| 4:03 pm on Feb 2, 2005 (gmt 0)|
Very interesting. I also rotate the shape/placing of ads to try to reduce the blindness problem, but your scheme sounds good and is much bolder than mine!
Thanks for the tip!
| 4:24 pm on Feb 2, 2005 (gmt 0)|
Out of interest do you start your counter script at 10 or 0? i.e. When you get a SE referral, does the first page have adsense on it, or do they only see it later?
| 4:38 pm on Feb 2, 2005 (gmt 0)|
Nothing until 10 page views. The first time someone finds us, if they're not sticking around, I don't want them to click either. Why? They probably won't convert. Sure, I've considered the "lets get the exit clicks" principle also, but I'm experimenting at the moment and want to try different takes.
The overal philosophy was based on not only increasing CTR but also finding the people that want to buy (the ones that are really digging around the site).
I figured, considering that "smart pricing" might actually be pretty smart, that if I provide a filtering mechanism and only supply the AdWords advertisers with good quality traffic (i.e. probably want to buy something), we'll get rewarded for it. I hoped to get around "Ad blindness" at the same time, as our regulars might well convert if they spotted a decent ad.
My figures are specific to my site and purpose - the overall idea of this post was to share the basic principle of rotating Ads with something completely non-ad related, and not displaying ads to the people that are not staying.
I will be reducing the 10 to 5 this month and I'll let you know what happens. But if you're looking to experiement you'll want to consider your own visitor stats and arrive at your own conclusions as to what to display and when.
| 6:57 pm on Feb 2, 2005 (gmt 0)|
How do you count the page views from the same person? Do you track IP address? Could you sticky me the script if you don't mind? I removed the adsense (in fact all the ads) from one of my favoriate community sites, since CTR was too low. I decided to make it non-profit because I wasn't getting significant money anyway. If the counting scheme works, then I may want to try it again. thanks,
| 7:22 pm on Feb 2, 2005 (gmt 0)|
No, we set a cookie.
|Could you sticky me the script if you don't mind? |
I don't mind at all. I didn't write it actually, my partner in crime did, but it's pretty straightforward so I'm sure he won't mind if I simply post it up here.
It's basically this (was 1 in 10, currently switched down to 1 in 6 (0..4 is 5 incrementations of the $showbanner variable and > 4 is 5):-
$showbanner = 0;
$showevery = 4;
$adsense_banner = 'our adsense code goes here';
if ($_COOKIE['adsense'] == '') setcookie('adsense', 0);
if ($_COOKIE['adsense'] > $showevery)
$showbanner = 1;
$cookie = $_COOKIE['adsense']+1;
if ($showbanner == 1)
CallOurBannerFunction is the banner engine that's set to rotate the images in the usual AdSense place. We didn't write this - it's part of our CMS, so I can't help you with that part, but it should be quite easy if you write it from scratch. It basically echo's back the HTML for the graphic image to be displayed.
I'd like to try to make it so that if the user has security set high, and we can't set a cookie, they get the Ads.
Can anyone help me with that?
|I removed the adsense (in fact all the ads) from one of my favoriate community sites, since CTR was too low. |
I can't promise you results as I'm sure this will vary from site to site, but I can definitely promise you this is worth a go. And I'd be interested to hear your results and what you plumb for.
| 8:08 pm on Feb 2, 2005 (gmt 0)|
I haven't played with cookies before. So it may be challenging. For the MCS part, I can use the banner rotation code that I used for my banners on other sites. I can't promise I will do this very soon. I am swamped with other important projects. But I will definitely try this later.
Thanks for the code! Flag up.
| 8:33 am on Feb 3, 2005 (gmt 0)|
trillianjedi- thanks for sharing your test with the rest of the forum! What an interesting way to prevent 'banner blindness' with your users. Hopefully this will encourage other pubs to work on optimizing their AdSense sites. There's always potential to improve and a lot of creative ways to go about it.
Also, you're always welcome to shoot an email to adsense-support if you have any questions about how to optimize your site or if you'd like someone from the support team to take a look at your site to suggest any tips.
| 8:39 am on Feb 3, 2005 (gmt 0)|
| 3:53 am on Feb 4, 2005 (gmt 0)|
TJ, thanks for posting that, very interesting.
