Conversion and Analytics

I came across this in my reader today...In case anyone's job hunting...

Some good tips on how-to write a web analytics resume.

Last night I faced a problem with Google Analytics: if you run multiple Adwords Accounts for the same site (for ex.: different languages) GA separates paid /non paid traffic just for one of them. I might be wrong?.

I actually got to listen to Joe Shantz from Turner Broadcasting speak about this very issue at a digital summit.

He went through the basic kinds of web analytic services:

Cookie Based (Fireclick, Omniture)
Traditional Panel Based (Nielsen Net Ratings, ComScore)
Convenience Panel Based (Alexa, Compete)
ISP Based (Hitwise, Quantcast, Compete)

He basically said they all sucked in their own way.

Traditional Panel Based are the best. Cookie Based is the next best. And of course you have to pay for both these services.

Interestingly enough, Mike Moran, of IBM fame, told the crowd that although you can't trust the numbers sites like Compete and Alexa give you, you can trust the trend.

I don't know how true that is though, I've seen Hitwise and Alexa give out two different trends. I guess you have to gather as many sources as possible and go with the average.

No doubt the most controversial of all online marketing measures is the Alexa ranking. Many say it is flawed becuase it's statistical bias - the sample of users are those using the Alexa toolbar. I know for a fact when our staff stopped using the tool bar to undertake submissions for clients our Alexa dropped off for no other reason.

Problem is there are no alternative reliable measures.

I'd say it is still a great ballpark measure - anyone with less than 5,000 are doing great.

5k to 100k plenty of variance

Anything more than 100k probably safe to say small traffic. (But I wouldn't bet my life on it)