Not for CTR improvement, but for eCPM improvement you might want to try ArtistMike's method: rename your channels. I tried it on some of the channels and it felt like it helped to increase EPC somewhat. Didn't care to do AB testing on it, it's possible that the increase was optical only.

I have also NEVER blended my ads, no matter how many times Google tells me to do that, and my ads are always seen as ads by the people that click them. The viewer knows full well that he is clicking an ad, and thus is not fooled and does not expect anything but to be taken to the advertiser's site. This keeps my CPC up.

I also have the ads on pages where people are trying to buy something, they are in a "buying mood" when they click the ad. This also helps keep my CPC up.

And I also rename the channels (as Radix mentioned) at least every Sunday when the algorithm resets it self lower, so that on Monday I am not stuck with low CPC on ads.

The channel renaming sounds pretty strange, but I guess it's worth a shot. Other things I'm considering:

Making pages shorter (less content per page)
Making pages longer (more content per page)
Moving to a smaller ad in the same size space (eg. a 250x250 ad block (or smaller) displayed in a 328x280 space.
Showing different size ads in the same space on a rotation.

Showing adsense ads only to U.S. visitors (or some other combination).
Rotating adsense ads with house ads for our own products in the same space (ads are flash ads). We're already doing this.
Making pages less wide overall so as to have the sides where the 160's are more prominently in view, particular at higher res.

...others?

I've blended my ads to some degree. Links are consistent throughout my site whether they're ads or content links. Ads are always separate from content by drawing a box around them or some sort of line or series of horizontal rules.

However, I think my CTR has remained high (at least double the average I've heard mentioned) because at least one ad unit is above the fold. 728x90 and 336x280 ad units are never used above the fold. I bend over backwards to ensure my sites don't look overly commercial. I also do everything I can to avoid accidental clicks (borders, horizontal rules, lines, etc).

But does any of this really help your CTR? What can you do to ensure advertisers write good ads or that Google is displaying well-targeted ads? That may be where high CTR originates and ends.

I think that if some ads format was working in the past, while your traffic source hasn't changed radically, the cause is mostly the lack of good advertisers and enticing ads for that topic.

One of my site can have the same traffic in one month and another, but the CTR will vary according to advertisers seasonal interests for that topic (ex: football and the superbowl).

Another site, with another topic, has exactly the same ad setup up and the CTR is triple, so I can't say "well, ads of the size A by B don't work anymore". I know for a fact that advertisers have a bigger interest in that particular topic.

Another site of mine with the same ads format, which did well for years, had its CTR completely crashed in september, a few weeks before the big "credit crunch" disaster. What changed with that site? Good advertisers were gone and replaced with low quality ones, maybe because of the economy, or Google's algorithm, who knows, but visitors weren't affected yet by the economy. I added a new bigger ad format and CTR improved, but my site look more spammy, to my disappointment.

PS: I stopped blocking advertisers in my competitive filters completely for a year now.

My penny:

- Rotate colors/ sizes. Works epsecially if you have a high percentage of returning visitors (forum, membership site etc.)

- Use your filter. Filter ads that are not interesting for your visitors.

- Use rectangles instead of horizontal units. The large rectangle rules; people develop ad blindness for horizontals.

- Avoid the 468x60. Ad blindness.

- Enable / disable image ads. See what ads turn up and if it works for you.

- Fewer ad blocks. Less spammy site = lower bounce rate = higher CTR.

- Use section targeting [google.com] for better targeted ads (exclude nav etc).

But does any of this really help your CTR? What can you do to ensure advertisers write good ads or that Google is displaying well-targeted ads? That may be where high CTR originates and ends.

Obviously we can't, but I do not believe, nor do I want to believe that we are helpless here. Besides, I can't wait until Google targets better, or does whatever to return ctr to former levels.

Hence, the question in the first place.

One observation that I've long wondered about. When I view my own sites (and I guess those of others), if I view, let's say ten pages, I'm pretty much going to see the same ads on each of those ten pages. In other words, even if I go from one page, to one completely on a different topic, I'll still see some ads regarding the first.

This is nothing new, though. I once thought that it was that ads were being cached, which would account for the same ads on each page.

It seems to me that even if there are 10 pages on a site on the same topic, that visitors should see a greater variety of ads than they seem to. After all, if you show someone the same ad ten times, do you think the llth will cause a click?

