First Post - Help with SE Strategy for Site Please

[Jack]

Hi,
I'm still learning about the SEO stuff and I would much appreciate your views on some decisions I have to make.
When I designed my site (still in development) my highest priority was best user experience. As it is a highly data intensive site (dynamic) I use a number of techniques to maintain state and thus minimise the number of database reads (minimise reads == maximise performance == best user experience).
I use frames, sessions and the query string - so a bit of a challenge for SEO.

I have already implemented freindly url's - so the SE's don't see any query strings (except perhaps for session id's - more of which in a moment).

I am currently dependent on sessions for the site to work. As I am using php sessions, if the user has cookies enabled there is no id in the query string. If not I can have php pass the session id in the query string. So, this is like being between a rock and a hard place, bots don't use cookies, but bots don't like query strings.

Question 1: What should I do about sessions?
a): do away with them altogether and accept many more db reads.
b): keep them but incorporate the session id into the freindly url, eg:
domain dot com/typeA/thingB/987596847398405843939474
c): check for cookies enabled (set cookie then reload and check) - then give cookie users (people or bots) a different experience to non-cookie users (is this cloaking?)
d): check for user agents which are bots and make it work without cookies just for them? (this is cloaking - yes?)
e): let the id's default to the query string as SE's will still work with them.
f): leave it as is because bots ought to be able to use cookies? (Note: I have previously been visited by 'Faxobot' which seemed to get down two levels of links in my structure, therefore, I think it must have used cookies - is this usual?).

Question 2: What should I put in 'No Frames'?
a): a full duplicate non-framed site which will not be as good a user experience but will be better for the bots?
b): just a load of keywords?
c): nothing because google says it can cope with frames (therfore it ignores 'no frames'?).
d): isn't putting different stuff in no frames just another method of cloaking - leading (some way down the line to the possibility of penalties)?

Thanks for any advice, Jack.

[seomike]

a): do away with them altogether and accept many more db reads.
b): keep them but incorporate the session id into the freindly url, eg:
domain dot com/typeA/thingB/987596847398405843939474

It wouldn't help to pass the SID in a mod. When the spider goes from page to page the SID will change making your site pretty much infinite when it comes to pages.

c): check for cookies enabled (set cookie then reload and check) - then give cookie users (people or bots) a different experience to non-cookie users (is this cloaking?)
d): check for user agents which are bots and make it work without cookies just for them? (this is cloaking - yes?)

This is the best route. and it's not cloaking. It's a user preference based on their technology settings. Plus you are showing the user or bot the same thing correct? :D

e): let the id's default to the query string as SE's will still work with them.
f): leave it as is because bots ought to be able to use cookies? (Note: I have previously been visited by 'Faxobot' which seemed to get down two levels of links in my structure, therefore, I think it must have used cookies - is this usual?).

If you don't want the bots to deep crawl or even come back then do this.

a): a full duplicate non-framed site which will not be as good a user experience but will be better for the bots?
b): just a load of keywords?
c): nothing because google says it can cope with frames (therfore it ignores 'no frames'?).
d): isn't putting different stuff in no frames just another method of cloaking - leading (some way down the line to the possibility of penalties)?

Don't do any of this.


You can have dynamic strings or mod rewritten strings. Google will crawl both as well as Inktomi (if the dynamic strings are not too long) As per Ask/Teoma and other smaller engines your strings will be pretty much invisible if you go dynamic.

It's good that you've gone static strings just make sure that you don't make your subfolders too deep. stick to 1-2 levels deep for better results when your site is crawled.

Here is my question. How important is the SID? Will your site just go to pieces without it? Could I view all your pages if I wasn't served an SID?

[Jack]

Seomike, Thanks for the comprehensive response.

To answer your last question first, yes - the site was completely depenedent on the SID, you couldn't have viewed the important pages without it.

I say that in the past tense as I have spent all morning working out how to become less reliant on SID and query strings.

With reference to my question 1 - your advice (c or d) - sounds right, and its re-assuring that you don't regard it as cloaking (exactly the same content would be served). Good point about b) - overlooked that. However, now I am now on a mission eliminate the SID for any 'browse only' of the main pages. This will be at a slight cost in terms of making a "query string 'directory' " a bit longer (though this is static) eg.:
I currently have:
domain dot com/typeA/thingB/[short codified internal info]
... it will become:
domain dot com/typeA/thingB/[a bit longer codified internal info]

I think this is a worthwhile trade off.

If I am not using session I will also have to put longer (very long!) real query strings in parts of the system where the bots don't really need to get to - I assume the bots won't penalise me if I have some difficult (query string) links, (and some javascript links) as long as there are some links (the important ones) which are freindly to them. Is this true?

Your comment on my Q2 is music to my ears - if the main bots will follow my modified urls through the frameset/frames, then this will save me a lot of time doing 'no frames' stuff.

Thanks, Jack.