Not long ago, Rand Fishkin mused about the possibility of a long-tail to the web’s referral traffic (Is There a Long Tail to Referral Traffic?).
…that only ~20% of the referrals that the average website receives comes from the tail of the distribution curve, whilst Matt felt that number should be considerably larger.
Rand’s analysis of several domains (broken out separately below because the number of sites used for the two groups differed) showed that the top 10 referring domains (between .33% and 1.47% of the total domains driving traffic) drove between 65% and 88% of all referral traffic. Interestingly, this percentage was significantly lower (37%) when looking at SEOmoz.org, presumably because it has been around longer and generates a wider swath of referral traffic compared to the other, less established sites. Yet, even for SEOmoz 9% of referring domains generated 79% (79%!) of referring traffic:
In the post, Rand threw out a challenge to anyone willing to look at this issue on a larger scale. At Conductor, we decided to take up this challenge and look at the referral traffic from a mix of fifty B2B and B2C websites over a 3 month period. In the end, we analyzed over 31 million referral visits.
First, the distribution of traffic types coming to the sites:
Rand’s original analysis excluded search traffic, but we decided to include it below for additional comparative purposes. In doing so, you can see that if accounting for organic search as a traffic referrer, it drives 7 out of 10 visits:
Ten Percent of Domains Drive 91 Percent of Referring Traffic
We then proceeded to tackle the question of referral traffic distribution. We grouped domains by ‘top x% of domains driving traffic’ rather than the ‘top 10, 50, 100’ that Rand used. We also decided to provide further granularity and broke out B2B vs. B2C sites to expose differences in how the two might receive referral traffic.
Analysis of the data seems to support Rand’s assertion that “…the distribution is strongly biased […] to the fat head of the referral demand curve.” The data shows 1% of domains drives 70 percent of traffic, increasing to 86 percent with 5% of domains, while 10% of domains drives a whopping 9 out of 10 (91%) visits.
A comparison of B2B and B2C distributions showed only a slight difference, with B2B referral sites driving between 4-6% less traffic than B2C sites.
In digging deeper into the top domains driving traffic we found what we’d consider to be amongst the most interesting findings of the analysis. While some might have hypothesized that there would be a group of ‘super referral sites—sites that we’d see as among the top referrers across a large percentage of industries–there was little to no overlap amongst the top referring domains between the domains analyzed. If our sample is therefore reflective of the web at large, this suggests that the top drivers of referral traffic are very much domain/industry specific.
When we added organic search to the equation the percentages rise to 91%, 97% and 98% respectively. This is not surprising given that organic search drives a large percentage of overall site traffic (as per our first few charts) and the fact that said traffic comes from only a handful of domains (google.com, bing.com, etc.)
Conclusion: The Referrer Fat Head is Alive and Well
Our analysis of 50 sites and 31 million visits over a 3 month period seems to support Rand’s original findings that a small percentage of domains drive a large percentage of referral traffic. Although it seems somewhat counterintuitive, when it comes to referral traffic, there’s actually a lot less going on than one might think. Juxtaposed against Search where the goal is to continuously move up the search rankings for existing keywords but also continuously expand the zone of coverage, it makes for an interesting contrast.