Google Trends For AdSense?

correlates with anything for you?

[ronburk]

One eye-grabbing feature of the new Google Trends (http://www.google.com/trends) is that it automatically graphs "news reference volume". One wonders if spikes in the latter correlate for some folks with phenomenon like EPC, CTR, etc.

I also note the ability to compare trends by separating the terms with commas. My risk-averse nature immediately leads me to start searching for terms within my niche that show any evidence of being hedge-worthy (i.e., when traffic for one goes down, the other tends to go up). If I can find such a thing, I will certainly invest effort in scoring well for the contrary search terms.

Realistically, it may be necessary to hedge different topics (e.g., separate websites) against each other to get the desired effect of stability. But, it's hard to prove that contrary terms don't exist within a niche -- maybe I just haven't been smart enough to locate them yet :-)

I also note the feature that labels spikes with the news information they correlate with. I will certainly study this long and hard the next time I'm attempting a new press release to stimulate traffic. I already see some obvious (well, it's obvious now) reasons that some PR produces bigger spikes for a particular search term than others.

[ronburk]

I've discovered I can just while away the hours surfing this tool and thinking. I just found a term (modest CPC) for which the top cities in descending order are: Ljubljana, Slovenia; Bangalore, India; Chennai India; Tampa, U.S.

Wow, Ljubljana, Slovenia beats Tampa for percentage of search going to a common medical treatment term? Color me cynical, but I'm thinking this tool can be used to locate areas where click fraud is concentrated :-).

[ronburk]

More fun with trends: comparing holidays. Type in the name of three holidays, separated by commas. You get a cool instant comparison of the traffic volumes they generate, and also get nearly identical waveforms for all of them (there is nothing with higher seasonal variation than holiday names).

On the interesting side, these are kind of distorted sawtooth waves. All the holidays I've looked at seem to consistently have a nearly exponential slope up -- but then they do a straight-line spike right before the day.

Or at least I assume it's right before the actual date; I wish they would let us download actual values instead of having to peer at these low-res graphs.

In any case, it's interesting to compare volumes and length of buildup and length of decay for different holidays.