this month, The Atlantic posted a fascinating article on social media referral traffic.

In it, Atlantic senior editor Alexis Madrigal makes some bold claims about the history of the web and the way that we consider social traffic, coining the phrase ‘Dark Social’ in the process. 

It’s a great post and I urge you to check it out, because in many ways I agree with Alexis’ sentiments, although I feel that this may not be quite the huge revelation it’s been made out to be. 

Here, I’d like to take a closer look at this and the relative importance and attribution of this traffic….

Econsultancy is very active across several social media platforms, and it’s part of my job to try to measure that, so it’s in my interests (and, hopefully, yours as well) to take a deeper look at this. First of all, Dark Social sounds a bit melodramatic, so let’s start with something simpler: Direct traffic. 

Direct has traditionally been defined as traffic that comes to your page when a user types a URL directly into their browser, or from a bookmark.

In practice it’s slightly more complex, and the amount of traffic defined as direct is increasing. For example, in some cases a user switching from a secure (https) to non-secure (http) site may mess up your carefully defined tracking.

Apple has also recently made changes to the way that iOS 6 tracks referral data, which has added to the direct mountain. 

So far, so what?

The important thing here is that direct traffic converts, and converts well. If we can’t get some idea of where it comes from, then we may find ourselves optimising the wrong things. 

Here’s a look at Econsultancy’s multichannel conversion report, with social, direct and email highlighted:

As you can see, Direct accounts for an enormous amount of our conversions, almost 50% in fact. 

Social is referring around 16% of trackable conversions, so it makes sense that 16% of untrackable direct conversions would be from social as well. However, we should never make assumptions about user behaviour, so is there any way we could work out the ratio of social traffic that’s being masked by direct?

Sort of… 

A light in the dark

The vast majority of Econsultancy’s social traffic refers to individual articles on our blog, so if we check out our direct traffic landing pages and filter by the /blog/ extension, we’ll be on the right track:

We don’t tend to refer to the home page itself, so it’s probably fair to assume that this traffic is primarily from search, however  we’re still left with about 10% of total traffic coming to blog articles for this period, which is a fair chunk of traffic by any standards. 

Of course, this doesn’t account for people searching for phrases like ‘brand Google Plus pages’, and it won’t account for traffic that was driven by information sent in our newsletter The Daily Pulse. Email should be uniquely trackable, but as mentioned, If I forward it to someone, it may not always be. 

Who, what, where, when, why?

If I receive an email, strip out a link, and forward it to someone via Twitter, who bookmarks it and arrives on site later, which channel gets the credit? Is that email, social, or direct/search? 

The point I’m awkwardly making is that it’s probably down to all three, but there’s a number of ways we could weight their importance. 

  • Email gets more credit, as it initiated the total funnel path. 
  • Social gets more credit, as it enabled the sharing.
  • A bookmark gets more credit, as it was the last touchpoint before conversion. 

However, we’re missing something important here. The person who shared. 

The recommendation itself was probably most valuable in creating a conversion, because it was relevant information, delivered in a timely, specific manner by a trusted source: A Friend. 

Of course, the friend wouldn’t be able to share if we hadn’t been putting together the product, and informing people about it via email. It’s circular logic but it’s unfortunately true. 

Multichannel attribution

If we take a look at our extended funnel paths on site, we start to see the real value of direct:

It appears that direct is the most valuable path. The more direct touchpoints in a funnel, the more likely the customer is to convert? 

The magic mix

So what are we to do?

If we take these figures at face value, we can sack our entire marketing and sales teams (and me), because users are far more likely to convert if they get a recommendation from a friend.

Of course that isn’t true, as the friend wouldn’t have heard about us if they hadn’t been wasting time on Facebook, or receiving emails or phone calls in the first place. The trick here is in finding the magic mix that suits each customer best. 

The way customers behave throughout these extended paths makes me assume that bookmarking is important.

For some, an initial email, and a follow-up was enough to convert, but if we look at source six, we can see that they got an email and probably bookmarked something from it for later.

This doesn’t mean the email was less valuable, merely that the customer was busy at the time and came back later. 

Call it what you will

All of this leads to a fairly obvious conclusion. Dark Social is a case of semantics. It’s not all social; it’s the accumulation of efforts by diverse team members, across multiple channels, over time, combined with a final push from an existing customer. 

It’s also not particularly accurate to call it social at all. Perhaps we could define a new source as ‘human’ or ‘Word of mouth’, but even this doesn’t cover it accurately.  

Some of it will just be people happening across things and bookmarking them for later of their own accord, some will be people recommending products and services in the pub. Trying to gauge ratios for your direct traffic is useful, and if you have various ‘non-landing page’ content (Again, we’ll struggle with the language here… “intransient content” possibly?)then you may be able to get a slightly clearer picture of things, but ultimately this comes down to paying more attention to the entire conversion path.

Because everyone is so hell-bent on proving social ROI currently, social is a useful rallying point for this, but to assume it’s the only channel suffering from direct masking is a rather limited view.

Email helps highlight this. I’ve mentioned in the past that we expect email to convert, so we largely measure results on the final conversion rate, but it’s still uncommon for marketers to look at email’s place in an extended path where the final conversion may come from social. 

Ultimately, there may be no answer to this. When attempting to track our campaigns, there will always be unknowns, and in many cases they won’t even be known unknowns, in which case we have to rely on the fact that we’re creative people, and go with your gut. You can attempt to clear things up as much as possible, but in the end it may be more trouble than it’s worth. 

As always I’d love to hear your responses to this in the comments, and if anyone does have any clearer ideas about how to track this kind of traffic then it would be fantastic to get your input.