Background for the question:
I have to admit I have not paid much attention to advertising on my
site prior to the last couple of months.
I'm showing ads to guests only to avoid clicking by members that think
they support me by clicking ads.
This is what I'm doing to test:
I have setup several channels to track performance of ads located in
different locations of a forum. With 75% of the traffic coming
directly to a post I have the highest CTR in that area so this is
where I'm concentrating my efforts. Setting up ads on a forum to cover
all entry areas and not breaking any AdSense TOS takes some effort
since ads can not be shown on login , sign up pages etc.
For the A/B testing I'm using the area between post 1 and 2 because it
is an area where guests reading a post will go by while reading.
Currently the Link Units on top prior to the post and the ad below the
post receive more clicks.
The A/B testing setup is rotating 728x90 Text Ads and Image Ads to see
if any of the two will perform.
Prior to the A/B testing I tried to see if the alignment of an ad
played any significant role when it came to CTR. Because of forum
layout any ads centered seem to perform better than left or right
alignment. Text ads (Link Units) with font color and background color
matching the forum is performing better than image ads
What are your AdSense Pro's opinion when it comes to the A/B testing?
I am not a pro, which is who's response was asked for regarding the
OP's question, but A/B testing is useful and practical. Set up one
channel of image only, one of text only, then use a server side script
to randomly display one or the other script to the browser. I've done
it many times and it does not break TOS and it works.
However if you randomize it client side, with JavaScript, you run the
risk of violating the TOS by altering the delivery/script.
> Channells are a way to know which of your website(s) or webpage(s) are
> performing more than the other. Suppose, you have a website in which
> you place articles about Dogs and Cats. Now while putting adsense ads
> in these webpages, how would you know which of your webpage, i.e,
> either about Cats or Dogs is fetching you more money than the other?
> Enter Channells. With Channell feature, you would create two Custom
> Channells for Cats and Dogs. While generating the Ad Code, there is an
> option to associate a chanell with that code. Thus you can associate
> different channells (Cats & Dogs in this case) to ad code and place
> this code in appropriate pages. This way you can see how much
> impressions/clicks each webpage(s) (Cats & Dogs in this case) are
> getting and try to optimize webpages specifically.
> > still don't understand how you set up channels, I am really new to
> > this
> > On Sep 7, 12:51 pm, NilsC wrote:
> > > Background for the question:
> > > I have to admit I have not paid much attention to advertising on my
> > > site prior to the last couple of months.
> > > I'm showing ads to guests only to avoid clicking by members that think
> > > they support me by clicking ads.
> > > This is what I'm doing to test:
> > > I have setup several channels to track performance of ads located in
> > > different locations of a forum. With 75% of the traffic coming
> > > directly to a post I have the highest CTR in that area so this is
> > > where I'm concentrating my efforts. Setting up ads on a forum to cover
> > > all entry areas and not breaking any AdSense TOS takes some effort
> > > since ads can not be shown on login , sign up pages etc.
> > > For the A/B testing I'm using the area between post 1 and 2 because it
> > > is an area where guests reading a post will go by while reading.
> > > Currently the Link Units on top prior to the post and the ad below the
> > > post receive more clicks.
> > > The A/B testing setup is rotating 728x90 Text Ads and Image Ads to see
> > > if any of the two will perform.
> > > Prior to the A/B testing I tried to see if the alignment of an ad
> > > played any significant role when it came to CTR. Because of forum
> > > layout any ads centered seem to perform better than left or right
> > > alignment. Text ads (Link Units) with font color and background color
> > > matching the forum is performing better than image ads
> > > What are your AdSense Pro's opinion when it comes to the A/B testing?
based on your previous answers in other treads you are closer to a pro
than me. Maybe the "pro" part should have been replaced with
"experienced users"
I'm using the script supplied from the AsSense support files for the
testing and for some reason image ads are doing better than text ads
in that position. I'll give it another week to see if trend continues.
> I am not a pro, which is who's response was asked for regarding the
> OP's question, but A/B testing is useful and practical. Set up one
> channel of image only, one of text only, then use a server side script
> to randomly display one or the other script to the browser. I've done
> it many times and it does not break TOS and it works.
> However if you randomize it client side, with JavaScript, you run the
> risk of violating the TOS by altering the delivery/script.
