Forum Moderators: martinibuster

Message Too Old, No Replies

Smart pricing experiment

the effects of tesing scraper script

         

badtigger

9:35 pm on May 12, 2005 (gmt 0)

10+ Year Member



I have a site with lots of content, 400+ pages.
Recently I got the idea of an experiment with a php script to generate an Open Directory listings relevant to my topic, and threw adsense on it, to see what would happen...

I then had adsense on all of my 400+ pages (except for search/404s, etc.) and the dynamically generated ODP pages (which added another 350+ PP)

Guess what happended, almost immediately?

My eCPM nosedived from a very healthy figure (over 10.00) to an abismal couple of bucks, overnight.

I stuck with this for 6 weeks to see what would happen. Then, a week ago, I removed the adsense from the php-scraped Open Directory pages, and the eCPM went back up to levels which were previously normal, overnight.

I cannot help but conclude that "Smart Pricing" and page content are necessarily tied together. ( I know this is still causal reasoning, but...this seemed too coincidental)

spaceylacie

9:52 pm on May 12, 2005 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



I know it doesn't have anything to do with CTR, so it's gotta be the quality of the page. I've also experimented with the ads on generated pages and same thing... I think I only waited about a week to remove them.

RonS

10:52 pm on May 12, 2005 (gmt 0)

10+ Year Member



My eCPM nosedived from a very healthy figure (over 10.00) to an abismal couple of bucks, overnight. [...] I removed the adsense from the php-scraped Open Directory pages, and the eCPM went back up to levels which were previously normal, overnight.
What happened to total earnings (and CTR)? Did you get lots more page impressions and clicks making up for the lower eCPM?

Just wondering...

badtigger

11:07 pm on May 12, 2005 (gmt 0)

10+ Year Member



the really odd thing is that the page impressions were more or less the same and so was CTR.

Its just the total earnings were down, due to eCPM decline.

I doubt it was due to major changes in advertisers also, as most of the ads stayed the same, more or less all month.

creepychris

11:07 pm on May 12, 2005 (gmt 0)

10+ Year Member



I was wondering the same thing. Did the CPM of the old content also nose dive or was it just the average that nose dived (which could be expected)? This is a very good study if the CPM (tracked through channels) nose dived everywhere. It would mean that putting adsense on one part of the site affects the revenue from adsense on other parts, which seems like a silly result.

However from Google's point of view they may use global statistics to set global smart pricing parameters out of practical necessity: they don't get enough information about each page (which may only be displayed a few hundred times per month) to do smart pricing on a page by page basis. That could cause the results described by you.

europeforvisitors

11:32 pm on May 12, 2005 (gmt 0)



I cannot help but conclude that "Smart Pricing" and page content are necessarily tied together.

When smart pricing was introduced more than a year ago, Google explained that pricing would be based on "expected value to the advertiser." We were told that, for example, a click on an ad for digital cameras on a page of photo tips might be worth less than a click on the same ad next to a review of digital cameras. So if the results of your experiment are accurate, they confirm with what the "AdSense Update" e-mail said in April of 2004--at least to some degree. (The difference is that the e-mail gave the impression that smart pricing was determined by individual page content, while your experiment suggests that smart-pricing discounts are based on an average of all pages on the site.)

badtigger

12:30 am on May 13, 2005 (gmt 0)

10+ Year Member



I agree with both europeforvisitors and creepychris; it is impossible to tell from such a limited experiment how the SP is triggered.

What I do know is that the CPM dropped across all channels, so maybe it is safe to infer that part of a site can affect an average across all pages somehow.

Nitrous

12:40 am on May 13, 2005 (gmt 0)



Whats the definition of site?

Sometimes my sites grow to the point where a third of them live on my own computer (cable/iis) and may be spread around 4 domains, or more!

And some due to bandwidth issues have the site in one place and the images in another!

It all works but creates unfathomable (to me!) structures eventually...

Problem here is that most of my sites are spread around multiple domains and subdirectories...?

How does this work with smart pricing then?