Please check my search engine friendly URL

Converting query string URL to search engine friendly.

[demello]

I've been having problems getting any of my database generated pages on my site get indexed in Google or any other search engine. My site does not make use of cookies or session IDs. After much research it is clear that this may well have been due to the size of my query strings and perhaps the fact that I was using a CartID value with a long number.

I have attempted to address this problem by using the following solution:

1. Use search engine friendly URLs which replace query string characters "?" and "&" with "/" which then get processed back to query strings by use of a server side ISAPI filter.
2. Create a site map containing links in the above format to the database driven pages.
3. Ensure the links in the site map do not use the long CartID value.

Before re-deploying the site, I would like feedback as to whether the new query strings look ok and whether I should refine them further. I am particularly concerned as to whether the URL is still too long and whether the fact that I use a variable called CategoryID is going to cause problems. I understand from previous posts to avoid query strings variables of "ID".

To illustrate the changes better, below are 2 examples of the old format query strings which would not get indexed :

[MySite.co.uk...]
BackPageKind=PropertyListPage&
CartID=27105349165031087003224588027596

[MySite.co.uk...]
BackPageKind=PropertyListPage&BackCategoryID=39&
CategoryName=Camberley&
CartID=27105349165031087003224588027596

The above have now been changed to the following within my new site map page :

[MySite.co.uk...]
CategoryID/39

[MySite.co.uk...]
PropertyID/45/CategoryID/39/CategoryName/London

I would appreciate any comments if I have done the right thing. Should I reduce the URL size further?

Many thanks

[edited by: Marcia at 9:40 pm (utc) on July 30, 2003]
[edit reason] Inserted line breaks for sideways scrolling [/edit]

[Daryl]

Well you're certainly on the right track!
But I think your structure is still too deep. I personally haven't seen the Google spider(s) any deeper than 3 levels (www.yoursite.com/htm/product/prodID).

You could always go the trusted feed route.

[andrewrab]

Hey there...

in terms of depth, don't mistake depth of the URL (with regards to forward slashes) with depth of links from the home page...

I, personally, utilize an ISAPI.dll that uses alphanumeric character replacement in place of directory naming structure (e.g. '/') and while I can't, then, necessarily state that using multiple forward slashes is either better/not as good/or the same, I feel confident that your focus should be on depth of links from the home page and NOT worry about the actual URL.

With regards to the past issues associated with long, dynamic URLS, that had nothing to do with the URL lenght, but instead, with the number of parameters associated with the queries utilized.

As such, you should be able to have long URL's to your heart's content as long as they link within three steps (and preferably two) from the home page.

[eztrip]

Hey andrewrab,

Can you explain what you mean by "depth of the URL (with regards to forward slashes) with depth of links from the home page"

If I have a link from a sitemap which is linked from the home page that links to a subpage like:
/dir1/dir2/somepage.html

is that "worse" then having the sitemap point to just somepage.html?

Is it spidering that is the problem or is it Page Rank reduction or both?

Thanks
Mike

[Jesse_Smith]

::But I think your structure is still too deep. I personally haven't seen the Google spider(s) any deeper than 3 levels (www.yoursite.com/htm/product/prodID).

I see it do it ALL the time, when it's static, looking like a .html/.shtml/.htm and so on file.

[Rick_M]

[dmoz.org...]
Acoustics,_Ultrasound_and_Vibration/Ultrasound/
Resources/Acoustic_Material_Property_Tables/

seems to be indexed just fine! I agree - the number of subdirectories is not important. It is how many links you have to follow from the high pagerank site that matters.

Back to on topic - I'm really trying to find a way to get my CMS to make search engine friendly URL's. I have found solutions for phpnuke and postnuke, and I'm wondering if people can recommend other CMS systems, as mine is great with getting the dynamic links spidered, but I think my SERPS are suffering. I am basing this on the fact that I converted my directory from static to dynamic, but left the static directory up - after around 1 year, my static directory still often gets selected for the SERPs, even though I've tried to get all my backlinks updated to the dynamic page. I think I would have been better off just keeping the static directory, even though it was much harder to update.

