Forum Moderators: martinibuster
One of our channels is our homepage where we show leaderboards. CTR for this page was x before the image ads begun showing on May 26. From May 26 up until yesterday, June 14, our CTR decreased by 23%. Since the channels feature was offered, the CTR on our homepage was fairly constant, moving within 1-5% range. This is the first time that the change was this significant.
One factor could be that there is only one advertiser using this that appears on our homepage, and so only one image leaderboard ad gets to be shown to our users. That must be some quick banner ad blindness :o(
Other channels showing image ads are showing extremely poor CTR rate relative to our other channels. One newly created channel (about 27 pages) that has been consistently showing image skyscraper ads have extremely minimal CTR. For a week now, we are barely getting clicks from these pages when traffic is comparable to other channels getting very good CTRs.
Good thing that we are increasing the number of our pages using purely text ad formats, as well as increasing traffic to make sure that our revenues remain on target. We may disable the image ads to see if the CTR goes back to its normal range. We will also change the skys to maybe leaderboard to see how it will affect CTR. EPC, though, is still doing very well.
You have only ONE ad instead of FOUR-FIVE for your users to choose from.
I think there are sectors where image-ads should perform better, i.e. when the advertiser wants to show a picture of the product.
1. Create 2 AdSense channels (A - text only, B - image/text).
2. Randomly select which channel and google_ad_type to include in the JavaScript code such that over the long run each channel receives about 50% of the impressions.
3. Compare behavior for each channel.
Without knowing what percentage of channel B's impressions displayed image ads (unless channel B *only* shows image ads and does so 100% of the time), you still won't know the absolute effect on CTR (or EPC), but you will be able to determine with statistical confidence whether image ads have a measurable effect on CTR since the randomness of the distribution of the impressions b/w channel A and channel B over a big enough number of impressions will balance out the effect of other factors while testing.
I've done this in PHP using the following logic, though it can be done in any language using numerous different methods.
$rand = rand(0, 100);
if ( $rand % 2 == 1 ) { include( 'as_code_1'; } else { include( 'as_code_2' ); }
But we'll continue the experiment to get a month's data. Then we'll make a decision whether to junk the image ads altogether.
We are expecting that within the next two months some serious big players will get involved and that should hopefully help earnings.
The only image ad I have seen is for some funny site, which was not as targetted as we would like.
If EPC or CTR is falling cancel the image ad idea for a while and see if either increases.
control for any other factors .... statistical confidence
Warms my heart to see a valid experimental design.
Remember it is still very early days for image ads on AdSense. We are expecting that within the next two months some serious big players will get involved and that should hopefully help earnings.
Hmmm....Care to reveal the source of your information? :-)
I agree that it's early days yet--and I think the target market for "image ads" consists of large advertisers and their ad agencies who aren't currently using CPC ads.
To existing CPC advertisers, image ads offer no obviously compelling benefit--especially since they display only on AdSense partner pages, not on Google's SERPs. For existing CPC advertisers, the safe thing to do is to stick with ads that have been proven to work: i.e., text ads, preferably on Google's own pages.
To large corporate advertisers and media buyers at advertising agencies, AdSense image ads do offer a compelling benefit: They offer a safe, comfortable way to test the CPC waters while getting better results from the display-ad formats that they're already using.
Side note: On my own travel-planning site, I can easily imagine some of existing corporate advertisers testing image ads to see how they work. For an airline, tour company, or cruise line that's accustomed to running display ads in newspapers and magazines plus text ads on AdSense sites, the idea of buying display ads through Google's contextual ad network shouldn't require a great leap of faith.
Warms my heart to see a valid experimental design.
What can I say - I'm an engineer by training. As you'd probably agree, there's plenty of opportunity to use experimental design (and statistics in general) to optimize sites.
Although CTR may be lower and viewers may be blind by these ads, i thought it looked quite sexy. Out of coincidence it matched my sites color theme perfectly, and still had the user defined border color.
They certainly may be garbage for CTR, but they look nice.
What I really like about image ads is they are not (yet at least) constant. Some pages will show image ads and others the regular text ads, this makes me look, and I can only presume others on the site will as well, this I hope after time will lead to a higher CTR rather than a drop in CTR.
True, and in addition to catching users' attention by their presence, they may also help the text ads to grab attention as users go from pages with image ads to pages with text ads. In other words, they help to solve the problem of "ad fatigue" by providing visual variety in the ad space.
Yesterday was my first full day with image ads, and daily earnings were higher than they'd been all month even though impressions were a bit lower than they usually are on Mondays. EPC was up, too, though not by a huge amount.
I may even be temtpted if I found all image ads to block some big players for a day or two to see if the reappearance of text and image ads has a positive effect.
