Use Google Analytics to report on the visitors which matter. For advertising response analysis, I tend to favour just measuring visitors from your sales area, who haven't been before, and who stick around.
Just as a matter of interest, I asked a couple of marketing managers this month how they use their website visitor traffic data, and although it wasn't exactly a statistically significant poll, they really weren't using it well. Indeed, I'd go so far as to say that what they were getting out of it was potentially misleading. Both were using Google Analytics, and the fundamental problem was that they were accepting the out-of-the-box settings for most reports. This is a mistake.
For example, if you only sell in the UK, why would you look at a report on "all visitors" to your website? You should simply be looking at visitors from the UK by default. The rest are unimportant. Or perhaps you're appearing high up in a Google search for something irrelevant, like a company with a similar name? You'll be getting lots of visitors who take one look at your site, realise you're not the company they want, and leave. Again, you need to filter these out if you're to see what's really happening on your website. One of the marketing managers above pointed out to me how Google Analytics showed his website traffic had increased from 1,800 visitors a month to over 3,000 recently. Very nice. I asked what his total universe of prospects was, and he reckoned between 1,000 and 2,000. The penny dropped. It would seem that either every single potential customer in existence is visiting his website every month, or he's measuring a lot of visitors who are no good to him.
Those of you who use us to manage your Google AdWords campaigns will know that we use your Google Analytics to report on the visitors generated by the advertising, but we look at the visitors which matter. I tend to favour just measuring visitors from your sales area (e.g the UK), who haven't been before, and who stick around on the site. When you're looking at the response from an advertising campaign, that's a sensible benchmark. For other exercises, you might want to set up different definitions of "quality visitors".
I appreciate this can take a bit of thinking to set up. Indeed, many of our Google AdWords management clients have said that the help we give them with their website visitor traffic analysis is as useful to them as our main job of taking the AdWords campaign off their hands. But measuring website traffic properly is worth doing.