The LinkedIn Algorithm: How it Works
By Joshua BoydDec 13th
Published September 8th 2017
Dark social traffic is traffic that comes from private social channels, such as email and messenger apps. Analytics tools are therefore unable to track their true origin, often classifying it as ‘direct’ traffic instead.
That’s obviously a problem. The more we know about our traffic the more we can understand how our site is performing. If 20% of your sales are coming from ‘direct’ with no context there’s very little you can do with that.
So what social platforms will send traffic that shows as direct? Here are the main types:
So, what’s the solution then?
As a caveat, there’s currently no way to track 100% of dark social traffic. We’re mostly looking at educated analyses or tactics for mitigating the lack of data. But we’ll discuss the limitations of each approach as we go.
There are some things you can do in GA to get any idea of how much dark social traffic you’re getting.
Let’s start with why traffic from WhatsApp or your emails ends up being labelled as ‘direct’. Direct traffic is meant to be when someone comes directly to the site of their own accord. For example, by typing a URL into the address bar or clicking on a bookmark they’ve saved.
Due to how they operate, when you click or tap on a link in an email or messenger app, GA sees this as though you’ve just typed it into the address. When we’re talking about URLs like the below that seems unlikely someone would write them out:
Based on that we can start making some educated guesses about how track dark social traffic.
Start by creating a segment in GA that only includes direct traffic. Then exclude pages that could be typed in or are likely bookmarked. This will probably include your homepage, contact page, and similar.
Here’s our one:
We’ve included the direct traffic in the ‘Traffic Sources’ tab. In the ‘Conditions’ tab we’ve got two filters set up that exclude traffic to certain pages (this is just a sample of the ones we exclude). Working out which ones to block will be down to you and your best judgement.
You’ll notice for most that we’re only excluding traffic from return visitors.
This is because these are pages where it’s unlikely someone will type in directly, but could easily be returned to directly again by a user. One case might be someone being sent a page on Signal, and then navigating back to it using their search history when they need to.
What to include under the returning visitors filter is something to experiment with. Have a play around to see what works.
For different types of sites, different approaches will be needed. For example, a recipe site will have to think about how their direct traffic will include a lot of bookmarked recipes along with dark social.
Once you’ve created your segment you can start to see how much traffic might be coming from dark social sources. It’s not a perfect solution but it certainly offers insights. Make sure to have a dig around to see what pages are being included. Some can often sneak through.
If people are going to share your content on WhatsApp, why not make it easier for them and increase your ability to track it at the same time? It’s a win-win.
Adding share buttons for more platforms achieves both of these aims. First it obviously creates a smoother process for people to share your content, but secondly, you can use the buttons to add a tracking code to the link. Light shone upon the darkness.
If you’re not familiar with tracking codes, they’re pretty straightforward. If you’re not using them yet, get started as soon as. We’ll use Google’s tracking codes as an example for what’s possible.
Here’s a blog URL:
Now, if you share that on Twitter and someone clicks, GA should automatically classify it with a medium of ‘social’ and a source of ‘twitter’ by itself.
The problem is Google can sometimes make mistakes, while someone could take the link from Twitter and share it on a forum. Now any traffic from that will be labelled as ‘referral’ even though its original source was social. Depending on your approach, this could be the right or wrong result.
What tracking codes do is to take Google’s guesswork out of it, and tells GA exactly where the traffic is coming from. We can also add extra details, like a campaign name, to further enhance our tracking.
Using Google’s own Campaign URL Builder, we can create this:
When GA tracks someone coming to that URL it will classify the data accordingly. That means far more accurate reporting.
The share buttons we mentioned earlier can be used to append this tracking onto a link. If we used WhatsApp as an example we could create this:
GA would use that to assign traffic to that URL with a medium of ‘social’, a source of ‘whatsapp’, and a campaign of ‘data_blogs’.
Obviously you can customize this however you like, adding in whatever you prefer.
Of course the caveats are clear. First, a person might not use the share buttons or there might not be one for the the platform they want to share it on. Secondly, people can always cut off the code from a URL if they want to.
Just like with our dark social segment, it’s not perfect, but it does part of the job. Even getting a little bit more accurate data is useful. Combining this approach with the dark social segment can take you a long way.
Brandwatch doesn’t track dark social traffic specifically, but you can use it to discover an unclear source.
Using Brandwatch you can set up queries to track your own content, mentions of your brand, and whatever else you like. Aside from being valuable from a social media monitoring perspective, it can give you some insight into your dark social traffic.
For example, you’re looking in GA with your dark social segment and see a huge spike of valuable traffic coming to the site. It’s converting well and you know finding the specific source is important.
So you hop into Brandwatch Analytics and look for mentions of the specific piece. Around the time of the spike, lo and behold, someone submitted your blog post to a subreddit on Reddit, it got upvoted, and the subscribers seem to be sending it onto their friends.
Now you know where it came from you’ve got a community of relevant people to engage with or target ads to. And it only took a few minutes of research time. This means even if your URL tags get removed, you can still find out where it came from.
We’ve touched on this already, but we thought it would be worth really driving home the importance of dark social traffic.
RadiumOne claims “84% of consumers’ outbound sharing from publishers’ and marketers’ websites now takes place via private, Dark Social channels such as email and instant messaging”.¹
That’s obviously a huge amount. This will obviously differ from site to site, so don’t worry that you’re only tracking 15% of your traffic properly. RadiumOne’s methodology isn’t bulletproof, but it shows dark social can’t be brushed off.
It isn’t just about volume either. Dark social traffic can be very valuable too. For example, on our own site we see a higher conversion rate with dark traffic. This makes sense as you share articles with people you know will like them. It’s hyper-targeting brands can’t achieve on their own. RadiumOne’s report also supports this.
And that’s all aside from the fact that having an in-depth understanding of site’s data is always important for optimisation and improvement.
So there you have it. Get tracking.
¹RadiumOne (2016, September). The Dark Side of Mobile Sharing. Retrieved September 7, 2017
Social is a rich source of data but only if you know how to conduct good social media research. This post explains the methods for finding consumer insights