A part of my URLs ("/_c/..." for catalogue pages) triggers ads for C and C++ programming-related stuff. Which doesn't fit the site at all, especially as it's implemented in Java! B^>

This only happens as far as I can tell where there would not be very good ads anyway (but in some cases I'd then prefer as PSA or a blank/redirection).

I have several times brought this up with various bits of G, most recently in response to an unsolicted mail from G in Ireland offering to help "optimise" my use of AdSense. The people in G variously respond with boilerplate or that this targetting miscue is my imagination or that it cannot be fixed, and my last email conversation stopped dead!

ASA: please get your CSR and "optimisation" and targetting people:

1) To get their story straight (and stop denying what is a very definite and comletely understandable mis-feature in an otherwise clever/good heuristic).

2) To allow occasional manual fixes (essentially a regex to zap URL components for some sites, and a small list of -ve keywords too) to overcome small wrinkles in the algo. Note that these should be fairly abuse-free as they will be BLOCKING ads rather than attracting undeserved ones...

Very little about G irritates me, but calling the customer a moron or stopping the conversation dead, is one of those things that does.

Rgds

Damon

OK, I can't stay mad at G for long: I've just had a nice email from them addressing some of my points.

Let the G leurve-fest resume. B^>

(Oh, sorry, maybe not on WW!)

Rgds

Damon

A part of my URLs ("/_c/..." for catalogue pages) triggers ads for C and C++

Obviously, Google is using the folder name as part of the "content" it analyzes for ad-placement purposes.

I should have mentioned that the ads for dog products mostly contained the word "cur", just like my file structure.

I am seriously considering changing the "cur" to a word that would help Google categorize my site.

I run a blog who's every page is mod-rewritten, for the purposes, among others, of making the urls more human freindly. As such, the public url can fit any format I'd like. In this case, I've made the urls like

/www.mydomain.com/my_niche_keyword/something.html

There's no purpose to the intermediary directory name other than sticking the keywords in there. In my experience it definitely helps targeting on new pages.

My observations, which may be completely wrong, show that directory names will target the first ads that appear when a page is brand new. How relevant they stay after the mediabot visits I'm not sure but my URL's with keywords in the folder names target very well.

By the way, in what language does cur = dog?

By the way, in what language does cur = dog?

English for one:

They . . . like to village curs,
Bark when their fellows do.
-- Shak.

G should definitely NOT remove this feature altogether, it is generally very useful.

I am only suggesting that it be possible to have a small number of overrides to ignore components of the URL and/or a "negative keywords list" per AS domain/site/subtree.

Rgds

Damon

ADS BY GOOGLE
my domain name
first folder name
second folder name
page.htm

none of which lead to anything but empty pages. I did have to send them a screen shot to convince them I wasn't imagining this.

Fortunately this usually goes away as soon as the bot gets there. But the bot doesn't always get there right away.

I am only suggesting that it be possible to have a small number of overrides to ignore components of the URL and/or a "negative keywords list" per AS domain/site/subtree.

I'd love to have that feature. There are too many times I've seen horrible targeting that was obviously based on some string in the URL that had nothing to do with the topic of the page.