[edited by: Marcia at 2:19 am (utc) on July 31, 2003]
[edit reason] Line breaks to fix side scrolling. [/edit]

[andrewrab]

Hi all...

clarification was provided in the last post by Rick_M... essentially, what he and I mean is that the URL itself is not the indicator of depth... but instead, it's how many jumps from the home page that is the concern...

As such...

Home --> Link 01
Link 01 --> Link 02
Link 02 --> Link 03
Link 03 --> Link 04
Link 04 --> Link 05

... that is bad... five 'steps' to get from home to link 05... regardless of the location of that link (in terms of directory levels.

For further clarification...

[xyz.com...]
[xys.com...]

Those are FINE... in fact, not actually (in terms of uniform resource locationing) not differentiated from one to the next.

Further...

[asdb.com...]

... is ALSO fine... though there is 'SOME' anecdotal experience that alphanumeric characters in place of the traditional query strings is still somewhat better, even today.

[sdkdf.com...]
=pasta&dog=crap...

THAT IS WHERE the problem resides... much more so BEFORE than today. The reason for that is NOT the number of characters or slashes or anything else in the URL... it is quite simply the number of multiple parameters (and the requisite 'overhead', like session id's, etc. associated with it) being utilized in the query.

So that, especially, is where an ISAPI.dll will work...

Now, for Rick_M... that is what you want, either (1) an ISAPI.dll (write your own or I can send you to a free one that works fine, or (2) dynamically WRITE .html/.htm/etc. pages. The first is easiest but requires either access to IIS, or ability to install custom dll's or both. The second one is more complicated and requires much more effort.

You may, in fact, be experiencing a problem with duplicate content, served both by dynamic and static pages... as long as your links (or most of them) are fixed (and have been for some time) you should be okay...

HOWEVER, that is also the problem you CAN run into when using an ISAPI.dll... why? because it interprets query strings (e.g. '_A_' works instead of '&') But does not require you to hardcode it (which is easy in and of itself)... the net effect is that if you DON'T hard code and don't take extra precautions to change the query strings in your pages, then BOTH types will be working and running side by side, potentially producing duplicate content.

Anyway, if anyone needs any mroe help or the dll... just pm me.

eztrip... yes, you have that right... now I will say to everyone that with my partner we've pulled off some outrageous rankings and I've ALWAYS used ISAPI.dlls to do it (even before it became popular)...

Additionally, you can find examples of 10, 15, 20 levels deep (which again, is a misnomer because they can all be in the root) from ISAPI.dlls that utilize '/' rather than alphanumerica characters.

I could get a site in within 10 days (though the last week has been weird!) that went 30 'levels' deep... but, that would be stupid because it would be poor organization for a website in my opinion.

The one .dll that uses '/' would convert this:

[xyx.com...]

to the following (or its general equivalent)...

[xyz.com...]
id/543343/type/4673747/color/red/weight/1847 -- it's something like that anyway... but I prefer the dll I use which would create:

[xyz.com...]
543343_a_type_e_4673747_a_color_e_red_a_weight_e_1847...

Either way, those URLs are independant of the directory structure and will not impact your rankings... I ought to know, I have enough that are that long that rank extremely well.

Bye all!

See ya.

[edited by: Marcia at 4:32 pm (utc) on July 31, 2003]

[Net_Wizard]

I'm really trying to find a way to get my CMS to make search engine friendly URL's

Rick,

Have you tried pMachine? The new version have an option for search engine friendly URL.

Cheers

[demello]

Thank you all for your responses. To summarise I have picked up the following points :

1. If at all possible avoid URLs more than 3 levels deep although others have not had problems with this.
2. Use an ISAPI filter which converts the query string "&" character to another alphanumeric combination instead of another "/".
3. Beware of duplicate content although I do not see this as a problem as it will only be the links in the site map that are converted to be search engine friendly, the main site links will still have the long query strings and so are not expected to be indexed.

