Hoping to pick the brains of some old pros!

RodG:

google.com - 8 of the top 10 results returned do NOT have the "Wurldbitzer Debibulator" keywords in the URL

Purely anecdotal of course ;), but it doesn't look good for the theory that 'Keywords in the URL' are beneficial for Google rankings. If I were to make a conclusion at this stage (I'm not), I'd have to conclude that the **evidence is suggesting that the keyword loaded URL's are performing noticably worse than those without the keywords. **

Or perhaps Lankeeyankee and feznizzle are just better at SEOing sites than we are.

Take you, RodG. Insofar as I see, you have more "ancillary links" than anyone. Yet right now, my google server return shows that BOTH of your test products are toward bottom of page two. So how does this tend to indicate that keywords in the url are doing worse? (And I've never noticed you higher than page one bottom.)

I think another tentative conclusion is that having many "ancillary links" DOES NOT push one to the top. If that were true, we'd expect to see both of RodG's test products at the top.

Another tentative conclusion is that, for some reason, dynamic urls do produce alot of "ancillary links".

My three "pretty url" sites have been holding steady mid page one. I don't yet consider that a failure. I recently added a CEON entry, but although google has indexed a page or two, it has not yet found the test product:

http://webcache.googleusercontent.com/search?q=cache%3Awww.prom-mart.com%2Fdebibulators%2Fwurldbitzer-debibulator&sourceid=ie7&rls=com.microsoft:en-us:IE-Address&ie=&oe=

So last night I installed the sitemapxml plugin and this morning a robots.txt with the sitemapxml in it. Not yet in webmasters "sitemap submit." I want to see how CEON performs, if like Feznizzle's. Before I start adding more text or something.

Returning to Lankeeyankee. I am assuming that he went all out with his SEO knowledge on drumfactory, so when google finally finds and indexes his test product I will not be surprised if it goes to the top.

But with sabianstore, it looks like that probably isn't particularly optimized, being more of a test site. So it has crossed my mid that perhaps he's up there because of something else...even the Rubik template. Afterall, sabian store is above even Fez's dynamic url entry bizesite. Why I'm think it could even be the template.

I have a Rubik template entered on one of my subdomain "demo" entries, and if and when google finds it will be curious as to any result.

Perhaps the main conclusion so far, RodG, is that we're not that great at SEOing a site? ;)

EDIT: I also suspect that if Fez's CEON entries were still involved that they'd still be on top making more "pretty urls" on page one. That would be a good contest: Fez's CEON site vs Lankeeyankee's dynamic url Rubik template site.

niccol:

Some factors that may influence the results :

• pages do not have the same domain name
• the domain may ( or may not ) have some google ranking
• the domains have all existed different lengths of time
• the content on those pages is different ( this is my favorite by the way )
• a significant length of time has not passed so this may just be noise
• various people may have visited those pages
• various people may have clicked on the link to those pages in the search results
• the list could go on an on

My point is not that this does not prove what you are saying it proves. My point is that it proves nothing.** Unless you eliminate some of the variables listed above then the experiment will continue to mean nothing**.

Really, Google say that they have over a thousand factors that go into their algorithm. You really have to start to understand that you have to reduce the number that are present to be able to draw any reasonable conclusions. Any one of the factors that I have listed above could determine what you are seeing and the truth of it is that they probably all have some influence. I could list scores more factors that might be, and probably are, in play.

So if you are really trying to investigate the effect of keywords in URLs then you need to improve your testing environment. As I have said before my way of doing this which is kind of industry standard is to use matching sub-domains. In the past I have named them something like aaaaaaaaah.domain.com and aaaaaaaaai.domain.com . **That is to try and reduce the influence of the actual domain name. ** Obviously I have tried to create them at similar times. And I have created them on domains that have no history in google. All of that is just to try and eliminate the effect of domain names. It is not perfect but it is a lot closer to perfect than what is going on now. I have taken similar measures to isolate the factor that I am looking at with regards to content, links and any other factor I can think of. One has to do that otherwise the results and observations are not valid.

What you are going for is causality. That is the holy grail. What it means is that you are trying to demonstrate that, for instance, 'keyword loaded URLs benefit ranking'. ** If anyone can say 'but couldn't that effect have been due to an alternate factor' then your results do not really pass the test of scientific rigour** ( trust me I have experienced this in the scientific world as have many people. that is the benefit of scientific peer review. it asks the difficult questions )

So, for example, in a peer review process someone will ask 'are you sure those results weren't dependent on variations in ambient temperature?' If your answer is 'no, I eliminated the consideration of ambient temperature during the experiment in the following way' then all is good. If your answer is '* I guess so, maybe*.' Then not so good. I know this for a fact because it happened to me in my M.Eng oral examination which is more years ago than I care to think about. They told me to go away and redesign the experiment that I had been running for six months. ( It was about the stresses of structural elements when subjected to dynamic wave loading - if anyone cares)

So, Rod, if you were defending your results I would ask you, 'Are you sure that the effects you are seeing are not due to the different content on the pages?' If you are honest then I think you will say 'I don't have the information to prove that one way or the other at the moment. Because I have not actually analysed the content differences and/or we have no mechanism for excluding the effect of those differences on the results.'

At which point I say 'Go away and redesign the experiment'.

You can take that advice or not as you please.

Right on Niccol!!! the points I highlighted in your post mirror my own thoughts. These are the reasons why I maintain that because of how this "test" is being run, that any responsible siteowner/site designer cannot walk away thinking that anything was really proven here. This "test" lacks the proper controls to provide any real MEANINGFUL data. Sites with different domains.. Sites with different content (not everyone could even agree on the so-called control content and varied their "entries") Some sites are established domains, other are not.. Too many variables to make any meaningful conclusions here. For example SPH's repeated inquiries about the role of the template in seach engine results speaks VOLUMES about the lack of controls that abound in this so-called "test". Like you I hope that newbies don't stumble across thread this thinking that anything is to be gleaned from the "test results".

But I do continue to be entertained by it all..:laugh:

And to reply to or edit myself:

SPH:

I think another tentative conclusion is that having many "ancillary links" DOES NOT push one to the top. If that were true, we'd expect to see both of RodG's test products at the top.

On the other hand, sabianstore also has a ton of "ancillary links". So if there is ANY truth to the theory that many "ancillary links" help push the main product to the top, that could also be a factor with sabianstore's current high ranking. Along with the Rubik template. ;)

Drumfactory also shows a ton of "ancillary links" which may help boost it's test product when google finds and indexes it. (I don't yet see it's test product in the listings.)

SPH:

Another tentative conclusion is that, for some reason, dynamic urls do produce alot of "ancillary links".

And this just might turn out to be the real benefit with dynamic urls. If they do churn-out more "ancillary links" than do "pretty urls", and if these "ancillary links" help boost the product, this would be the real reason to go dynamic url.

And, of course, some of these sites are going with different product descriptions. My prommart.com (ultimate seo url) and lankeeyankee's sabianstore (dynamic urls) I believe are the only two entries with this exact description. Drumfactory has yet another description.

But as I noted the other night: right now dynamic url sites rank highest for either of the two descriptions now indexed for the test product. Lankeeyankee's sabianstore with one of the new descriptions and Fez's dynamic url entry "bizsite" with the original description.

RodG:

I was hoping to see the KeyWord laden URL's to take ~8 of the top 10 spots, which would be in accordance with the theory that KeyWord laden URL's do offer ranking benefit.

Of course, there would have to be 8 or more "pretty url" sites in the test to produce that many page one results. And "no", I'm not going by "ancillary links." I'm looking for final ranking on the test product itself. My observation is that when a site's test product does hit, it goes ahead of all "ancillary links" from that site. And test products from other sites tend to move ahead of "ancillary links" from other sites.

I only began with three sites - two Ultimate SEO and one Simple SEO URL. All three of mine have been holding well mid page one so far.

Feznizzle had TWO CEON entries top spots, but he pulled out. So we went from five back to three.

I subsequently entered two more Ultimate SEO and one CEON entries. Google is still looking for my two ultimate seo entries, and has yet to find the test product on my CEON entry.

So that potentially makes six page one placements for me. :)

Too bad Fez pulled out.

@ Feznizzle,

Can you re-enter your CEON entries? It would be interesting to see if they can achieve same ranking as earlier, considering the "history" other sites now have.

