I'm still trying to find my best combination. For years I ran a 468x60 just above the content and 300x250 in the sidebar; Google sent me an email predicting a revenue increase if I swapped out the banner for a 728x90 at the top, and darned if they weren't right - CTR and revenue improved immediately. On extra long pages, I have a 468x60 at the bottom too; I might swap that out for another 728x90 and try that.

Link units used to work really really well for me, but they crashed last year almost across the board. Not sure why, but I took em all off.

Right now I'm back testing borders again. I just hate how they look, but if they work, they work.

I probably should test plopping an ad in the middle of my content, but I just haven't been able to bring myself to do it yet.

What used to work amazingly well was a pair of large rectangles just below the site navigation, on the order of double digit CTR. Fortunately I took those down before all the Panda updates and my sites didn't catch any of those SEO cooties.

I'm curious about this as well. On a site I manage there is a 728x90 at the top of the page, a tiny 234x60 text ad in the middle of the page, and another 728x90 lower on the page.

The 234x60 has a terrible CTR and earns very little extra money, but a larger ad would not fit where I have it. My question is do you believe the extremely low CTR is hurting my earnings from the other ads on this page?

An AdSense rep explained to me that the top ad on pages typically are highest possible payout. The second ad should payout the 2nd best, and so on. I get a lot of clicks on this page from the 3rd ad on the page...but is the poor performance from the 2nd ad reducing the potential RPM of the 3rd ad?

I guess I should just pull the under-performing ad and find out.

It's essentially all about math and habits of visitors.

Let's do a simple example:
- You have a grand total of 2 pages
- page 1 links to the page 2 and nothing else
- page 1 is your landing page (all incoming links point here)
- page 2 has no outgoing links
- page 2 only has one incoming source fo traffic: page 1
- Each page has 2 ads on it. ad1 and ad2.
[I'm trying to make it less complex :) ]

Scenario:
Visitor lands on page 1.
Visitor has 4 options:
- go way (bounce)
- clink link to page 2
- click ad1
- click ad2
if they go to page 2
they again have 5 options:
- go way (bounce)
- click ad1
- click ad2
- click back

Now if you know your pages very well you could guess at the chance of each visitor taking each action (or run google analytics to get the data).
You could also learn the expected CPC for each ad.

You also get a good idea of how often a

From that you should be able to predict the expected value of a visitor.

Let's take the example:
Visitor lands on page 1.
Visitor has 4 options:
- go way : odds: 44% revenue: 0
- clink link to page 2: odds: 50%: revenue: TBC
- click ad1: odds: 5%; revenue: $0.15
- click ad2: odds: 1%; revenue: $0.01
if they go to page 2
they again have 5 options:
- go way (bounce): odds: 90%: revenue: 0
- click ad1: odds: 3%; revenue: $0.50
- click ad2: odds: 1%; revenue: $0.25
- click back: odds: 6%; revenue: TBC

You can then go on with the scenario, but let's assume we know we have negligible visitors that actually do anything alse but bounce after they click back (I want to keep it simple ...)

Adding it up:
page 1 - ad1: 1.00 * 0.05 * 0.15
page 1 - ad1: 1.00 * 0.01 * 0.01
page 2 - ad1: 0.50 * 0.03 * 0.50
page 2 - ad2: 0.50 * 0.01 * 0.25
Make a visitor worth $0.01635

Next, we redo our page 1:
we make link to page 2 stand out better, make it look more promising and get rid of ad2

This shifts our average visitor's behavior a bit, not just on page 1, but also on page 2 as our site is now more attractive and perceived differently.

Visitor lands on page 1.
Visitor has 4 options:
- go way : odds: 40% revenue: 0
- clink link to page 2: odds: 55%: revenue: TBC
- click ad1: odds: 5%; revenue: $0.15
if they go to page 2
they again have 5 options:
- go way (bounce): odds: 80%: revenue: 0
- click ad1: odds: 5%; revenue: $0.50
- click ad2: odds: 2%; revenue: $0.25
- click back: odds: 3%; revenue: TBC

Adding it up, our average visitor is now worth $0.024 - a 47% increase! But it's due to page 2 being visited more, not due to the revenue of page1 in itself (which lowered a tiny bit).

Now this scenario is obviously oversimplified and it's hard to understand the motivations of "the average visitor", yet having that insight is what enables one to cover the visitor not just on a page by page basis (what Google does already - but from their viewpoint), but also as a whole over your entire site and over more than one revenue stream (e.g. if you have affiliate sales on there).

What I've learned by trying to understand this is that less is more ... esp. if you do it in the beginning of their visit on your site - esp. if the site is large and does sell well on affiliates: it's not worth it to let a visitor escape from you on the first page they land on for cents: I'd rather try to sell them a book or two on the subject.
Yet at the same time the typical landing page is also where those less interested seem to find a way out in a prominent ad block - so you need to balance it all, but you need to do it for a site as a whole, not a page at a time as they are all linked ...

Amazon links work even better.

This has been my experience with Amazon.

swa66...I really like your example as it's kind of what I'm experimenting with right now...reducing number of options and trying to control visitor's flow more by funnelling or channelling them in the direction I want them to go.

However, unless I'm missing something I think the number of Options for page two is 4 not 5? And after the redo of page 1 the number of options is 3 not 4.

unless I'm missing something I think the number of Options for page two is 4 not 5? And after the redo of page 1 the number of options is 3 not 4.

Yeah, sorry - I simplified the example a bit more before posting.

why does second ad on page 2 pay 25x as much as second ad on page 1?

You may be able to get rough guesstimates for CTR on both ads and page links, bounce rates, etc, but you're just guessing CPC on ads and your whole formula is based on that.

It's not a formula, it's an example! And I said to start with that it was over-simplified.

The data was not constructed to proof any point - nor did it come out of thin air. The 2 pages was actually what I images a MFA site would do - not that I run any myself.

Feel free to use your own data to calculate it for your site (but you have way more than 2 pages I hope and way more paths that users use. Enjoy the calculation.
Changes you make are predictions in this kind of thing - unless you execute the changes and measure them in real life.

The core point I was trying to make was that you need to look at *all* your revenue (not just adsense), and *all* your pages - not just one at a time.
I was not trying to draw a conclusion as to less is more or not - only advocating to calculate it instead of guessing.