Understanding Your Traffic Sources, Part 2: Direct Traffic

October 16th, 2013 by Josh

This post marks the second entry in our series on Understanding your Traffic Sources. In Part 1, I introduced the metrics through which we’d be looking at traffic and talked about the split of traffic into direct, social, external, and search. Today, we’ll be talking through what’s probably the most mismeasured segment of your audience: direct traffic.

Defining “direct”

When we say “direct traffic”, we’re referring to those visitors who come to your site’s homepage deliberately, as opposed than those who come to your site via a link to a particular news story.

Typically, people measure direct traffic by looking at the HTTP referrer for each page view: when you visit a page, your browser usually records the referrer that sent you to the page; if no referrer is recorded then it should mean that you either typed in the name of the page or visited via a bookmark.

There’s a bit of subtlety to that story — many sites that use HTTPS (email sites and some apps are notable examples) don’t send HTTP referrers and are frequently misclassified as direct. The phenomenon of external traffic masquerading as direct has been termed “dark social” and discussed often on the web (e.g. Alexis Madrigal’s defining piece for The Atlantic and Lauryn’s post on the Chartbeat blog). In short, it’s very unlikely that a person would truly come directly to a news article, that they’d actually have typed in the URL for that particular article, and it’s much more likely that this traffic is coming from a source that doesn’t send HTTP referrer information.

So, a certain amount of traffic is often over-attributed as direct when it’s actually coming from external traffic sources. But what about cases where traffic is misattributed the other way, where direct traffic is classified as external?

I’d argue that this is the case with what’s often called branded search, but is better termed direct search. Look at search data in your favorite analytics dashboard and you’ll likely see that all of your top search terms are words related to your domain. If someone searches “Chartbeat” and clicks on a link to our homepage, I’d argue that that’s essentially equivalent in spirit to them typing chartbeat.com into a browser — that person intended to visit our site. Indeed, we see that visitors who come to a site’s homepage via a search for the site’s name exhibit very similar browsing behaviors to those who are traditionally measured as direct, and in fact these visitors are highly likely to come back “truly” direct the next time they visit.

quote-1

Direct traffic’s quality

Since direct visitors typically land on home pages and section front pages — pages that link to other content rather than providing content themselves — it doesn’t make sense to look at engagement on the landing pages themselves because these landing pages are not designed to produce engagement. Rather, it makes more sense to ask how deeply people read upon arrival and how often they come back. From both of these perspectives, direct traffic far outperforms the norm: on a typical site, we see that visitors who come directly come 3-4 times per week and view 9 articles across the week — a drastic difference compared to the average visitor, who visits 1-2 times per week and views 2-3 pages.

quote-2

In that sense, we don’t need to think about whether direct visitors are coming back or whether they’re reading stories, we want to think about how frequently they’re coming back and how deeply they’re reading. This is, perhaps, best expressed in terms of the number of daily direct visitors versus the number of weekly or monthly direct visitors — we’d like to push this ratio as high as possible, so those who come directly are doing so every day.

How much should you get?

If direct traffic represents your most loyal audience, it’s natural to ask how much direct traffic you should hope to get. The answer, unfortunately, is that there’s a very wide spread. The figure below shows the breakdown of how much direct traffic sites get.

On average, about 50% of visitors to a site come as direct traffic, but the distribution has an interesting bimodal shape — there is a set of sites for which about 30% of visitors come direct and another set for which the direct rate is about 75%.

quote-3

If your site is in the former camp, increasing the fraction of direct traffic should be a top priority. That means outreaching to the sources whose readers are most likely to convert to direct (see here for a figure analyzing this statistic across the web), increasing the branding on your site to make sure that side-door traffic knows what site they’re on, and publishing more content to encourage visitors to come back more frequently.

A caveat and conclusions

The picture that we’ve presented of direct traffic so far is a positive one — direct visitors visit often and account for a disproportionately high amount of the total content consumption on a site. There’s one caveat, though, which is that direct traffic is almost by definition composed of people who are not new to your site. Those sites with the highest direct traffic have the lowest rates of new visitors. If that issue rings close to home, the first place to look for new visitors is the social web, which will be the topic of the next post in this series. I hope you’ll stay tuned.

  • chrispirrotta

    Hi Josh, great article and thanks for the insights. I have an interesting scenario where we see 60% of our direct visits are considered by GA as ‘new visits.’ One possible explanation I was considering was these 60% new visits would be a customer returning to our homepage after a purchase? i.e., coming out of a https into http? Any thoughts?

    • Josh Schwartz

      That seems possible, though I’d be surprised if GA didn’t measure correctly through an internal transition across https. Are both sites on the same subdomain?