[xekul]

I don't know exactly how the content matching algorithm works, but I know that search traffic produces a higher conversion rate than content. A big part of that is because a content click doesn't have the same intent to buy that a good search keyword does. For my business, content clicks convert about 1/3 as well as search clicks, so I just cut my content bid to 1/3 of my search bid, and that way I'm paying the same per customer regardless of the source.

[baltcode]

Thanks! I guess my question was, how do you differentiate between context clicks? Say you are a health insurance company. You could bid 30 bucks for a search click on a search of "self employed health insurance plans", and say 1 dollar for a search click on a search of "medical bills". Fair enough.

Going with the 1/3rd analogy, it seems you should pay 10 bucks on a click with a blog post of someone comparing and sharing their experience of buying health insurance while freelancing. You should bid 33 cents per click along with news story about rising medical costs in the nation. But how do you differentiate between these two types of "specific" vs "general" contextual content while placing bids on contextual ads? Or is this functionality just missing in current bidding engines?

[xekul]

I'm aware of two ways to do this with the content network:

- Bid on specific placements. If there's a blog or blog post post I want my ad to appear on, I can tell Google that I want to raise my bid for that specific URL.

- Bid on specific keywords. For example, I might include the keyword phrase "freelancing health insurance personal experience" in my content campaign, and give it a higher bid price than the keyword "medical bills." In this regard, it's the same as with search, except that your ad is triggered based on Google's scan of the blog post, rather than what a searcher types into Google.

To get an idea of the keywords that Google might extract from a blog post, type its URL into the Google Keyword Tool:

https://adwords.google.com/o/Targeting/Explorer?__u=10000000...