{"items": [{"author": "Kleber", "source_link": "https://plus.google.com/114937925666302803969", "anchor": "gp-1436406597156", "service": "gp", "text": "What User Agents are the sources of the HTTP traffic?", "timestamp": 1436406597}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1436446557587", "service": "gp", "text": "@Kleber\n\u00a0Looking into this, I think there's bias from my site being http-only and browsers coming from https sometimes suppressing the referrer entirely. \u00a0For example I see google traffic on Firefox as 99% http, but I think that's probably suppression.\n
\n
\nThat aside, here's what I see, looking at the fraction of accesses that are https where the referrer is google. \u00a0(Restricting to UA groups with > 200 google referrals in the time period.):\n
\n
\n97% (n=18219) Safari on iPhone\n
\n96% (n=31399) Chrome on Windows\n
\n96% (n=9517) Chrome on Mac\n
\n89% (n=5976) Safari on iPad\n
\n86% (n=1188) CrOS\n
\n70% (n=1734) Chrome on Linux\n
\n41% (n=18927) Chrome on Android\n
\n37% (n=4952) Safari on Mac\n
\n7% (n=488) MSIE 10\n
\n2% (n=2356) Android Browser\n
\n1% (n=2105) Trident/7\n
\n1% (n=6486) Firefox\n
\n0% (n=522) MSIE 9\n
\n0% (n=450) MSIE 8\n
\n0% (n=255) Opera Mini", "timestamp": 1436446557}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1436447519644", "service": "gp", "text": "Yes, it does look like the problem with my stats is browsers that hide referrals when going https -> http. \u00a0Looking at traffic empty referrers I see 2.9x more FF traffic than Chrome/Windows traffic, while with non-empty referrers I see only 0.7x as much.\n
\n
\nHere's the major non-bot UA groups for traffic with empty referrers:\n
\n
\n304496 Firefox\n
\n105425 Chrome on Windows\n
\n102089 Safari on iPhone\n
\n87571 Safari on Mac\n
\n66057 Chrome on Mac\n
\n32617 Chrome on Android\n
\n27684 MSIE 6\n
\n22951 Chrome on Linux\n
\n17742 MSIE 8\n
\n14559 Safari on iPad\n
\n12610 Android Browser\n
\n12019 Trident/7\n
\n11842 MSIE 7\n
\n...\n
\n
\nAnd for non-empty referrers:\n
\n
\n397015 Chrome on Windows\n
\n315716 Chrome on Mac\n
\n277064 Firefox\n
\n206530 Safari on iPhone\n
\n182378 Chrome on Android\n
\n119266 Safari on Mac\n
\n85842 Chrome on Linux\n
\n81119 Safari on iPad\n
\n47388 Android Browser\n
\n44550 Trident/7\n
\n25911 MSIE 6\n
\n23257 Googlebot\n
\n16630 MSIE 9\n
\n13510 CrOS\n
\n13414 MSIE 8\n
\n10502 MSIE 10\n
\n...", "timestamp": 1436447519}, {"author": "Kleber", "source_link": "https://plus.google.com/114937925666302803969", "anchor": "gp-1436449148217", "service": "gp", "text": "All browsers strip the organic referrer upon clickthrough from HTTPS to HTTP. \u00a0Google has used a variety of techniques to deliberately add back in a referrer header -- there's the <meta> referrer tag and there's a javascript-based HTTPS-to-HTTP downgrade redirector on \nhttp://google.com\n, and maybe some others, depending on the capabilities of the browser. \u00a0I expect the downgrade redirect explains most of your HTTP \nhttp://google.com\n referrers.", "timestamp": 1436449148}, {"author": "Jan-Willem", "source_link": "https://plus.google.com/100580955183019057735", "anchor": "gp-1436450819705", "service": "gp", "text": "Huh. \u00a0Didn't know about the downgrade redirect. \u00a0Does it provide search terms? \u00a0I assume not.", "timestamp": 1436450819}, {"author": "Kleber", "source_link": "https://plus.google.com/114937925666302803969", "anchor": "gp-1436453104144", "service": "gp", "text": "I think it only indicates that the source was \nhttp://www.google.com\n, but I'm not positive.", "timestamp": 1436453104}]}