Forum Moderators: open
What are the chances of getting banned from cloaking? I know Air's response earlier was that it depends how good you are, and even if you get caught it might not be a bad thing. My problem is, obviously I'm not to good right now, since I'm even still vague on the whole process.
Maybe one thing that would help me get a better grasp would be to know the pitfalls. You know, what do I do and what do I definately NOT do?
Make some practice pages and host them on the experimental domains and see what you can do.
>What are the chances of getting banned from cloaking?
My admittedly limited experience suggests that chances of banning are low to very low if the cloaked, optomized pages are relavent to the "real" site. Spam is spam.
My own experience this summer convinced me that Inktomi will bury your listing but not ban you if they detect cloaking.
The real experts will be along shortly! :)
Welcome to WMW too!
Make sure your targeted keyword phrases are relevant to you content (Don't fudge an inch here). Make sure whatever software you use is comprehensive from the standpoint of the ability to add IP's and UA's on the fly. Also, a nice software feature capability is to be able to over-ride UA, and do IP delivery only for engines such as FAST, which uses an incremental UA, I think they are up to "Pre-23" or so, at this point.
AV is the only SE that we know of doing "serious" cross-checking from mibh.net IP's, and comparing to crawls from "Scooter".. Inktomi, Excite, and Google haven't been an issue, and FAST is probably one of the most spammed SE's out there, so no worries there.
Of course, everything changes, and part of this game is to stay on top of your logs. That's another feature to really focus on with whatever software platform you use for IP/UA delivery. It's nice to have "norefers" write to a seperate file from normal click throughs, and also spider crawls write to their own file. I would advise against using any software that doesn't have these basic capabilities.
Using IP/UA delivery requires much more monitoring, for new UA's and IP's.. Catching a new one through forums like these is great, but sometimes costly, and much too late in the game. As a SE like Google could come in w/ a new UA/IP and crawl through hundreds of URL's. So be prepared to dedicate some extra time to monitoring your norefers. More sophisticated software has alerts built in, that tell you when a variation of a known SE user agent is crawling your URL's, or a known SE IP is crawling your URL's with a new user agent that this IP has not shown before.
Some use the "host" environmental variable at the server level, but we've always used either IP/UA, or IP.
In summation, there are a lot of PERL scripts out there, that do a basic job, though I wouldn't use/recommend the majority of them. IP/UA delivery isn't a instant success solution. We've been doing it since before it got the nasty reference of "cloaking". It all comes down to how many keywords you are optimizing. The larger the number, the more IP/UA delivery may be a possible solution.
That said, here are my two cents:
I have had sites banned - but they were pushing the edge. Some here would even call them spam. I don't believe there is much risk for the average site, but there is some. I keep sending people to
Air's post [webmasterworld.com]--it is wise advice. If you are going to push the envelop, I would do what Oil said and buy a couple of disposible domains to play with.
You really do need to stay on top of your logs and as Redzone said, size does help. It is much easier to look for patterns when you have 200 domains than when you have two. But two domains will give you a lot more info than just having one - and I really don't think 200 domains are all that much more useful for watching patterns than 20.
As far as programming languages go, I don't think Redzone and I will agree. I like scripting languages. They give you the ability to go in and make changes quickly and launch them within a couple of hours. It has helped me get a leg up on the competition when they were running programs through a compiler. Poeple will tell you they are too slow, but this isn't nessesarily the case. I've had 230k+ hits a day to one server using a relatively simple 31 kilobyte cgi script without any problem or perceived delay. And if you take advantage of mod_perl [perl.apache.org], speed will be your friend.
=Do not cloak many pages to one visitor page,
keep the relationship one to one as much as possible. The only exception is where you can have a few pages that remain highly relevant to the target page.
+ So for every cloaked page, there should be a corresponding "real" page that matches it? What I have in my mind is that you use cloaking to optimize and serve one page for this SE and another page for the next, right? If you have an equal for every cloaked page, aren't you in danger of having "similar/duplicate" pages, which is what got my client banned from AV? Or do you have several cloaked pages for one KW, but each page is targeted to only one SE and all those pages for the one KW have one corresponding "real" page?
