Month: November 2014

Online Analytics: URL referrals lost as function of HTTPS adoption

The other day we talked about URL referrars and how they (might) affect Search and Ads engines. Let’s put some math into those ideas.

The default web browsers rules state that only HTTPS to HTTP connections do not carry URL referrars (if you don’t know what we talk about, please read our last 4 blog pages first).

Assume that the web pages are linked in a relatively uniform distribution, and that we have a certain percent of web pages under HTTPS. Let that percent be S.

This is the equation that describes how many URL referrars are missed by Online Analytic Engines : S * (1 – S), with S in [0 .. 1] interval. S = 0, means 0% HTTPS adoption, S = 1 means 100% HTTPS adoption.

https

Interpretation of the formula: if all the websites are under HTTP, then all of them would carry URL referrar as HTTP -> HTTP is fine, if all the websites are under HTTPS, then all of them would carry URL referrar as HTTPS -> HTTPS is fine. The only problem is when we have some websites under HTTP and some under HTTPS. At worse, we would lose a quarter of the URL referrars when we have half of websites under HTTP and half under HTTPS.

Here is an interactive view. Please keep in mind that most clicks are to pages inside of the same website, these estimations affect external links only.

HTTPS adoption (10%) URL Referrals lost %

HTTPS -> HTTP % HTTPS -> HTTPS %
HTTP -> HTTP % HTTP -> HTTPS %

This May, Wired wrote:

Early last year–before the Snowden revelations–encrypted traffic accounted for 2.29 percent of all peak hour traffic in North America, according to Sandvine’s report. Now, it spans 3.8 percent. But that’s a small jump compared to other parts of the world. In Europe, encrypted traffic went from 1.47 percent to 6.10 percent, and in Latin America, it increased from 1.8 percent to 10.37 percent.

http://www.wired.com/2014/05/sandvine-report/

Which means that by now we probably have about 10% of websites under HTTPS, so we are somewhere in the region 1 from the previous image.

Given the pace of adoption of HTTPS, and the fact that many websites have little resources or reasons to adopt HTTPS, the Analytic Engines might see themselves soon in the worst place possible, somewhere on a 30-70% HTTPS adoption, which is going to maximize missing URL referrars (region 2).

Since there is no way back to a 98% HTTP web, the only way forward to maximize Analytic profits, is to move to region 3, full HTTPS adoption, OR to change the default browser behavior regarding URL referrars.

A loss of 20-25% of the URL referrars is bad, but it is not going to break the bank. Probably, Analytic Engines will lose on a low double digits million dollars of annual revenue (just a wild guess).

What does this means for the future, is that we should expect either of these two scenarios to happen in 2015:

  • Someone might be interested to subsidize your free HTTPS certificate and offer you support to manage your secure server
  • OR as Chrome and Firefox are used by the majority of the users, and are backed by the same revenue streams, we might see a change in the default Web Browser behavior regarding URL referrars.

OR, who knows, both.