More proof that popularity counts

More frequently accessed pages get better ads

[androidtech]

Due to a web mastering snafu, we ended up with two copies of an identical web page, one with an .html extension the other with a .php extension. There is no content added by php scripts in the .php page version. The extension was change when the whole site went to using .php extensions. The pages are indeed identical.

The .html extension page is still referenced through some off links throughout the site, but not via the main links like the .php page is.

The .php page gets approximately 1200 unique visitors a day.

The .html page gets approximately 170 unique visitors a day.

The .php page consistently gets four excellently targeted ads from AdSense.

The .html page gets 1 reasonably targeted ad per day and 3 weakly targeted ads. In addition, it frequently gets only 1 or 2 ads shown in a 4-ad skyscraper strip, even though the .php page is still getting 4 good ones.

Just some more grist for the data mill.

Thanks.

[Yidaki]

An important information that might relativate the numbers: how long has the html version been online and how often has it been visited by the mediapartners bot vs. the php version?

[creepychris]

But do they have the same linking structure (same pages linking to them)?

I doubt it. Linking structure also helps to determine what a page is about. The pages may be identica, but the pages linking in might be completely different. I know Mediabot follows links to determine relevancy. I also know my single most popular page gets the worst targeted ads.

[Glen_Murphy]

I think we're onto something you know..

I recently put together a DMOZ directory. Before launching to the outer world, I made two versions of the site.

1) with all .cgi pages
2) with all .html pages using mod rewrite.

The content of each version was identical, and hosted on a new domain name.

Within minutes, the .cgi version was showing revelant paid ads.

3 weeks later, the .html pages are still showing charity ads.

Bare in mind neither version has been launched to the outer world.

[Jenstar]

androidtech & Glen_Murphy, what were the mediabot requests like for the two different sets of pages?

[androidtech]

Jenstar,

maybe it's just because I'm real tired right now but what do you mean by "what were the mediabot requests like"?

[Jenstar]

How often did the Google mediabot request (visit) the .html pages versus the .php pages? Did a higher number of visits per page by the mediabot have any relation to the better ad targeting?

If the html page only had one mediabot visit, while the equivelant php page had 12 mediabot visits, it could also mean that the more times the mediabot visits a page, the better it is able to provide targeted ads.

Hope that made sense because I'm really tired too ;)

[4eyes]

On a similar theme to Jenstar.

The page with the most visitors would allow them to build a better profile for that page and get a clearer idea of which ads are providing the best CTR.

I have no idea if their system works this way, but it would make sense for the ads to be initially targeted based on the page content and then refined by the click throughs - the more click throughs the better the targeting gets.

As I said, pure guess work, but feasible?

[jomaxx]

The same thing occurred to me. That's the way the banner networks supposedly optimize their campaigns, so it's certainly feasible.

But doing it on a page-by-page basis would be daunting. I'm not sure Google would feel the need to include that factor in their matching algorithm anyway.