FOUL! sabianstore has two top ranking listings for the same product: one without and one with the "en". Not to mention the category/single product redirect. Plus an exact match category name with the test product. :wink:

are these ancillary links not produced by google seeing the product in a sidebox or cross sell nothing to do with the url ?

bn

bn17311:

are these ancillary links not produced by google seeing the product in a sidebox or cross sell nothing to do with the url ?

bn
100% spot on. They have nothing to do with URL

cheers
Rod

SPH:

And,** of course, some of these sites are going with different product descriptions**. My prommart.com (ultimate seo url) and lankeeyankee's sabianstore (dynamic urls) I believe are the only two entries with this exact description. Drumfactory has yet another description.Which only serves to highlight the issue with coming to any meaningful "conclusions" from this "test".

As niccol put it: "Are you sure that the effects you are seeing are not due to the different content on the pages?' If you are honest then I think you will say 'I don't have the information to prove that one way or the other at the moment. Because I have not actually analysed the content differences and/or we have no mechanism for excluding the effect of those differences on the results.'"

SPH:

But as I noted the other night: right now dynamic url sites rank highest for either of the two descriptions now indexed for the test product. Lankeeyankee's sabianstore with one of the new descriptions and Fez's dynamic url entry "bizsite" with the original description.Proving what??? Given that this "test" and it's (lack of) controls, varying data, etc, what exactly does this show?? How do you know that these results aren't affected by (as niccol aptly put it): "the domain may ( or may not ) have some google ranking" or other factors?

SPH:

Too bad Fez pulled out.

---

@ Feznizzle,

Can you re-enter your CEON entries? It would be interesting to see if they can achieve same ranking as earlier, considering the "history" other sites now have.I have serious doubts that Feznizzle will be participating in this "test" anymore..

DivaVocals:

Which only serves to highlight the issue with coming to any meaningful "conclusions" from this "test".

As niccol put it: "Are you sure that the effects you are seeing are not due to the different content on the pages?' If you are honest then I think you will say 'I don't have the information to prove that one way or the other at the moment. Because I have not actually analysed the content differences and/or we have no mechanism for excluding the effect of those differences on the results.'"

One thing the various descriptions MIGHT have shown is that google really doesn't like our original "gibberish" description as much as lankeeyankee's more normal text, even though the "gibberish" description contains much more text/content.

But we can ignore the sites with different description text from the original "gibberish". That would be lankeeyankee's two sites - sabianstore and drumfactory (dynamic urls) - and my prommart.com (Ultimate SEO URL). Doing that might even put RodG on page one. :smile:

Proving what??? Given that this "test" and it's (lack of) controls, varying data, etc, what exactly does this show?? How do you know that these results aren't affected by (as niccol aptly put it): "the domain may ( or may not ) have some google ranking" or other factors?

Maybe not pure science, but I think useful for various purposes. Sure, I've suspected that my prommart.com does have a google disability, but I'm happy to see it made it. And we can even get in to differences with different carts or platforms. I have a Prestashop and Opencart (fairly recent but not most recent versions, both non-production demos) that I may take the liberty of tossing our original test product on tonight...to see what might happen. What would it possibly tend to "prove" if they happened to move to the top?

I have serious doubts that Feznizzle will be participating in this "test" anymore..

I'd like to see him reenter, though. We know his sites were impressive while in this test. I wonder if that can be repeated, now that there are many more listings. Could potentially tell us that it's best to get one's products up first, before your competitor?

Diva, do you have a CEON site or two, hopefully that google likes, to enter with our original product? Or perhaps or original product and one of the new descriptions?

SPH:

One thing the various descriptions MIGHT have shown is that google really doesn't like our original "gibberish" description as much as lankeeyankee's more normal text, even though the "gibberish" description contains much more text/content.

But we can ignore the sites with different description text from the original "gibberish". That would be lankeeyankee's two sites - sabianstore and drumfactory (dynamic urls) - and my prommart.com (Ultimate SEO URL). Doing that might even put RodG on page one.

