Many of us in the marketing community are at this point familiar with the subject of dark traffic and dark search. If not, check out the post we published earlier this year about Google Analytics attribution, or another post from our friends at Search Engine Land about dark search. In short, not everything is as it seems in the world of Google Analytics.
Today we were experimenting with something totally unrelated regarding how a bounce rate is defined in GA, when we made a rather odd discovery. If you have ever dug into the various causes for dark traffic, you'll generally see examples such as chats on websites and in apps mentioned as something that can be bucketed as "direct traffic." While traffic that comes from chats such as G-chat, Slack, and Facebook are thought to be totally "dark," today we discovered that that's not really the case. To analyze this, we conducted some experiments.
How is Facebook Chat Traffic Grouped?
According to our tests, Facebook traffic that comes from a link shared in a chat is grouped with the default channel grouping of social. The Medium is social and the channel is Facebook. Luckily, we are testing this on a site that has a valid HTTPS set up, but still loads most of the site under HTTP. This means we can test both states, secure and non-secure. In our experiment, we found that regardless of HTTPS or not, clicks to a link in Facebook chat are grouped as social, not direct as some studies indicate or suggest.
How is Google Hangouts (formerly G-Chat) Traffic Grouped?
Surprisingly URLS coming from Google Hangouts chats tested in our experiment under HTTP and HTTPS were reported as ORGANIC TRAFFIC. Of course, traditionally, we'd expect that this was either grouped as a referral from Google.com or simply bucketed as direct, when - in fact - this couldn't be further from the truth. We had multiple people who have never before been to this website click the demo link and all of them were reported as Organic.
What About Slack?
In our experiment we found all messages sent through the App slack showed up in Google Analytics as direct traffic. This is more along the lines of how we expected other chats to behave.
Firstly, remember that not everything is as it seems in Google Analytics. Referrers can get lost, some browsers play by their own rules, and HTTP and HTTPS are handled differently in some, but not all cases. In this case, it's not likely that anyone website is getting so much chat traffic that it is throwing off their real organic traffic numbers. If you are concerned about data pollution you could try using UTMs in more places, but people organically sharing your links without tracking is inevitably going to happen. If anything, we just wanted to point out this observation and see if any of you have noticed this as well. We will be sending this to the folks at Google for their feedback.