Intuitive InSite

by Will Bontrager
Copyright 2006 Bontrager Connection, LLC

I’ve been testing something for 6 weeks and am now ready to make a report.

Some of our sites do pretty good with AdSense. A 7-10% click-through percentage is not uncommon for some of the pages. This is less than good numbers I’ve seen others report, but still highly acceptable.

However, some of our sites have an abysmal click-through rate. I mean, really bad.

At the beginning of February, I decided to do something about it, in spite of being busy with other projects.

The area I chose to pay attention to first is where over 250 Possibilities articles are archived, articles from July 1999 to July 2005.

Page views for the archives index and articles range from 800 to 1100 every day. During the month of January, the AdSense click-through rate was 0.8%.

Note that these click-through numbers are not coming from Google. It is my understanding that revealing their numbers would be against their TOS. Instead, we use ClickWatch™ for our link click reporting software. (Even if it wasn’t against Google TOS, I would still use ClickWatch for the instant reporting it provides — for all link clicks, not just AdSense.)

I had to do some testing. And the quicker I got results, the quicker our income would increase. Every 1% AdSense click-through increase would mean getting paid for an extra 300 clicks every month.

The quickest way I can think of to rotate some ads to see which are the best is with the free Master Merry-Go-Round.

I set up 4 ads. Two were square, one black on white and the other white and light blue on dark blue. And there were two towers, with the same color variations as the squares.

Those 4 ads were set up to alternate between the left side of the content and the right side of the content.

This made a total of 8 ads. Each was loaded into Master Merry-Go-Round, which dutifully rotates each, in turn, displaying the next ad every time a page loads.

Go to http://willmaster.com/archives and click on a link. Load the page repeatedly. You’ll see what I did. (The ads are different, now, as I’ll explain later, but you can see the idea of how the ads are being tested.)

The first two weeks of this doubled the click-throughs — double!

Actually, more than double. The average click-through rate of those 8 ads, for the 16-day period ending on February 19, was 1.8%. Each ad was shown approximately 1400 times.

The lowest performer was 1.2%, the highest 2.5%.

Now, that really isn’t a high enough number for conclusive results. But I thought I could see a trend, so I swapped out the lower-performing ads for some brand new ads, ones with white, green, and yellow text on a background of browns.

That flopped. I could see it right at the beginning. But I let it run until each ad was displayed twice as many times as the first test.

That took 27 days. Each ad was displayed nearly 3500 times. The total click-through rate was 1.4%, 0.4% less than the first, short-term test.

The black text on white background ads did the best of any color combinations during both tests. So that’s what is rotating now, all black on white ads. I decided to see what positions and sizes did the best, not mixing colors in with the test. It seems that the less variables are being tested, the sooner conclusive results might be obtained.

It’s too early to predict with certainty what the latest test results will be. It was started only yesterday. The first 735 page views have had 15 AdSense clicks for a click-through rate of 2.0%. That percentage will most likely change by the time the test has run its course.

Some Observations

These are observations, not necessarily conclusions. And the observations might not apply to your web sites. Each site is different.

Observation # 1:

The same ad that was doing 0.8% during all of January, did 1.4% during the first two weeks of February. The difference, other than the time of year, is that the ad was rotated with seven others.

Now, I don’t know about you, but I find it interesting that an ad’s click-through rate changed from 0.8% to 1.4% simply because it was rotated with other ads. That’s a 75% increase. It’s not that January just happened to be a low month, the previous October was low, too. (November and December were used to test different types of ads, different vendors, so aren’t comparable here.)

It appears that the mere fact of rotating ads can have an effect on click-through rates.

After thinking about it some, it started to make some sense.

On web sites where visitors rarely visit more than one page, rotation by itself might not make much of a difference. But on a site where visitors load several pages (willmaster.com sitewide averages 6+ page views per visitor), rotating ads could have one or both of these effects:

1. One ad shape, color, or position might be more noticeable to certain visitors than others are. Rotation provides a better chance of displaying the one that is noticed.
2. Rotating ads may not allow site visitors to subconsciously place ad appearance and location to then ignore. Or, if so placed, the next ad display, being different than expected, then gets noticed.

Whether or not the actual rotation of ads increases AdSense click-through rates for you, testing ads by rotation can give you a knowledge edge. Set up a different channel for each AdSense ad in the rotation. Google’s stats can tell you which are doing better than others.

I’ve read somewhere that an ad should be seen a minimum of 10,000 times before results are certain enough to act upon. Whether or not you do 10,000 impressions per test, replace the least effective with a different test ad.

Keep doing it. Retain the best and replace the least effective with a new test.

Gradually, your click-through rate should become better and better. Also, other factors, including displaying the same ad over an extended period of time, might cause click-through changes.

Observation # 2:

The blueish ads in the first test did as well as the whitish ads, in some positions even better. However, when the brownish ads were introduced in the rotation, the bluish ads’ click-through rates immediately dropped.

So that’s another thing to test, if you’ll be testing colors. Some color combinations can do better with each other than other combinations do.

Designers know color has an emotional effect on people.

Observation # 3:

Mari and I have had AdSense on many of our sites for several years.

It has been my observation that click-through rates of AdSense ads slowly decline if left alone. I don’t know why that is, unless repeat visitors no longer see the ads and, the longer the ad is in place, the larger percentage of visitors are repeats.

Sometimes a small change, like a text color, for example, can increase click-through rates enough for a spontaneous, “Wow!”

Will Bontrager
Programmer, Master Series CGI programs
(Article syndication powered by Master Syndication Gateway V2)