Maybe not pure science, but I think useful for various purposes. Sure, I've suspected that my prommart.com does have a google disability, but I'm happy to see it made it. And we can even get in to differences with different carts or platforms. I have a Prestashop and Opencart (fairly recent but not most recent versions, both non-production demos) that I may take the liberty of tossing our original test product on tonight...to see what might happen. What would it possibly tend to "prove" if they happened to move to the top?If you say so.. rock on...:bigups:

BTW, though it's been explained more than once, Google doesn't know "gibberish" from German.. You are hell bent to make this claim and though several folks have explained why this isn't true you are like a dog on a bone in pushing forward the notion that Google somehow can tell "gibberish" from German or English..

SPH:

I'd like to see him reenter, though. We know his sites were impressive while in this test. I wonder if that can be repeated, now that there are many more listings. Could potentially tell us that it's best to get one's products up first, before your competitor?Yeah.. I think he's done.. Good for him..

SPH:

Diva, do you have a CEON site or two, hopefully that google likes, to enter with our original product? Or perhaps or original product and one of the new descriptions?
HE|| NO!!!:laugh: I am not DRUNK:cheers: or HIGH:smoke: enough to consider entering this "test"!!!

Sorry... I know you think this "test" is useful, but I don't. I have no plans to enter any site into this "test" because I echo Niccol's sentiments on the methods used to conduct this "test" making any conclusions guesses at best. Based on the inconsistencies of this "test", I maintain that a responsible siteowner/designer cannot seriously walk away from this "test" thinking that they really gleaned ANYTHING useful.. and besides all that, I sure will NOT enter any of my client's sites into this mix, nor would they allow me to. They are all doing very well on their own (excellent steady sales) and I won't risk that on this "test".

DivaVocals:

You are hell bent to make this claim and though several folks have explained why this isn't true you are like a dog on a bone in pushing forward the notion that Google somehow can tell "gibberish" from German or English..

I see the results, but yea, they may be inconclusive. And not pushing anything. As I've also said: if "gibberish" is okay with google, that will make writing descriptions for thousands of products much, much easier.

BTW, why so mean!

HE|| NO!!!:laugh: I am not DRUNK:cheers: or HIGH:smoke: enough to consider entering this "test"!!!

Sorry... I know you think this "test" is useful, but I don't. I have no plans to enter any site into this "test" because I echo Niccol's sentiments on the methods used to conduct this "test" making any conclusions guesses at best.

Then forget the test vis-a-vis urls. Why not test and refine your SEO skills*?

Here's the mission: you and Niccol sell, or have clients who want to sell, Wurldbitzer Debibulators. Now, do whatever you have to do - metas, great descriptions, etc. - to get to the top in this little internet world we have created. If you make it to the top, I/we can respond.

You're thinking too small re the possibilities here.

• There are alot of SEO experts around. This little "test" and product can put that to the test, possibly going out on a limb.

PS: I think that lankeeyankee is already doing some other, independent, testing with his two descriptions.

SPH:

I see the results, but yea, they may be inconclusive. And not pushing anything. As I've also said: if "gibberish" is okay with google, that will make writing descriptions for thousands of products much, much easier.

BTW, why so mean!Not mean at all.. not interested in participating in this "test" because the manner in which you all are testing can only yield a flawed result which I find wholly useless.

SPH:

Then forget the test vis-a-vis urls. Why not test and refine your SEO skills*?

Here's the mission: you and Niccol sell, or have clients who want to sell, Wurldbitzer Debibulators. Now, do whatever you have to do - metas, great descriptions, etc. - to get to the top in this little internet world we have created. If you make it to the top, I/we can respond.

You're thinking too small re the possibilities here.

---

• There are alot of SEO experts around. This little "test" and product can put that to the test, possibly going out on a limb.

PS: I think that lankeeyankee is already doing some other, independent, testing with his two descriptions.
Yeah.. No thanks.. Still not high or drunk enough.. :P I would rather test and refine my SEO skills in my own properly conducted test without all the noise that I think is being generated from this uncontrolled one. When my time permits I may do just that. (and NO if I do conduct my own test, I won't be using Wurldbitzer Debibulators as my test product -- the market for this product is too crowded:laugh:)

bn17311:

hey rod

just wondering how many sites joined in test how many used standard zencart url

bn

Congratulations!

Of all the possible 'variables' that may or may not influence the observable trend, this question is by far the most pertinent & relevant (and it doesn't even have anything to do with the 'variables' at all).