Do you remember how long it took before you noticed the jump in revenue? If the good results are because of "smart pricing" changes I'm curious how it takes for that to update. thanks
| 11:24 am on Feb 4, 2005 (gmt 0)|
|Do you remember how long it took before you noticed the jump in revenue? |
Within 24 hours.
| 11:51 am on Feb 4, 2005 (gmt 0)|
Very interesting. I'll give it a try.
btw: Why not use a random algorithm where the ad will show in one out of ten. This will save all the cookie-coding effort.
| 11:57 am on Feb 4, 2005 (gmt 0)|
|Why not use a random algorithm where the ad will show in one out of ten. |
We considered that - I don't actually remember the reason we went for a rotation by page views in the end.
Might well try out a random interval on another site.
I have to say, so far, 1 in 6 is working out well though.
| 1:13 pm on Feb 4, 2005 (gmt 0)|
|if I provide a filtering mechanism and only supply the AdWords advertisers with good quality traffic (i.e. probably want to buy something), we'll get rewarded for it. |
I congratulate you on thinking for the advertisers first. This is great thinking. One can always take the easy way of just focussing on getting the maximum clicks but if everyone can think like you and focus on conversions for the advertisers, it is a win-win situation for everyone.
thanks tj for sharing it.
| 2:30 pm on Feb 4, 2005 (gmt 0)|
tj: Maybe because your rotation method is more reliable. I've just implemented your idea with randomly showing ads and went down to 1 of 4.
However, sometimes 5 ads appear one after another...
| 2:42 pm on Feb 4, 2005 (gmt 0)|
Thanks for the idea trillianjedi.
This has just been put on my "Must Try" list!
| 3:35 am on Feb 5, 2005 (gmt 0)|
> I suspect you have to be a little creative with what you display in the AdSense "space" - that is what I think is reducing the effect of Ad "blindness" - people, including the regulars, are now actively looking to see what crops up there next.
| 7:24 pm on Feb 6, 2005 (gmt 0)|
What forum system are you using? Having trouble adopting this for phpbb.
| 9:43 pm on Feb 6, 2005 (gmt 0)|
This is fairly easy to do with phpBB if you use the eXtreme Styles mod (Google it), which allows you to use PHP code in your .tpl template files.
| 2:26 am on Feb 7, 2005 (gmt 0)|
I'm thinking of rotating ad colours. Perhaps have a colour that complements my site's colour scheme and then every 10 pageviews display ads that contrast with the site's colours.
I don't get consistent CTRs so I can't really test it. Would someone else like to give it a try?
| 7:50 am on Feb 7, 2005 (gmt 0)|
we are trying out the rotating colors since some time, it works. banner blindness does get reduced and clicks increase. conversions? dunno but i think so, since EPC hasnt dropped but has remained as before.
| 10:52 pm on Feb 25, 2005 (gmt 0)|
What does mediabot think of this? Is it still getting relivent ads throughout your site? Or maybe there is a better way by adding another if statement onto it to allow it to always show ads to mediabot/google?
| 11:05 am on Oct 1, 2005 (gmt 0)|
I saw this one got ressurected in another thread so perhaps time for an update 6 months or so on.
I've tried pretty much everything from 1 in 10 to 1 in 2 and find that 1 in 6 works great for me.
Pictures work and silly things work. Some of the best results I've had are have been from silly comments : "This space intentionally left blank" to "Shine your widget up or the cat get's it" (in cut out newspaper typeface).
If you look at the AdSense "heat-map":-
The top right hand corner of the page, traditionally the place where 468x60 banners go, is stone cold. This is where I'm rotating my images with Ads and I've held a 1.8% CTR average in this space since March (that's very high for banners on a now 10k page views/day site mostly comprised of regulars).
You can achieve better CTR's in other page locations, but for me, in terms of page design, that top right hand corner just works really well as a space for half-size banners.
