# The site that refers the most visits to Physics Stack Exchange is…

I was recently talking on chat about the analytics information available to 25k+ users. This is normally pretty boring, but I went to have a look and I was somewhat surprised. These are the top ten sites that have sent us traffic over the past month:

(full statistics here.) A lot of this comes internally from the Stack Exchange network - I suspect, primarily, through the Hot Network Questions Sidebar - but we do have a fair amount of other external sources.

In particular, I was surprised to find that the topmost non-SE referrer is news360.com, a web recommendations app, which is responsible for about 1% of visits. (This is because that chart only accounts for the 3.1% of visits that come referred from external sites. 7.2% are marked as direct visits, and 89% come from search engines, with 97% of those coming from Google.)

Remarkably, the news360 traffic is even bigger than what we get from Facebook: that's also around 1%, but a distinct 5% less than the news360 counts. This is for a site that's incomparably small (4M users according to Wikipedia compared to Facebook's billion+ users, so there's probably something specific at play here.

So, what do I want to say about this? Well, first, this is a curious enough observation as is. Secondly, though, I would like to know if there are any ways we can explore what kind of questions that news360 traffic is landing on. With Stack Overflow, we mostly know where it's going on (HNQs, with all the good stuff and all the bad stuff that entails), but I'm in the dark as to what kind of content is getting extra attention from that external source, and I'm not sure the same kind of intuitions about facebook-driven traffic also apply here.

So... is there something useful we can say about that traffic? And, if so, does it tell us something useful about the site and how we can improve it?

• Looking at your full table, there's some funny data-binning going on; it is improbable that the bottom three referring sites, like the bottom five search engines, would each have sent us exactly the same number of referrals. (The same thing happens in the moderator analytics view, but I haven't been bothered by it enough to complain.) So I wouldn't have too much confidence in the 5% excess of referrals from news360.com over SuperUser. Your question is still interesting, but it's about a teensy-weensy fraction of site traffic. – rob Jan 29 '17 at 15:09
• @rob the binning thing comes from a sampling effect and its discussed in the meta.se post linked at the beginning of the question. You're right that it does introduce a fair bit of uncertainty but I think the conclusion that it's essentially as important as Facebook referrals stands. It is indeed a small fraction of site traffic but I think it's worth investigating where it goes. – Emilio Pisanty Jan 29 '17 at 15:20
• More on the question of sampling; I think I understand what's happening there now. Thanks. – rob Jan 29 '17 at 15:33
• news360 is the 2nd top referrer other than search engines for this site. @rob why is that "teensy-weensy," that seems to contradict the facts. an immediate question is whether news360 is driving a lot of traffic into other SE sites (in which case its general, somehow news360 users are triggering it) and if not, why is Physics special wrt news360? there is some point that news360 seems about the same as combined facebook hits across different facebook referral sites... so maybe news360 is just comparable to facebook in reach... – vzn Jan 31 '17 at 18:37
• @vzn See the full table. Referred traffic is 3% of the total, and news360 is 12% of that, so in fact my initial assessment of 1% of site visits coming from there is too high - it's more like 0.3% of total site traffic. – Emilio Pisanty Jan 31 '17 at 18:49
• ok, oops! did not see/ realize that, am a bit shocked that "referrals" are only 3% of total. then agree with rob. and the answer to your question seems to be that its not worth thinking much about referrals by any sites (the entire pie graph posted) because they are heavily eclipsed by search engine traffic, and search engine traffic is heavily eclipsed by google. and presumably this distribution is fairly uniform across all SE sites. then to improve search engine hits is basically SEO optimization, and yeah it would be nice if SE mgt had a way to give more guidance to sites on that... – vzn Jan 31 '17 at 18:56
• Isn't duckduckgo.com a search engine? – Joshua Lin Feb 1 '17 at 1:17
• @JoshuaLin Again, see the mother-meta post linked at the beginning of the question. Essentially it's just Google Analytics being mean. – Emilio Pisanty Feb 1 '17 at 1:22

I think the intra-SE and external traffic should be handled differently, but for the same goal: to attract more visitors. Maybe two circle diagrams from the internal and external sources could be more useful.

The goal of the site would be to increase the visitor rate. It is an universal rule. Even the never acknowledged "quality over quantity" rule is clearly for that, because LQ content expels visitors, HQ content attracts them.

Attracting more visitors has these major effects:

• Popularizes the natural sciences.
• Helps people to learn physics.
• Attracts new users with a capability and intention to produce high quality content here (which attracts yet more users).
• Makes money for the SE Inc, helping it to take care more to our wishes.

Knowing where the most hits are coming from, it opens the possibility to know the mechanisms and behavior of the source sites, and tune our content and traffic to help it to send more.

An example: news360.com reflects probably to current events, these current events could be focused more on the front page. It could serve as a hint to know, that the site frontpage should reflect more the current events. If the top source site would be, for example, the wikipedia, then it would serve as a totally different hint.

The same could happen also in intra-SE relations, although here the rules are quite different. First, in the case of the intra-SE content, the goal shouldn't be to "rob" visitors from the other side, but a cooperative one: working together to extend the visitor rate of both sides.

• I think it's premature to conclude that we should focus more on current events without at least identifying which page these links are coming from. – David Z Jan 30 '17 at 10:53
• @DavidZ The browsers normally send the full previous URL where they are coming from. Not only the previous hostname. It depends on the SE system, if it uses it for anything. I don't know, how the site stats are organized, this circle diagram uses only the host part. It is information loss. But, the point of my post: yes, it is highly useful data, and it should be used to attract visitors. (Side notes: 1) also tuning the treatment of the newbies is an useful possibility to vitalize the site – peterh - Reinstate Monica Jan 30 '17 at 13:40
• @DavidZ (contd) 2) If the power users / mods of the different SE sites would cooperate in handling the question migrations, it would work also as an intra-SE cross-advertising. This was a reason, why was I always against the current de facto taboo on the question migrations). – peterh - Reinstate Monica Jan 30 '17 at 13:44