Bio not provided
@crestodina Thanks Andy!
6 months ago on Cartoon of the Week: Don’t Put All Your Marketing Eggs in One Basket
@Smbatyan If you use "ppc" then you will see facebook / ppc in the All Traffic Sources report and the data will not be included in the Advertising/Paid Search section of the reporting.
1 year, 8 months ago on Google Analytics: 21 Inaccurate Traffic Sources, Setup Mistakes …and Fixes
@Smbatyan Yes, marking the medium as cpc is a feature in GA that allows the data to be included in the Paid Search section. Cost data won't be included outside of AdWords though. If you don't want it to show up in Paid Search (even though one would argue that it still is a form of paid search), simply change cpc to something else.
Regarding #5, this primarily happens as a result of the analytics code being improperly set up across multiple domains or subdomains, or in worse cases, someone having stolen your website and keeping all of the code intact including the GA code. More info can be found here: http://support.google.com/analytics/bin/answer.py?hl=en&answer=1011827. If you want to see examples of how this could be happening or where the issues are coming from, view the Hostnames report by going to the Audience section of GA, click on Technology, then the Network report. Once on that report, below the graph and above the table data is a link for Hostname. Nice and easy to find, huh? This will show you all of the URLs that triggered your GA code and help diagnose what is going on.
Perhaps one of the biggest inaccuracies within GA is time on page and time on site. It's not just what you pointed out in #16 that people will leave a page open, but unless you visit a second page of the website, GA doesn't calculate the time on the page or the time on the site and will use 0:00. This obviously skews the information downwards for time on page/time on site, especially if a site is high in blog traffic (people that tend to visit a single page and go on their merry way).
As you pointed out, bounce rate can be screwy as well and many people look at bounce rate as a problem, whereas something like a phone call would count as a conversion and a bounce if the site visitor only visited one page (also time on page/site would be 0:00 as well). I'll take 100% conversion and 100% bounce rate any day of the week! Including bounce rate across an entire website and across all mediums could realistically be considered an inaccuracy of GA.
Andy - I'm not sure I'd agree with #20. It's definitely true that you can't get data back if you have filtered it out, but best practices for setting up multiple profiles within an account is to keep 1 profile unfiltered and including all of the data. Then create new profiles and apply those filters only to those specific profiles. That way if you wanted to share specific pieces of data with certain departments or people, while hiding other data from them, that could be done. I've not heard of that data being removed from the master profile to fill the new profile.
Awesome idea for an article.
1 year, 10 months ago on Google Analytics: 21 Inaccurate Traffic Sources, Setup Mistakes …and Fixes
I would agree that these statistics are not valuable when looking at your overall data, but these are great things to look at when segmenting and breaking down the data. For example, traffic to the blog should have a higher bounce rate than the rest of the website. If you are working on lowering or increasing bounce rate, you should focus on segmenting out the areas that you are trying to lower to measure the output.
Another example would be a company focusing on generating phone leads - a user visits one page on the site and decides to make a phone call and purchase services. In this case, that will be a bounce visitor, but it will also be a conversion (for those doing phone call tracking). Another interesting piece is that it will register that visitor as 00:00:00 average visit since they have to visit two pages before that statistic can be calculated.
So I think the real questions to ask are what am I trying to improve, what skewed data should I segment out, and what stats should I look at to measure the changes?
2 years, 8 months ago on Moneyball Analytics: 3 Stats That You Should Ignore