> On Sep 8, 5:06 am, Maria Janet wrote:
> > Channells are a way to know which of your website(s) or webpage(s) are
> > performing more than the other. Suppose, you have a website in which
> > you place articles about Dogs and Cats. Now while putting adsense ads
> > in these webpages, how would you know which of your webpage, i.e,
> > either about Cats or Dogs is fetching you more money than the other?
> > Enter Channells. With Channell feature, you would create two Custom
> > Channells for Cats and Dogs. While generating the Ad Code, there is an
> > option to associate a chanell with that code. Thus you can associate
> > different channells (Cats & Dogs in this case) to ad code and place
> > this code in appropriate pages. This way you can see how much
> > impressions/clicks each webpage(s) (Cats & Dogs in this case) are
> > getting and try to optimize webpages specifically.
> > > still don't understand how you set up channels, I am really new to
> > > this
> > > On Sep 7, 12:51 pm, NilsC wrote:
> > > > Background for the question:
> > > > I have to admit I have not paid much attention to advertising on my
> > > > site prior to the last couple of months.
> > > > I'm showing ads to guests only to avoid clicking by members that think
> > > > they support me by clicking ads.
> > > > This is what I'm doing to test:
> > > > I have setup several channels to track performance of ads located in
> > > > different locations of a forum. With 75% of the traffic coming
> > > > directly to a post I have the highest CTR in that area so this is
> > > > where I'm concentrating my efforts. Setting up ads on a forum to cover
> > > > all entry areas and not breaking any AdSense TOS takes some effort
> > > > since ads can not be shown on login , sign up pages etc.
> > > > For the A/B testing I'm using the area between post 1 and 2 because it
> > > > is an area where guests reading a post will go by while reading.
> > > > Currently the Link Units on top prior to the post and the ad below the
> > > > post receive more clicks.
> > > > The A/B testing setup is rotating 728x90 Text Ads and Image Ads to see
> > > > if any of the two will perform.
> > > > Prior to the A/B testing I tried to see if the alignment of an ad
> > > > played any significant role when it came to CTR. Because of forum
> > > > layout any ads centered seem to perform better than left or right
> > > > alignment. Text ads (Link Units) with font color and background color
> > > > matching the forum is performing better than image ads
> > > > What are your AdSense Pro's opinion when it comes to the A/B testing?
Whoa, they have a script back there for testing? I didn't know that.
That said, are you just comapring revenue, or, are you comparing the
other factors as well, maybe images are getting more impressions? Just
remember to look at the full picture...not just revenue
> based on your previous answers in other treads you are closer to a pro
> than me. Maybe the "pro" part should have been replaced with
> "experienced users"
> I'm using the script supplied from the AsSense support files for the
> testing and for some reason image ads are doing better than text ads
> in that position. I'll give it another week to see if trend continues.
> Thanks for the answer,
> Nils
> On Sep 8, 1:45 pm, Dan B. wrote:
> > I am not a pro, which is who's response was asked for regarding the
> > OP's question, but A/B testing is useful and practical. Set up one
> > channel of image only, one of text only, then use a server side script
> > to randomly display one or the other script to the browser. I've done
> > it many times and it does not break TOS and it works.
> > However if you randomize it client side, with JavaScript, you run the
> > risk of violating the TOS by altering the delivery/script.
> > On Sep 8, 5:06 am, Maria Janet wrote:
> > > Channells are a way to know which of your website(s) or webpage(s) are
> > > performing more than the other. Suppose, you have a website in which
> > > you place articles about Dogs and Cats. Now while putting adsense ads
> > > in these webpages, how would you know which of your webpage, i.e,
> > > either about Cats or Dogs is fetching you more money than the other?
> > > Enter Channells. With Channell feature, you would create two Custom
> > > Channells for Cats and Dogs. While generating the Ad Code, there is an
> > > option to associate a chanell with that code. Thus you can associate
> > > different channells (Cats & Dogs in this case) to ad code and place
> > > this code in appropriate pages. This way you can see how much
> > > impressions/clicks each webpage(s) (Cats & Dogs in this case) are
> > > getting and try to optimize webpages specifically.
> > > > still don't understand how you set up channels, I am really new to
> > > > this
> > > > On Sep 7, 12:51 pm, NilsC wrote:
> > > > > Background for the question:
> > > > > I have to admit I have not paid much attention to advertising on my
> > > > > site prior to the last couple of months.
