Dec 19, 2008

Collective Interest Test:: Reading the Stats

As I've been telling you, I'm testing campaigns for my website Collective Interest. It's a site on hybrid cars and solar energy. I had experienced some success once I got serious and started to concentrate on the AdWords and AdSense campaigns. I wanted to figure out whether a 49-state strategy was the best thing for me to do, or should I concentrate on areas of the country where hybrids were already hot.

My budget is meager, up to $12.00/day. I'm paying for this out of my own pocket.

After looking at the numbers, I decided to go with the segment called "Collective Interest Good States". This campaign had the 30 best states/metro areas for hybrid car sales.

After looking at the numbers, here's my decision structure:

"Good" represented 34% of the cost, gave me 34% of the clicks, 45% of the impressions, and an average ad position of 3.1.
"Control" represented 34% of the cost, gave me 35% of the clicks, 22% of the impressions, and an average ad position of 4.4.
"Underperforming" represented 32% of the cost, gave me 31% of the clicks, 32% of the impressions, and an average ad position of 3.0.

To recap, "Control" was a 49-state campaign, and "Underperforming" was the 22 worst states for hybrid car sales. All 3 campaigns had the same demographics, the same keywords, etc.

Since this is a content-only campaign, impressions were a little bit more important to me. My ad position was basically the same between "Good" and "Underperforming". "Underperforming" cost less, but really didn't perform with the clicks.

With that settled, the next test is search-based. Previous to the content test, I found that content was giving me clicks, while search was giving me impressions. For every 10 content clicks I received, I got .5 search clicks. Search left me with abysmal ad positions.

So I went to content only.

Now I'd like to look at how to rejuvenate search. I may not have the money to make it work properly, which wouldn't be bad; because it would give me an idea of the threshold I need to be at in order to make it work.

I'll let you know when that one starts up.

No comments: