SEO test: Subdomains vs. unique domains

P

phixx

New member
Jul 4, 2007
311
2
0
Hi,

I'm planning to do some SEO testing. My plan is to try 20+ approaches to making things rank.

Do you think I can just create 20 subdomains on the same domain - or am I better off buying 20 unique domains?

(problem with buying 20 unique domains is that I prefer to used aged domains. If I don't, I believe the risk of getting a penalty is to big)

Thanks.
 


CCarter said:
That's doesn't sound like a good idea. How do you have 20+ SEO approaches?
Click to expand...

Why is it not a good idea?

Finding 20+ different approaches isn't difficult. Just think different types of links, different amounts of links, etc.
 
Your idea will be fine.

Your subdomains won't impact/hurt the main domain if one of your ranking tests get penalized by google. ( I can't comment on the effect it could/might have on the shared IP though. I've had no problems so far, but I can see how there is a chance it could negatively impact the IP with Google if too many subdomains at that IP get penalized. If that were to ever happen to me, I'd just move that Priamry Domain and all subs over to another host with a CLEAN IP.)


BUT........the subdomains will be treated as a NEW DOMAIN and the AGE of the Primary domain will NOT transfer/carry over to the subdomains.


Unlike most on here who Poo Poo the use of subdomains, I feel that they are a good way to do a lot of things. Such as your idea to test out new strategies so as to NOT BURN a new or primary domain.



CCarter meant well in his post. He was NOT busting your chops. If you were to be building an Authority Site you would NOT want to do it with a subdomain or if your are setting up sites that you might one day want to sell, a subdomain would not be the way to do it.


I think he was more curios about about your "20 SEO Approaches". lol



I'm assuming you are counting, let's say....Keyword Stuffing as One Approach. Certain types of Link Building as another for examples.



You should be FINE using subdomains for what you have in mind.
 
The problem is... you are using 1 domain for 20 different test. I don't see how that's a controlled environment. I would have each test on a different domain, different hosting, and see what happens. Lets say you get 20 domains, and have 4 hosting companies. Divide up the way you are spreading out the test, so it's less likely to lead back to you. You might find, that your hosting company is a problem, or a shared IP is a problem, or a network lagg is a problem. But testing 20 variations at ONE time? NO ONE ON EARTH, DOES THAT! NO Scientists, NO one can handling that much data.

If you notice, whenever you see double blind studies or anything, they are testing out 1 variable. These people with billions of dollars, can't figure out how to test 3 or more variables, and you are going to come in and test 20?

Good one, Boss :updown:
 
CCarter said:
The problem is... you are using 1 domain for 20 different test. I don't see how that's a controlled environment. I would have each test on a different domain, different hosting, and see what happens. Lets say you get 20 domains, and have 4 hosting companies. Divide up the way you are spreading out the test, so it's less likely to lead back to you. You might find, that your hosting company is a problem, or a shared IP is a problem, or a network lagg is a problem. But testing 20 variations at ONE time? NO ONE ON EARTH, DOES THAT! NO Scientists, NO one can handling that much data.

If you notice, whenever you see double blind studies or anything, they are testing out 1 variable. These people with billions of dollars, can't figure out how to test 3 or more variables, and you are going to come in and test 20?

Good one, Boss :updown:
Click to expand...

Can't see the problem.

Test 20 different versions of SEO stuff. 1 version/subdomain.

See how they rank compared with each other.

As far as I can see, subdomains might actually give a better picture than different domains and hosts since different domains might have different histories, etc.
 
Alright, I'm going to try this ONE more time, and I want you to understand beforehand, I'm attempt to help you.

When you are testing something, you test, YES or NO.

You approach is erroneous. You are testing subdomain1, subdomain2, subdomain3, subdomain4, and so on through 20.

What happens if Something you did on Subdomain 13 takes down your WHOLE website, and therefore makes all your experiments null and void? How will you be able to tell what happened where?

Also, the reason I say different domains, and lower variables, is because you are attempting to find a magic bullet. Let's be blunt. You want to find a formula that will work more than 80% of the time. So putting it on 20 different domains and Going After 20 different niches, with 2 variations of the technique is going to get you closer to your objected goal. You are attempting to test 20 techniques on 1 domain, IN ONE NICHE. What works in one niche, may not work in another. What I propose you do, is attempt 2 or even 3 techniques, with 20 domains in different niches, so you can see what combination of things work.

Anyways, this whole thing is not going to work in the end, because there will be so many variables in an experiment with 20 variations or 2 -3 ,with 20 domains. You'll get a good understand of what does work, AT THE CURRENT MOMENT.

You are going to have to account for, what happens when one techniques pulls the main domain down or up, along the others. IF you believe, that each subdomain act and rankings completely independently, you are mistaken. You've got a shared IP, shared server, you've got a shared database, and if one subdomain's load is too heavy, it will effect the performance of the others, AND your overall efforts. So would you account for that, and track what's what?

This is why most scientists test 1 variables, and figure out if it's YES or NO. You are attempting 20.
 
Good point.

My plan was to do 2 x 20 versions, i.e. to each technique for 2 different niches.

I don't think the shared IP, shared server, etc. matters that much. After all, most sites are on shared IPs, servers, etc.
 
No... It does matter. Most small shitty sites are on a shared IP. No serious site is going to keep their site on a shared IP once they get traction. Also, what if subdomain13 takes down the whole experiment by banning the domain??? How would you know?
 
You must log in or register to reply here.
Top Bottom