> > > > > I'm showing ads to guests only to avoid clicking by members that think
> > > > > they support me by clicking ads.
> > > > > This is what I'm doing to test:
> > > > > I have setup several channels to track performance of ads located in
> > > > > different locations of a forum. With 75% of the traffic coming
> > > > > directly to a post I have the highest CTR in that area so this is
> > > > > where I'm concentrating my efforts. Setting up ads on a forum to cover
> > > > > all entry areas and not breaking any AdSense TOS takes some effort
> > > > > since ads can not be shown on login , sign up pages etc.
> > > > > For the A/B testing I'm using the area between post 1 and 2 because it
> > > > > is an area where guests reading a post will go by while reading.
> > > > > Currently the Link Units on top prior to the post and the ad below the
> > > > > post receive more clicks.
> > > > > The A/B testing setup is rotating 728x90 Text Ads and Image Ads to see
> > > > > if any of the two will perform.
> > > > > Prior to the A/B testing I tried to see if the alignment of an ad
> > > > > played any significant role when it came to CTR. Because of forum
> > > > > layout any ads centered seem to perform better than left or right
> > > > > alignment. Text ads (Link Units) with font color and background color
> > > > > matching the forum is performing better than image ads
> > > > > What are your AdSense Pro's opinion when it comes to the A/B testing?
> Whoa, they have a script back there for testing? I didn't know that.
> That said, are you just comapring revenue, or, are you comparing the
> other factors as well, maybe images are getting more impressions? Just
> remember to look at the full picture...not just revenue
> On Sep 8, 1:04 pm, NilsC wrote:
> > Dan,
> > based on your previous answers in other treads you are closer to a pro
> > than me. Maybe the "pro" part should have been replaced with
> > "experienced users"
> > I'm using the script supplied from the AsSense support files for the
> > testing and for some reason image ads are doing better than text ads
> > in that position. I'll give it another week to see if trend continues.
> > Thanks for the answer,
> > Nils
> > On Sep 8, 1:45 pm, Dan B. wrote:
> > > I am not a pro, which is who's response was asked for regarding the
> > > OP's question, but A/B testing is useful and practical. Set up one
> > > channel of image only, one of text only, then use a server side script
> > > to randomly display one or the other script to the browser. I've done
> > > it many times and it does not break TOS and it works.
> > > However if you randomize it client side, with JavaScript, you run the
> > > risk of violating the TOS by altering the delivery/script.
> > > On Sep 8, 5:06 am, Maria Janet wrote:
> > > > Channells are a way to know which of your website(s) or webpage(s) are
> > > > performing more than the other. Suppose, you have a website in which
> > > > you place articles about Dogs and Cats. Now while putting adsense ads
> > > > in these webpages, how would you know which of your webpage, i.e,
> > > > either about Cats or Dogs is fetching you more money than the other?
> > > > Enter Channells. With Channell feature, you would create two Custom
> > > > Channells for Cats and Dogs. While generating the Ad Code, there is an
> > > > option to associate a chanell with that code. Thus you can associate
> > > > different channells (Cats & Dogs in this case) to ad code and place
> > > > this code in appropriate pages. This way you can see how much
> > > > impressions/clicks each webpage(s) (Cats & Dogs in this case) are
> > > > getting and try to optimize webpages specifically.
> > > > > still don't understand how you set up channels, I am really new to
> > > > > this
> > > > > On Sep 7, 12:51 pm, NilsC wrote:
> > > > > > Background for the question:
> > > > > > I have to admit I have not paid much attention to advertising on my
> > > > > > site prior to the last couple of months.
> > > > > > I'm showing ads to guests only to avoid clicking by members that think
> > > > > > they support me by clicking ads.
> > > > > > This is what I'm doing to test:
> > > > > > I have setup several channels to track performance of ads located in
> > > > > > different locations of a forum. With 75% of the traffic coming
> > > > > > directly to a post I have the highest CTR in that area so this is
> > > > > > where I'm concentrating my efforts. Setting up ads on a forum to cover
> > > > > > all entry areas and not breaking any AdSense TOS takes some effort
> > > > > > since ads can not be shown on login , sign up pages etc.
> > > > > > For the A/B testing I'm using the area between post 1 and 2 because it
> > > > > > is an area where guests reading a post will go by while reading.