I dunno. Just an observation re: why people may be getting more ad blind.

PS. Another thing we are doing is making our sites more horizontal than vertical moving away from vertical ads, which historically haven't done that well for us. Thought a switch to squares and rectangles would help. We aren't seeing a clear trend.

In fact, NOTHING we've tried in the last weeks has caused an upturn in ctr.

Again, hence the question.

I use rotating colors for backgrounds and borders in most of my ad settings?

Even on a site with mainly new visitors I suspect that visitors quickly learn that the big blue/white/yellow/green/whatever-unchanging-color blob full of text is ads, or something else they are not interested in, so varying the colors may help a help a bit.

Avoid the 468x60. Ad blindness.

Not our findings. Just compiled the November stats, and found out (to my big surprise) that 468x60 has replaced 300x250 as the most performing ad block. They are not getting the majority of traffic though (these go to 728x90 and 160x600), but the key learning is: change the ad format to 'something else'.

I see two major reasons for the drop in CTR:

1) Economy. When people are scared by the future and struggle to pay their mortgage, they do not buy (in our case) luxury goods and services. Some still do, for sure, but those who in the past were considering to buy, don't.

2) Ad quality. On our sites, the targeting is worse than, say, a year ago. I often get 100% unrelated ads on pages that used to attract correctly targeted ads. Sometimes, the preview tool suggests for a page on blue widgets that just red widgets show. If this is indeed the case (I can't know), then it's no wonder that noone clicks. (And yep, the diet ads don't help.) Filtering helps to a certain extent, but we can only probe here and there to see what ads come up. We impossibly can check all pages for all geographic locations.

So, it's not a big surprise that the traditional ways of optimizing Adsense do not work for you. Sad, but true.

Try using at least 2 different background colors for ads. Say you have a 120/600 skyscraper and a 125/125 square. Make the 2 totally different from each other.

I blame the credit crunch for the drastic drop I saw over the past few months.

As for practical advice: I havn't done any AB Testing myself, but heard that it should lead to a significant improvement.

Oh, and another thing: Try to do some kind of heatmap analysis to see where your users click in your specific layout. There are free programs available for that or you invest a few dollars for a proper account. It's worth it. The one thing Google Analytics won't do for you.

CTR - I usually have at least five different ad background / color combinations in rotation during my prime season. For me the the LEAST blended ads (and in my view, the ugliest) seem to have the highest CTR. I dunno, maybe people are so revolted by them that they'll click on them to get them off the page.

In my experience, more content on a single page hurts CTR. YMMV.

My CTR did go up once I restricted it to one ad block and one ad link block per page, where previously I had two ad blocks. But it hasn't stayed up.

You're right about seeing the same ads from page to page - I see that too. I think it's part of the whole personalization thing.

I'm working on putting together a little network of friends and colleagues across the country who will take screen shots of the ads displaying for each, so I can get *some* idea of what other people see.

The only other thing I thought of to try (except it kind of makes me feel icky) is to randomly change up the layout of the pages. So that the ads aren't in the same place every time. But that means then the content might not be in the same place every time, and from a usability standpoint - well, like I said. Icky.

Note that I just ran some numbers, and my CTR is up about 10% so far in 2008 over the same time period last year. But it's been pretty stagnant for a while, and is starting to slip now. So yea, I'd like to come up with a way to bring it up.

Hmm, just occurred to me, this year I'm rotating some adsense blocks with affiliate ads. Need to look at how that might affecting things.

One thing I tried and didn't have any effects. Feeling pretty much insulted by the ridiculous EPC I started to get I removed AS (the single ad) from my main earner network of sites for 6 days early November. Putting them back did not help much, EPC values resumed to their lows.

I suspect that visitors quickly learn that the big blue/white/yellow/green/whatever-unchanging-color blob full of text is ads, or something else they are not interested in, so varying the colors may help a help a bit.

If the ads are something your visitors are not interested in, that's a problem within itself. I realize I'm stating the obvious, but if I review a page and find ads that wouldn't be interesting to my visitors, I replace AdSense on that page with something they will be interested in. I used to be afraid to do this because I "might" miss out on a high value click here and there - but not anymore.

Google claims the ads they display will be so well matched to your content your visitors will find them useful. Experience has shown me that that is often rodeo residual. We have to manage things ourselves.

