Where higher CTR can lose you money - beware. - Google AdSense - Display Ads forum at WebmasterWorld

Forum Moderators: martinibuster

Message Too Old, No Replies

Where higher CTR can lose you money - beware.

esllou

10:56 am on Aug 20, 2006 (gmt 0)

I've been doing a lot of A/B testing on my site and there has been a real ding-dong battle going on on my homepage between a 300 rectangle and the 336 rectangle. I've had about 10 full days of stats now and the results are as follows:

336 rectangle - 12.3%
300 rectangle - 11.5%

BUT, and here's the biggie, more money earnt off the 300 ad format. why? I imagine it's because very often, I only get 2 ads loading into it and they're the higher payers.

More choice in the 336 and therefore, more people end up clicking giving me a better CTR, but earnings should be your most important figure to keep an eye on and, for now, it looks like I'll be going with the 300 despite the lower CTR.

jaynl

11:42 am on Aug 20, 2006 (gmt 0)

It may also be that you're displaying image ads in those rectangles, and that advertisers in your market segment have designed more image ads for a certain rectangle format. With A/B testing, you can't measure this easily, I think.

esllou

12:18 pm on Aug 20, 2006 (gmt 0)

well, actually, I can seeing as I wouldn't allow an image ad within 20 miles of my site. So they're blocked.

OptiRex

1:08 pm on Aug 20, 2006 (gmt 0)

I imagine it's because very often, I only get 2 ads loading into it and they're the higher payers.

Quite possible, have you tried the preview tool (if it still works!) to see what ads are displayed elsewhere?

The only downside to your test is that I find the EPC can vary hugely from one day to the next.

Let it run for a month and see what happens, after all, if your thinking is correct you'll earn more:-)

esllou

1:14 pm on Aug 20, 2006 (gmt 0)

yes, I don't think 10 days is enough really, even if it is my homepage prime location and I have thousands and thousands of ad views to go by.

The earnings difference isn't exactly huge so I may just leave the A/B testing going anyway. A lot may depend on the presence of a couple of big paying advertisers so the pendulum may swing quickly to make the better CTR ad the bigger earner again.