I'll even go as far to say this is the only thing that would render the observation useless. Without this information, the observed results can be explained by the simple fact that there are approx 5 times as many links (not sites) using the zencart url (no keywords) than there are using the keyword laden ones.

If this is indeed the case, the conclusion would then be that there is no observable difference between the 2 URL types.

The 'conclusion' (that I stated I didn't come to, for this very reason) would only be valid if there were an equal number of keyword & non-keyword laden URL's.

Your question clearly demonstrates that you understand this very simple/basic principle. Shame I can't say the same about some others. :(

Cheers
Rod

RodG:

Congratulations!

Of all the possible 'variables' that may or may not influence the observable trend, this question is by far the most pertinent & relevant (and it doesn't even have anything to do with the 'variables' at all).

I'll even go as far to say this is the only thing that would render the observation useless. Without this information, the observed results can be explained by the simple fact that there are approx 5 times as many links (not sites) using the zencart url (no keywords) than there are using the keyword laden ones.

If this is indeed the case, the conclusion would then be that there is no observable difference between the 2 URL types.

The 'conclusion' (that I stated I didn't come to, for this very reason) would only be valid if there were an equal number of keyword & non-keyword laden URL's.

Your question clearly demonstrates that you understand this very simple/basic principle. Shame I can't say the same about some others. :(

Cheers
RodOh yes.. what a shame..:no:

niccol:

My point is not that this does not prove what you are saying it proves.

My point is that you somehow read things that haven't been written. I never said it proves anything (with good reason).

This is what I actually wrote: word for word. No additional highlighting or emphasis.

If I were to make a conclusion at this stage (I'm not), I'd have to conclude that the **evidence is suggesting that the keyword loaded URL's are performing noticably worse than those without the keywords. **

I didn't use the word 'proof'. I stated 'the evidence is suggesting' - A totally different meaning.
I also stated quite clearly that I was not making any conclusions at this stage.

I've no idea how you've managed to twist so few words so much that they have a totally different meaning. Please stop doing that. It is annoying to me, and destroys your own credibility.

I was so pleased when bn asked the** only** pertinent question that would invalidate the observation based on the information provided. Everything else is irrelevant.

If you ask nicely, I'm sure bn will be able to explain this too you. No point in me even trying. All I will say is that bn realises the value of 'normalization'. Not saying you don't, just saying that http://esl.about.com/library/glossary/bldef_130.htm explains the problem quite nicely.

Rod

DivaVocals:

Not mean at all.. not interested in participating in this "test" because the manner in which you all are testing can only yield a flawed result which I find wholly useless.

So why are you still here?

Rod

DivaVocals:

Yeah.. No thanks.. Still not high or drunk enough.. :P I would rather test and refine my SEO skills in my own properly conducted test without all the noise that I think is being generated from this uncontrolled one. When my time permits I may do just that. (and NO if I do conduct my own test, I won't be using Wurldbitzer Debibulators as my test product -- the market for this product is too crowded:laugh:)

This little one-product niche market is, I think, a very good way to test and refine SEO skills and theories. Either independently or participating with others who come along.

Similar to long-distance chess or strategy games.*

But yea, I can see how one might not want to participate. :wink:

*EDIT: If there's any money in something like this, I want a cut!

RodG! I now see BOTH of your products on page one. What are you doing? Possibly doing side-testing which has been holding you back somewhat?

Anyway, great job!

SPH:

This little one-product niche market is, I think, a very good way to test and refine SEO skills and theories. Either independently or participating with others who come along.

Similar to long-distance chess or strategy games.*

But yea, I can see how one might not want to participate.

---

*EDIT: If there's any money in something like this, I want a cut!Uh huh.. Okay..

RodG:

So why are you still here?

RodWhy?? are you gonna get the bouncers and force me to leave???

So to answer your question.. Why am I still here??

A. Because as long as I am not violating any forum rules, I can read/view/respond to ANY thread I wish to.. I will follow this thread until.. well... well until I feel like NOT following it anymore or until it dies (which ever comes first).
B. Because though I think this "test" is LOADED with faulty data/reasoning/controls and faulty findings as a result, I have to agree with niccol in that this is one of the most entertaining threads I've read in a while on this forum (possibly EVER)..