-------------------------------------------------

After all, if you show someone the same ad ten times, do you think the llth will cause a click?

There is some merit to that, although I have always heard the number 6 or 7. But sometimes it takes someone seeing an ad multiple times before it clicks in their mind and they click the ad.

In direct marketing, if you send out a mailer to a list and get a 1% response, you can often send out the same mailer to the same list a few days later and get another 1% response. The second 1% are those people that didn't respond the first time they saw your mailer.

-------------------

Try to do some kind of heatmap analysis to see where your users click in your specific layout. There are free programs available for that or you invest a few dollars for a proper account. It's worth it. The one thing Google Analytics won't do for you.

You can accomplish that with channels, assuming you have a few available.

You can also track the effectiveness of different sizes of ad displays using channels. It's been a long time since we discussed assigning multiple channels to one display.

-----------------------------------------

Other suggestions:

On a regular basis, check for content thieves. We all help ourselves and each other every time a thief's AdSense account is closed.

Check your placement vs. content report to see if you might want to stop placement targeting.

Instead of a skyscraper, try two 120 x 240 displays stacked on top of each other with each having a different color scheme.

I've found that simple (and sometimes ugly) pages perform the best. The less there is on the page, the fewer distractions from your content and your ads.

By the way, coachm, great thread you've started here - this type thread has been missed.

FarmBoy

The only other thing I thought of to try (except it kind of makes me feel icky) is to randomly change up the layout of the pages. So that the ads aren't in the same place every time. But that means then the content might not be in the same place every time, and from a usability standpoint - well, like I said. Icky.

I'm not completely sure that it has to be...icky. css and collapsing divs, I think it would work aesthetically and I don't see anything wrong with expecting visitors to pay some attention, and wake them up by moving things around.

(I understand the common wisdom of a consistent layout on a website, but really, whosaysitgottabeso?

Anyway, I'm not going to do it, personally just because it's too speculative for the work involved.

edit: One thing we're doing is creating a 336x180 space and rotating ads of all sized that fit within it. Can't remember if I mentioned that.

Another thing I'm noticing (maybe) is that we used to see a lot more single or double ads in a 336 space, (you know those ugly big print things). To me, although they were ugly, if the ad was relevant it would stand out, whereas if there's one relevant ad within the block of four ads, it doesn't.

That's one reason we're shrinking down our ad spaces by running different sizes through the same space, since the smaller ones will only show a single ad (albeit in small print).

PS. Our eCPM is now down to about 25% of what it was at the high point, and dropping like a stone. I completely shut adense down on one of my sites that used to do well with it, after seeing that I had ZERO clicks by 2pm, and that stats seemed buggered anyway.

On some sites, I'm not even going to bother cutting back. I'm cutting them out. Sadly, on the same day that adsense crashed badly on us (a few days before Tgiving), EVERY other Internet source of revenue crashed pretty badly also, so in essence we are more stuck than I would have thought possible.

Sales down. October best month in our history. Nov. down 40% total rev. Dec. probably down another 50% over Nov.

ah well, I'm Canadian. Our government is falling too! No chicken little story, that.

Experimenting with colors is definitely something worth trying. It's difficult to predict exactly which color strategies will work best on your site, though we have some suggestions [google.com].

As some of you have mentioned, publishers sometimes find that rotating color schemes perform better for them than any one scheme performs individually. There's a pretty easy way to do it, as our Help Center explains:

Tip for testing color palettes: add variety and freshness to your ads by rotating between several color palettes. All you need to do is choose the Use multiple palettes option when generating your ad code during the Choose Ad Format and Colors step in the ad code setup, then hold down the Control or Command key and select up to four color palettes.

ASA

I'm going to be reporting back now and again, but one thing that I just remembered and that is that any changes in results such as ctr, cpc, etc are usually delayed somewhat, so evaluating tests need to be done on a longer time span than daily.

We tend to be on a weekly cycle where any significant ups or downs start on a Wednesday or Thursday, which I assume has something to do with smartpricing. May not be applicable to ctr, but certainly to other metrics.

I think the answer will depend a lot of the type of site and its audience. For example, rotating colors to fight "ad blindness" might be a good tactic on a community or news site (which is likely to have a regular audience), but it probably won't achieve much on a reference site that most users reach through search.

With AdSense, testing is easy, so why not simply test instead of relying on guesswork or advice that may not be relevant to your site and audience?