Experiment: Will reducing the number of ADs improve earnings?

Hi

In some many threads I read the theory about how reducing the number of ADs on a page results in
increase of your earnings. In regard to this theory the less options for ads to click on you have the more
often the clicks occur on high paying ads.

Theory Example #1:
You display three Banners, the first gets served the highest paying ads and down it goes from there.
The more banners you display the more choice the user has to click on an AD which increases the
likelihood of clicks occurring on lower priced ads.

Theory Example #2:
If you reduce this number to only 1 AD the user only has one choice to click on a banner and therefor
(if he clicks) he clicks on the highest paying ADs served.

However I have never believed very much in this theory.

This theory presumes that if a user visits your site he clicks on an ad regardless of the number of choices. But is this really the case?

Let's take the above Example #2.
If a user visits my site and reads a thread and between two posts is a banner with 3 ads in it.
When the user reaches the AD he has a choice. Does one AD catches his interest and makes him click? If yes he clicks and you get money. If not then he reads further down the thread and when he found what he was looking for (or not) he is leaving the thread or even (worst case) your site.

Now with the above mentioned Example #1 let's assume he would make the same choice and don't click on one of the links in the first AD. He reads further down the thread and at the end of the thread there is another AD giving him three more links in this AD which have the chance to catch his interest. Now he again has a choice. If he clicks you get money. Yes it is less but non the less you get money from him where he otherwise would have left your site (in the worst case).

There is an article written about this theory that seems to confirm it but has one major design flaw.
The article is this: Double adsense earnings showing fewer ads

If this Example of a website would do as the article says it would earn more by reducing the number of ADs. However in this Example the best performing banner is the last one. All the Example shows is that the banner with the highest CTR should be loaded first to get the highest paying ADs.
Yes you can do that by removing any other banner on that page and you will increase earnings if that banner was not the first that was loaded in your html. However if you would simply switch loading order of your banners using CSS positioning you would not only get the increase in earnings by having the banner with the highest CTR filled with the highest paying ads you would also get the money from the 550 clicks on the other banners that you would not if you simply removed them.

This is exactly what I did on my forum I use for the experiment (see below).
I switched the order of the leaderboard below the navbar and the large rectangle placed in the content as second post of each thread because the rectangle had a much higher CTR then the leaderboard.

So in conclusion my theory is that displaying more ADs should increase the odds of users clicking on your ADs and increase your earnings not lower it as long as you see to it that the AD with the highest CTR is loaded first.

Also people debating that the CTR and eCPM has influence on your earnings and displaying multiple banners is lowering your overall CTR and therefor your earnings.

In regards to the CTR and PIs I am unsure as of now.
I never really saw a field test which demonstrated evidence to one or the other side.

This is why I run my own private test starting tomorrow for 10 full days.

The Experiment:

One of my forums earned me around 25USD (which is low enough to risk it for this test ) with
around 45k Banner impressions from 05.05. - 15.05.2010 (10 full days)

Banners displayed currently on this forum are:
- a Leaderboard below the forum navigation (all users) - 2nd highest CTR
(this Leaderboard is loaded AFTER the following two rectangles in the html and positioned via CSS.)
- a Large Rectangle as second post of each thread (guests only) - highest CTR
(Loaded first in HTML even though it is the second banner displayed for the user)
- a Large Rectangle as last post of each thread (guests only)
(Loaded second in HTML even though it is the third banner displayed for the user)
- a Large Rectangle after first blog post (all users)
- a Leaderboard below the thread list in FORUMDISPLAY (all users)
- a Leaderboard below the forums on index (all users)
- A Square Banner in the Blog side navigation (all users)
- a Link unit in the navbar (guests only)
- a Link unit below the thread list in FORUMDISPLAY (all users)
- a Link unit below the last post of each thread (all users)
- a Link Unit besides the Forum Rules (posting rules) below each thread

Banners displayed from now on for 10 full days counting from tomorrow are:
- a Leaderboard below the forum navigation (all users)
(this Leaderboard is loaded AFTER the following rectangle in the html and positioned via CSS)
- a Large Rectangle as second post of each thread (guests only)
(Loaded first in HTML even though it is the second banner displayed for the user)

I leave the following ADs running because the Impressions are lower then 50 and it's much more
trouble to remove them temporarily.
- A Square Banner in the Blog side navigation (all users)
- a Large Rectangle after first Blog post (all users)

If the CTR and Impression theory is correct I will earn significantly more then 25USD in the next 10 days.

After 10 Days I will post my results and conclusions here.

What is your point of view?

StarBuG

I ran this experiment years ago, and 2 ads has performed the best for me.
gives 8 options instead of 4 (normal adsense unit) and doen't show the 9-18 position units that pay only pennies

Thanks Starbug. Subscribed.

StarBuG, what did you learn from your experiment?!?

I am very interested to know as I am current doing similar testing.

Experiment results:

05. - 15.05. : 43.000 AD Impressions : 19,21 €
18. - 28.05. : 16.000 AD Impressions : 22,08 €

The CTR on each AD is exactly the same in both time periods.

Discussion:

To a degree reducing the AD Numbers increased my earnings.
20% less page impressions made 15% more money.

However this was a small scale experiment and its result is way
within the normal noise of adsense variability.

This showed that it is possible (to a degree) to remove
low performing ADs without losing money and thereby increase
the users experience on my forum.

After This experiment I added my 2 best performing link units
again but the rest of the ADs stay off for now.

After that I removed a lot of ADs on my high performance forum
and earnings dropped. I had to add them again one by one until
the optimal result was back.

This proofs that the small scale experiment is not big enough to
make any real conclusions.

My current philosophy is to use as few ADs as possible without reducing the possible earnings.
This comes down to typically 2 AD units and 1-2 link units per page.

Visit www.patientenfragen.net to see the number of ADs working best for me.

Feel free to discuss your own conclusion of the results.
I am interested to hear more opinions.

StarBuG

StarBuG,

I have three main ads per page: 728X90 below the Navbar, 728X90 at the bottom of the page and 300X250 (guests only).

The 728X90 below the Navbar brings in the most money but its CTR is usually 1/2 to 1/3 of the 300X250.

The 300X250 easily has the highest CTR, but because only about 1/2 of our visitors see it (visitors are split about 50/50 between members and non-members), it doesn't bring in as much money as the 728X90 below the Navbar.

The 728X90 at the bottom of the page is what I consider my "non-performer." It is generally getting a dismal CTR and is bringing in only about 1/10 to an appalling 1/100 (depending on the forum) of money as the the 728X90 below the navbar brings in. While it still brings in money, I don't feel like it brings in enough to justify keeping the ad in place.

Switch the loading order of the 728x90 and make it load after your rectangle.
You will increase your earnings.

If the rectangle is only seen by guests you only need to change the loading order for guests.

A perfectly centered Adsense banner below the navbar that is loaded AFTER other ADs - vBulletin.org Forum

I have a second rectangle as last post in threads for guests only.
Performs quite well for me.

Interesting discussion going on here. I am sure you guys are referring to adsense earning particularly. I have been experimenting with my blog and a mix of all ad type combination has worked for me the best. Those small link unit ads might be getting just pennies but they do get you something if you are using them at the end of the page and use is about to switch your site ;-)