> > > > > > Currently the Link Units on top prior to the post and the ad below the
> > > > > > post receive more clicks.
> > > > > > The A/B testing setup is rotating 728x90 Text Ads and Image Ads to see
> > > > > > if any of the two will perform.
> > > > > > Prior to the A/B testing I tried to see if the alignment of an ad
> > > > > > played any significant role when it came to CTR. Because of forum
> > > > > > layout any ads centered seem to perform better than left or right
> > > > > > alignment. Text ads (Link Units) with font color and background color
> > > > > > matching the forum is performing better than image ads
> > > > > > What are your AdSense Pro's opinion when it comes to the A/B testing?
interesting, I've never seen it, I've never wanted to mess with their
JS for fear of violating some rogue TOS rule...
remember, programming languages are incapable of being "random", it's
seeded and the results are not back and forth perfectly, over the
course of hundreds it should even out, but it doesn't always, so watch
for impression counts being even and such too.
> > Whoa, they have a script back there for testing? I didn't know that.
> > That said, are you just comapring revenue, or, are you comparing the
> > other factors as well, maybe images are getting more impressions? Just
> > remember to look at the full picture...not just revenue
> > On Sep 8, 1:04 pm, NilsC wrote:
> > > Dan,
> > > based on your previous answers in other treads you are closer to a pro
> > > than me. Maybe the "pro" part should have been replaced with
> > > "experienced users"
> > > I'm using the script supplied from the AsSense support files for the
> > > testing and for some reason image ads are doing better than text ads
> > > in that position. I'll give it another week to see if trend continues.
> > > Thanks for the answer,
> > > Nils
> > > On Sep 8, 1:45 pm, Dan B. wrote:
> > > > I am not a pro, which is who's response was asked for regarding the
> > > > OP's question, but A/B testing is useful and practical. Set up one
> > > > channel of image only, one of text only, then use a server side script
> > > > to randomly display one or the other script to the browser. I've done
> > > > it many times and it does not break TOS and it works.
> > > > However if you randomize it client side, with JavaScript, you run the
> > > > risk of violating the TOS by altering the delivery/script.
> > > > On Sep 8, 5:06 am, Maria Janet wrote:
> > > > > Channells are a way to know which of your website(s) or webpage(s) are
> > > > > performing more than the other. Suppose, you have a website in which
> > > > > you place articles about Dogs and Cats. Now while putting adsense ads
> > > > > in these webpages, how would you know which of your webpage, i.e,
> > > > > either about Cats or Dogs is fetching you more money than the other?
> > > > > Enter Channells. With Channell feature, you would create two Custom
> > > > > Channells for Cats and Dogs. While generating the Ad Code, there is an
> > > > > option to associate a chanell with that code. Thus you can associate
> > > > > different channells (Cats & Dogs in this case) to ad code and place
> > > > > this code in appropriate pages. This way you can see how much
> > > > > impressions/clicks each webpage(s) (Cats & Dogs in this case) are
> > > > > getting and try to optimize webpages specifically.
> > > > > > still don't understand how you set up channels, I am really new to
> > > > > > this
> > > > > > On Sep 7, 12:51 pm, NilsC wrote:
> > > > > > > Background for the question:
> > > > > > > I have to admit I have not paid much attention to advertising on my
> > > > > > > site prior to the last couple of months.
> > > > > > > I'm showing ads to guests only to avoid clicking by members that think
> > > > > > > they support me by clicking ads.
> > > > > > > This is what I'm doing to test:
> > > > > > > I have setup several channels to track performance of ads located in
> > > > > > > different locations of a forum. With 75% of the traffic coming
> > > > > > > directly to a post I have the highest CTR in that area so this is
> > > > > > > where I'm concentrating my efforts. Setting up ads on a forum to cover
> > > > > > > all entry areas and not breaking any AdSense TOS takes some effort
> > > > > > > since ads can not be shown on login , sign up pages etc.
> > > > > > > For the A/B testing I'm using the area between post 1 and 2 because it
> > > > > > > is an area where guests reading a post will go by while reading.
> > > > > > > Currently the Link Units on top prior to the post and the ad below the
> > > > > > > post receive more clicks.
> > > > > > > The A/B testing setup is rotating 728x90 Text Ads and Image Ads to see
> > > > > > > if any of the two will perform.