****************************
= Do match title and description of your cloaked and visitor page, including the first sentence or so when feeding engines that do not use meta description. This avoids the snoopers who like to climb the rankings by removing their competition.
+ So, make sure that the cloaked page has the same <TITLE> & <META NAME="description"> as it's matching "real" page?
****************************
= Do link your cloaked pages as if they were real pages (which they are), and take advantage of theming, link popularity, etc.
+ So, link the cloaked pages from your site map? If I have several cloaked pages for one KW, but they are each optimized for a different SE, wouldn't a spider see all those "similar" pages and ban me? Doesn't this defeat the purpose of cloaking? (besides the pupose of keeping you code hidden from thieves)
****************************
= Do not use links on your cloaked page(s) that do not exist on your human page(s). This will make your page show up for searches of that domain and they'll wonder why. Plus it's not nice.
+ So, if you have a link to another page from a cloaked page, have the same link to the corresponding "real" page?
****************************
= Always let you clients know the guidelines you will follow if cloaking is involved, be careful with any engine targetted pages they supply.
+ So do you tell them that you are "cloaking" or do you tell them that you are "serving up specific pages to specific SE" or what?
*****************************
+ Also, I'm beginning to form a piture of what I would need from cloaking software. I have a guy in-house who's planning on helping me write PERL scripts. I get the feeling, however that it would be stupid to try this, and we should just go get a canned software to do it for us. We have Webtrends. Can it tell me the exact page a spider has hit? or should the cloaking software do that for me? And, I know that this is a delicate question, since you guys don't like to give recommendations as to which SW to go out and buy, but where does a guy get a recommendation, when he's got such a vague picture still of what it all should do?
Let's see if I can confuse things further :)
>Or do you have several cloaked pages for one KW, but each page is targeted to only one SE and all those pages for the one KW have one corresponding "real" page?
Bingo. Yes that is what I meant.
>So, make sure that the cloaked page has the same <TITLE> & <META NAME="description"> as it's matching "real" page?
Yes, but don't sacrifice optimization over it, this is primarily intended to keep snoopers at bay.
> So, link the cloaked pages from your site map? If I have several cloaked pages for one KW, but they are each optimized for a different SE, wouldn't a spider see all those "similar" pages and ban me? Doesn't this defeat the purpose of cloaking? (besides the pupose of keeping you code hidden from thieves)
No, what I was attempting to explain (and poorly I might add) is that the optimized pages should not be doorway like pages, with click here type links. They should have the same links to other pages within the site as the real pages, so that the engines spidering them can move from cloaked page to cloaked page and they look like the real site to the SE. Remember that when a spider crawls from one page to another it will get another cloaked page if it exists, so it can crawl all of the cloaked pages, moving in and out of cloaked pages depending on which exist as cloaked pages and which exist as visitor pages only.
>So, if you have a link to another page from a cloaked page, have the same link to the corresponding "real" page?
I mean links to external sites, if you put a bunch of links on your cloaked pages to other sites to boost relevancy and they don't exist on your human pages then everytime a webmaster checks their links at a SE your page will come up as linking to them and they won't be able to find the link on your page. They'll get suspicious.
>So do you tell them that you are "cloaking" or do you tell them that you are "serving up specific pages to specific SE" or what?
This applies mostly if you are using the client's domain, then I would tell them you are custom serving pages. If you are on your own domain(s) then you don't need to tell them, it is just a caution.
You can do searches at the SE's for "cloaking" and "cloaking script" you should find both products and related infomational websites that can help with what you might need. I would say that the script should give you some method of determining what it served, you won't be able to tell from your server logs.
HTH
What do you mean, "...keep snoopers at bay."?
Here is a thread from not too long a go that talked about some of these snooping methods [webmasterworld.com...]
Thanks again Air. Very interesting stuff indead!
