atopic_page1.htm
atopic_page2.htm
help_index.htm
help_faq.htm
help_contact.htm
index.htm

you get the idea.. so that you can find pages easily. This will probably work OK up to 100 or so pages, then will be too unweildy, you may want to use folders then.

[Visit Thailand]

What concerns me is if this is true and it seems to be that way then I am potentially losing page rank on pages.

At the moment I can create a page to this site (site rank 6 under a penalty I believe) and add it to the root directory and I get a 5 and it stays that way after indexing etc etc but if I put it into a lower level directory and the page pointing to it is in the root I then get an automatic 4 which is relegated later to a three.

As it is then a three it does not count as backward links.

I have seen other sites that include an index page in that directory and so the links are more local. I have not yet tested it but it seems to keep the PR better.

Any thoughts ?

[Visit Thailand]

Electro - thanks I posted at the same time as you but what you are saying makes sense. This site has thousands of pages and it all used to be in the root but was an absolute mess so we cleaned it out and put directories in.

But now some of our most important pages which are now in sub directories have lost PR value which is highly infuriating.

[Visit Thailand]

I am surprised no one else has any thoughts on this.

Do you think creating an index.htm in each sub directory would help and getting a higher ranked page to link to that ?

[bird]

What ciml wrote is the commonly accepted wisdom about this topic. I have never seen evidence to support any other speculations. If you make your directory structure so that it helps your visitors find what they want, then you have done the best you can.

[Visit Thailand]

Yes but I disagree with what CIML wrote as I have proof of it happening just as Electro says it does.

[bird]

I don't think you have proof of a direct causal connection between the directory structure and PageRank. What you have is an observation that they sometimes appear to coincide. The reason for this can also be found in cimls post:

For a lot of sites, the URL structure and link structure are similar, which is why it _looks_ as though directory structure makes a difference.

PageRank is exclusively about links. As long as you don't change the link structure, you can move your pages around as you like, and the relative PageRank values between them will always remain the same.

[Marcia]

I have to agree with ciml and bird. I just tried two experiments to prove it out with two sites that have links pointing to only the homepage from outside the site.

Site-A has several pages in the root directory, and 3 pages that are /directory/index.html. The homepage is PR5 and so are all the other pages, including the ones in the /directory/ which have only index.html so far, no other pages so far.

As a matter of fact, a while back the /directory/ pages were PR5 (I added those to the site myself and limited the linking) and the ones in the root were PR4 - those were in a totally different linking pattern, done by the person who made and maintains the site. They all linked a lot into the shopping cart and many other pages, there was a lot of dilution. My pages didn't. When I made some changes to the overall site navigation there was an upward change, but the 3 separate pages were PR5 from the beginning.

Site-B has several pages in the root dirctory and a couple of /directories/. The homepage is PR5 and so are all the pages in the root directory, and so are the /directory/index.html pages

However, on Site-B I deliberately put a page within a directory and linked to it from the homepage instead of the index of the directory to see this for myself.

homepage=PR5
/directory/otherpage.html=PR5 (linked to from the homepage)
/directory/index.html=PR4 - linked to from /directory/otherpage.html (PR5) but NOT the homepage.

One click away from the homepage is keeping the same PR, two clicks away is losing a point - regardless of where it is. I use directories a lot, always have, and I've seen this over and over. Whatever's linked to from the homepage will have the same PR as the other pages linked from the homepage, regardless of whether it's in the root or in a directory. If they're the same as the homepage they'll all be the same, if they're a point lower, they'll all be a point lower.

I prefer to order navigation by a logical structure, putting a section into its own directory if it makes for better site navigation for visitors. Keeping it all in the root and linking to them all is giving them equal importance and distributing the PR more widely, where in actuality they're not equally as important.

Doing it so it makes for a better, more meaningful site from a visitor navigation point of view actually reflects in a more accurate distribution of Page Rank, imho. If the visitor comes looking for green widgets, he may look at the privacy policy, but it's not as important to him as the page with the green widgets on it.

Logically speaking, the more important pages should have higher PR. I think setting up the navigation and directory structure are the hardest part of doing a site, even with small sites like I do.

[edited by: Marcia at 12:59 am (utc) on Aug. 2, 2002]

[Visit Thailand]

Thanks Marcia, I think perhaps I should explain this sites structure perhaps the explanation lies there.

I have outside links to may different pages within the site.

Some internal pages are 6 others 5 etc.

I create a page and upload to the root it automatically guesses it as a 5 and it stays that way.

Then I create a page which is in directory X (no index.html in this directory) and link that page from a page which is linked from a PR 6 page (which is linked from the home page) this new page is guessed at being a 4 but later becomes a three.

So in other words my home page (6) links to X.html (6 as well) which links to dir/newpage.htm (which starts of life as a guess 4 and later becomes a 3).

I am concerned about this as a three affects backwards links etc, and I prefer to have every page at least a minimum of 4 for this reason.

Any ideas ?

[ciml]

Visit_Thailand, the PR of the linked to page depends on how many links are on the linking page, and whether it is a high six or a low six.

For PageRank point of view, backlinks from pages less than PR4 do count, they just don't show in a link: search.

[electro]

I have ti change my tune here and say that I beleive what ciml and bird are saying, and that perhaps I was mistaken in my conclusion. I am still wondering if it is easier to get a good pagerank if pages are in the same directory... Also, an interesting experement would be to put a page in, say 6 nestled folders, and see how its PR changes, if at all.