> > > > > > > Prior to the A/B testing I tried to see if the alignment of an ad
> > > > > > > played any significant role when it came to CTR. Because of forum
> > > > > > > layout any ads centered seem to perform better than left or right
> > > > > > > alignment. Text ads (Link Units) with font color and background color
> > > > > > > matching the forum is performing better than image ads
> > > > > > > What are your AdSense Pro's opinion when it comes to the A/B testing?
If they give us the code then I guess it's not breaking the TOS.
The script setup should close 50/50, it's not random but but the image
ads is receiving .94% more pi's. I flipped the order of the ads in the
A/B script to see if that changes anything for the next couple of
days. Page reload may affect the accuracy
Another interesting aspect of this testing is that I see there are
different parts of the forum that receive a lot page impressions and
some of those locations never have a click. I have to use those areas
for something else.
> interesting, I've never seen it, I've never wanted to mess with their
> JS for fear of violating some rogue TOS rule...
> remember, programming languages are incapable of being "random", it's
> seeded and the results are not back and forth perfectly, over the
> course of hundreds it should even out, but it doesn't always, so watch
> for impression counts being even and such too.
> If they give us the code then I guess it's not breaking the TOS.
> The script setup should close 50/50, it's not random but but the image
> ads is receiving .94% more pi's. I flipped the order of the ads in the
> A/B script to see if that changes anything for the next couple of
> days. Page reload may affect the accuracy
> Another interesting aspect of this testing is that I see there are
> different parts of the forum that receive a lot page impressions and
> some of those locations never have a click. I have to use those areas
> for something else.
> Nils
> On Sep 8, 3:08 pm, Dan B. wrote:
> > interesting, I've never seen it, I've never wanted to mess with their
> > JS for fear of violating some rogue TOS rule...
> > remember, programming languages are incapable of being "random", it's
> > seeded and the results are not back and forth perfectly, over the
> > course of hundreds it should even out, but it doesn't always, so watch
> > for impression counts being even and such too.
Well, currently google has provided support for A/B testing.
Undoubtedly it is a nifty addition especially for those who use free
webhosts like blogger and so on.
For A/B testing, Google suggest the following format,
<script type="text/javascript">
var random_number = Math.random();
if (random_number < .5){
//your first ad unit code goes here
} else {
//your second ad unit code goes here
}
</script>
<script type="text/javascript" src="http://
pagead2.googlesyndication.com/pagead/show_ads.js"></script>
So, I wonder can we use A/B/C or more testing by altering the code
similar way. I mean sometimes, say if you wanna test which of the
three perform more, like say whether 1)728x90 Image or 2)728x90 text
or 3)468x90
For this I am suggesting this model, as follows:
<script type="text/javascript">
var random_number = Math.random();
if (random_number < .33){
//your first ad unit code goes here
} else if (random_number > .33 && random_number < .66 ){
//your second ad unit code goes here
}else{
//your third ad unit code goes here
}
</script>
<script type="text/javascript" src="http://
pagead2.googlesyndication.com/pagead/show_ads.js"></script>
I wonder whether we are allowed to use like above mentioned. Any
Adsense Pros hearing?
> Thanks for sharing your experiences with A/B testing with us, and for
> providing the links to the resources we provide on the subject.
> It is indeed a useful experiment to carry out, and we hope that you
> will share your conclusions with us.
> Ashley
> On Sep 8, 11:37 pm, NilsC wrote:
> > If they give us the code then I guess it's not breaking the TOS.
> > The script setup should close 50/50, it's not random but but the image
> > ads is receiving .94% more pi's. I flipped the order of the ads in the
> > A/B script to see if that changes anything for the next couple of
> > days. Page reload may affect the accuracy
> > Another interesting aspect of this testing is that I see there are
> > different parts of the forum that receive a lot page impressions and
> > some of those locations never have a click. I have to use those areas
> > for something else.
> > Nils
> > On Sep 8, 3:08 pm, Dan B. wrote:
> > > interesting, I've never seen it, I've never wanted to mess with their
> > > JS for fear of violating some rogue TOS rule...
> > > remember, programming languages are incapable of being "random", it's
> > > seeded and the results are not back and forth perfectly, over the
> > > course of hundreds it should even out, but it doesn't always, so watch
> > > for impression counts being even and such too.