Now, can anyone out there recommend a FREE or cheap ISAPI filter that does the trick. I understand Andrewrab offered to send a free one. I would prefer a filter that would not change the existing query strings but only specially coded ones in the site map page.

Cheers

[demello]

I thought of another possible solution and that is to merge many query string values into one variable such that instead of :

[MySite.co.uk...]
PropertyID/45/CategoryID/39/CategoryName/London

I would instead use :

[MySite.co.uk...]
PropertyIDX45XCategoryIDX39XCategoryNameXLondon

I could reduce the length of the URL even further by cutting down the size of the variables as follows:

[MySite.co.uk...]
XCatIX39XCatNXLondon

As you can see I simply used the "X" character as a separator but could have used other ones like ":" for example. The advantage of the above is that it appears to be only one level deep in terms of perceived directory levels.

Any comments on the above solution, in particular can I use other characters like @,#,~,: or others as separators? Also will reducing the size of the URL as in the last example be an advantage?

Thanks in advance.

[edited by: Marcia at 4:33 pm (utc) on July 31, 2003]

[Daryl]

Okay,
The level of technical expertise posted here humbles me, but I'm still not convinced that pages that deep (be it through links or file structure) rank well in Google(outside of Amazon.com of course:)

I’ve done a number of searches trying to find support for what’s posted in this thread and have come up empty. Could one of you please give an example search term that demonstrates deep pages can rank well in Google?

Thanks for being gentle to my over simplistic reply!

[andrewrab]

demello -- that won't be graceful... you need a unique identifier for each parameter e.g. '&' = '-a-' and so on and so forth, otherwise the coding side of things will be too obtuse. There are only 4 you need to worry about. And no, don't use special characters.

Daryl... I'll see if I can find anything... the reason you don't see much is because unless it's using an ISAPI.DLL, a directory structure that deep just isn't well organize, so rather than SEO-poor, it would be webmaster/organization-poor... but I know where to go to look... I'll send it when I have a chance.

[Marcia]

Note: When typing in those long URLs, it helps to prevent wide sideways scrolling to insert line breaks for readability.

[Daryl]

Thanks andrewrab I look forward to it.

Amazing what you asp types have to go through, creating dlls and all. Makes me appreciate Apache/PHP!

[demello]

Andrewrab and others,

Leaving aside the coding implementation side of things, I am purely interested in your opinion about whether my new formatted URL appears search engine friendly at least as far as Google goes. I have re-examined my URL and I can reduce it down to the following and still retain all the query string parameters I need :

[MySite.co.uk...]

As you can see, this reduces the URL down to 3 levels. Do you see any problems with getting this indexed in Google?

For your information, the values can be easily extracted from the string by use of the VBScript SPLIT function which can be made to split at "-x-". The values are then passed into a single dimensional array and easily extracted.

Cheers

Vince

[andrewrab]

demello.... all things the same (e.g. if your seo is as good as ours!)... you'll crank with that....

This comes from experience, not conjecture.

Good luck!

Just bear in mind one thing... IF those content pages have previously been served up and spidered with traditional query strings (e.g. '&' and '='), then you may have issues with 2 URL's having the exact same content...

As such...

search Google xyzdomain.com -asdf to track down any potential issues...

THEN, well... I'm sure you know what to do after that, primiarly with robots.txt and remove URL at Google.

See ya.

[edited by: andrewrab at 7:09 pm (utc) on July 31, 2003]

[demello]

Hi Andrewrab,

Thanks for your response, I am a little more worried now especially about the last "ps" paragraph.

First of all regarding the duplicate entries, the original pages have never been spidered and are unlikely to be spidered as they simply have too many query string parameters and one of them contains a long cart ID value. I am therefore confident that the links in the site map are the only ones to be indexed.

Please elaborate about your "ps" response. I can use "#" in place of the "-x-" separators, would this help?.

Thanks

Vince