Google Analytics is the most robust free web analytics software package out there. Setup is extremely easy, but it can have problems. The most common mistakes I come across are with tracking traffic sources. Traffic Sources are supposed to show you how people got to your website. In Google Analytics, these are broken down into three main traffic categories:
- Search: Users who found you via a search engine, whether it’s from an organic listing or an ad (PPC)
- Direct: Users who typed in your address directly into a browser
- Referral: Users who landed on your site by clicking a link from another website
In this article, we will explore why some of the data is incorrect. This can be because the traffic was incomplete, mislabeled, or incorrectly categorized
Search Traffic
For the most part, search traffic shows everyone who found your site through a search engine, and what keywords they used when they found you. There is a problem with this though.
1. Keyword data is hidden
We aren’t getting the whole truth from GA here. In November 2011, Google stopped displaying keyword data in their Analytics package. Here are a few scenarios that will cause (not provided) entries to show up in your Google Analytics pacakge:
- When a user uses Google with an https:// (rather than just http://) in the address bar of their browser
- The search field box in Firefox browser
- The Google Chrome “Omnibox” (entering their search directly into the address bar)
- Anyone logged into any Google product
Even though Matt Cutts originally said it affects a signal digit percentage of searches on the web, within a year, 39% of keyword data is “not provided”.
One thing you can do is used this custom Google Analytics report to help you keep track of the impact of this change over time. Also, here are a couple of steps to estimate the impact of Google http > https change. Remember, we are comparing this traffic vs. Google traffic only, not total site traffic, or even total search traffic.
- Open GA
- Go to Traffic Sources > Sources > Search > Organic
- Look for a keyword called: (not provided)
- Compare the total Visits (scorecard under the word Explorer on top) with the Visits you see in this row
Direct Traffic
If you consider this as the definition of direct traffic:
All traffic where a referrer isn’t specified.
This goes beyond just users who manually entered (copy & paste, clicked a bookmark in their browser, etc), it is any visit that isn’t from a link on a website or search engine, including URLs that do not have a campaign code. So that is essentially any visit that isn’t from a website. Direct traffic is for catching “everything else”. So there can be some odd stuff in there
2. Traffic from Mobile Apps & Social Sources
Think about how many people use apps to follow their Facebook or Twitter stream on their mobile devices. Apps like these don’t always pass a “referrer”, so it looks like direct traffic. Some apps will automatically shorten links, adding a referrer by redirecting the user through another address (the official Twitter apps send traffic through t.com domain). In these instances, it allows social traffic to be categorized more accurately.
3. Traffic from Email Programs
Send out any newsletters? Not everyone uses some sort of webmail, some of that traffic you receive can be opened by installed programs like Outlook. Desktop software doesn’t pass referral data like browsers do. In instances like this, make sure you add campaign tracking codes to any email marketing.
4. Bad Campaign Codes
If GA sees a campaign tracking code, it ignores the referrer. But if there is any problem with the tracking code itself, it doesn’t roll back to the referrer (if any), it automatically categorizes that visit as direct traffic.
5. Traffic from You
If you don’t have filters properly setup, there’s a high probability there’s a lot of traffic on your site coming from you or your team. What you need to do is tell GA to ignore your public IP address so it is removed from your stats. Hopefully you have an internet connection with a static IP address. Most offices do. I use Verizon FIOS at home, and while it’s technically not a static IP address, it rarely changes except when the router has been unplugged for an hour or so. But that only happens during power outages for me.
6. No Cookie, No Data
Back in the day, webstats would analyze log files on the server, but GA nneds cookies to work properly. If a visitor rejects cookies, you’ve got a hole in your data. Or in other cases, any time a previous visitor clears their cookies, it’s like they are a new unique visitor on their next visit. Currently there is no way around this.
Referral Traffic
In theory, referral traffic shows visitors who came from other websites, but in practice, it can often include traffic driven from social media efforts, email, or sometimes even a user visiting one page of your site to another!
7. Traffic from Social Sources
With all the effort that goes into your social media campaigns, you need to make sure you measure it’s true effectiveness of those efforts separately. Social traffic is categorized as referral or direct. While Google is working on making their social tracking better (it has it’s own node under “Traffic Sources”), any clicks on tweets or posts without a tracking code and within a browser will be lumped into referral traffic. There are two things you can do: (1) Add the URLs of your “Social Sources” in the Admin > Social Settings area. This will help GA segment social traffic from other referrers. (2) Some social media tools (like the pro version of HootSuite) can be setup to add campaign tracking code to shortened links. If you consistently tweet or post from a service like this, you can have this code added automatically to all shortened links.
8. Traffic from Web Mail
Even if your newsletter we mentioned above isn’t opened in a program like Outlook, webmail users such as Gmail or Yahoo! Mail, have their traffic recorded as a referral even though the real source was your newsletter. This is something you really want to track separately as well. This will always be true without adding your campaign tracking code. In these cases, use something like the URL Builder from Google in every link in your emails. This helps keep your direct traffic clean as well as tracking what visitors from your emails are actually doing.
9. Full URL links vs Relative Links
Link issues often lead to inaccurate referral traffic data. When you create a link from one page to another, the little decision about giving it the full address vs. the relative address (just the URL path without the domain name) can lead to inaccurate referral traffic data. Normally this issue doesn’t pop up for most users, but if when looking at your referral traffic report you see pages from your own site in there, check the links on those pages. In the instances I have seen this occur, changing the full URL to just the relative URL has cleared up the issue. Luckily enough, this doesn’t affect most users.
Conclusion
Google Analytics is an amazing tool, and is constantly getting better. Traffic sources are extremely diverse, and any type of stats using Javascript & cookies along means there are inherent limitations. With an improper setup, it only gets worse. I hope to make this list more comprehensive over time, but these issues are the ones that have occurred often enough in GA profiles I view, that I wanted to share common mistakes people make. If you know of something missed, please let us know in the comments below.