Welcome to WebmasterWorld Guest from 188.8.131.52
Those ads were below the fold so most people click on the ads if they havent found what they want on the page itself.
I am still running tests to see if significant separation from the first ad block to the second ad block removes the down-selling effect. My guess is that it will but the data's not there yet.
Best advice is to perform A/B testings and go with the results.
A recent discussion about having 2 or more adsense ads concluded overall that it was NOT FAVORABLE. It seems the majority of people had less CPC with TWO ads than just having ONE ad.
It is hard to form your own conclusions based on what other people post here in the forum. For one, the most vocal are almost always those who have suffered a decline. The posts are skewed towards those with unfavorable experiences.
Plus, what is CPC at the end of the day? You don't bring that to the bank. You look at the revenues. The revenues are what you cash at the bank. In our case, using multiple ads contributed to an increase in revenues.
The only thing to consider is to experiment multiple ads on your own site -- then base your conclusions on the metrics that you deem important. Not what other folks are saying.
But from this one-day test, it seems as if my increase in earnings (doubling my previous earnings!) is coming from the square ads, and not the inclusion of a second banner on the site. Hmm. I will keep you up to date after a week.
On the other hand, it may be that you really don't have any downselling going on and only pick up revenue. Without a scientific test, you are just talking out your arse and fooling yourself.
And, by adding ads to the page, you may be making your page too commercial, turning off users.
I do seem to be getting additional revenue after testing it over a few months. Love the money. Hate the actual ads.
As for the issue at hand, if you are performing one day tests, I do hope you are testing with at least 10,000 impressions. Anything less would be "uncivilized".
Remember, your data is only as good as where you harvested it from.
Generally speaking, your margin of error is 1 divided by the square root of n, where n is the number of events that you are monitoring. In this case, you are monitoring clicks, not impressions, so if, say, you get 100 clicks on 10,000 impressions, your margin of error is 10% (1 over the square root of 100, or .1). Therefore, if your test only results in a 10% difference, you really have no idea if it's statistically significant or just caused by the natural variation expected.
Only by continuing the test until you get a statistically significant sample can you determine for sure whether there is a real difference (and even then, it's not "for sure", just statistically likely, since after all, you aren't guaranteed to win the baseball pennant race in a 7-game series even if you have won the first three games).
Finally, I would add that doing the test one day at a time - one day one way and one day the other - is poor methodology. A better way is to completely randomize each ad showing, so that you have a 50% chance of running just one ad at any one point and a 50% chance of running multiple ads. Varying it by day adds an unnecessary complication, which is that there could be some effect based on day of the week (people might click more often on Monday when they get to work than on Tuesday, say).
I tried multiple ad units for a week and decided that they hurt revenue. Then two weeks ago I was updating a site that had 3 specific areas of each page (very long web pages), and I used three ad units, each one in a specific way that matched the area (two are single ads, one is a skyscraper).
After 2 weeks, AdSense revenue has more than tripled. I guess it hasn't been long enough to say that the revenue will remain this way, but for now, it's all good.
What I did is to set up a channel for each ad unit in the key page and after three days of data, I can tell whether it works or not.
In some pages they just don't work clearly. For example:
1. Ads show in all three channels (Top, middle and bottom for the long pages) since they have the same number of images.
2. EPC in the top channel =~6x to 10x bottom channel
3. The most CTR in top and the least in bottom
4. Overall not much change in earnings.
I removed the multi-ads since I don't want the cheap ads shown on my site.
My conclusion is that the multi-ad for a page with high or mid pay ads doesn't give any advantage over a single ad at the best location.
[edited by: FromRocky at 3:02 am (utc) on Oct. 22, 2004]