Order of display
I've found this to be the most interesting insight in to how smart pricing might be working, and into maintaining a high conversion rate for advertisers. There's no method or science to this, so I accept I'm making some big assumptions, but as far as I can tell from the bottom line, knowing that ad prices have remained largely static, these ads must be converting reasonably well.
The key is to show the real ad on page view #6 and not page view #1. This is what I've found from experimentation. What I think is happening is this; if I have a visitor that finds this site by mistake, or realises that he isn't going to find what he wants there and heads for his back button, I don't want to display him an ad - an exit-click strategy does not work for this site.
Displaying the ad at page view #6 tells me that this guy is digging around and he's looking for something. These are the guys that are making me money and keeping the advertisers happy.
The interleaved images are a means of keeping that guy interested in that top right corner space. For the regulars it's probably a case of "What stupid thing is TJ going to come up with next" curiousity, for the new arrival, it's probably more of a "What the ****?!", but it works.
| 11:26 am on Oct 1, 2005 (gmt 0)|
Yep, very interesting.
One thing that I've been tinkering with and thinking about is showing fewer ads on "first-hit" pages and low-paying pages to make the user experience better by reduing page-load time.
That may also be a small factor in your success, with 1:6 being your sweetspot.
| 12:06 pm on Oct 1, 2005 (gmt 0)|
| 12:27 pm on Oct 1, 2005 (gmt 0)|
Thanks TJ for this valuable information.
If ad prices have been stable, though, couldn't this suggest that conversion rates for SE visitors are about the same as your repeat visitors? The reason might be that if they have gone to the trouble of finding and clicking on an ad out of the hot zone, then they are seriously looking for something.
On that basis, why not try always displaying the ad on the first visit, and thereafter every 1 in 6?
| 12:31 pm on Oct 1, 2005 (gmt 0)|
|On that basis, why not try always displaying the ad on the first visit, and thereafter every 1 in 6? |
Tried that, wasn't as effective.
|...clicking on an ad out of the hot zone, then they are seriously looking for something. |
The principle of this exercise was to make that ad space a hot zone, rather than accepting the AdSense heatmap as the be all and end all.
It's about making rotational changes to influence the areas on the screen that users eyes fall upon - changing standard patterns.
| 1:38 pm on Oct 1, 2005 (gmt 0)|
|Nothing until 10 page views. The first time someone finds us, if they're not sticking around, I don't want them to click either. Why? They probably won't convert. Sure, I've considered the "lets get the exit clicks" principle also, but I'm experimenting at the moment and want to try different takes. |
I read about this 1-out-of-n banner idea a few months ago and brainstormed on it for a long time.
The points I'm considering are:
- which are the entry pages?
- which are the exit pages?
- which is the best way to avoid banner blindness?
- will a click an AdSense cause me to lose that visitor for ever?
The last question is quite interesting to me. Depending on that answer I would act one way or a completely different one instead. It would be nice to know whether people opened the ads in a different window or not, for example.
If you have a web site where people enter on a specific entry page, visit at most 2 pages, then exit, you can't, obviously, use the show-the-tenth-banner approach.
One idea I had was to create a page where there are two places where I can put banners, but I fill only one of them with the banner and the other one with "interesting spots". This would (should?, could?) reduce banner blindness while still avoiding to show less banners than page-views.
| 6:33 am on Oct 2, 2005 (gmt 0)|
|On that basis, why not try always displaying the ad on the first visit, and thereafter every 1 in 6? |
|Tried that, wasn't as effective. |
What I don't get is how you were able to determine that? I do understand that you were trying to influence smart pricing by not showing your visitor the ad right at the first chance. I can imagine being able to see which rotation causes higher CT but smart pricing? We have no idea how long till it takes effect, there is a lot of randomness around when someone will happen to buy, not to mention that not all adwordvertisers have that confirmation code on display.
| 6:46 am on Oct 2, 2005 (gmt 0)|
Since ASA has noticed this thread and the method is cleverly logical, how about incorporating the feature as an option in AdSense itself. Would save thousands of collective hours of programming and would be utilized by more publishers. Being optional would mean that no one would loose.
| This 34 message thread spans 2 pages: 34 (  2 ) > > |