3 ms·
Hard to know for sure, but the discrepancy between clicks and referrals is probably because not all of the promoted tweets will be shown on HTTP twitter. DuckDu
by yahelc 15y ago
Hard to know for sure, but the discrepancy between clicks and referrals is probably because not all of the promoted tweets will be shown on HTTP twitter. DuckDuckGo should know better than anyone that HTTPS to HTTP traffic doesn't leak referrer.
- epi0Bauqu 15y agoRight, that's why I counted up the requests of the images.
- yahelc 15y agoNow I'm confused. Won't the images have the referrer of the requesting page, not Twitter.com? http://d.pr/ZFHF http://d.pr/ZFHF EDIT: HN won't let me reply directly, but realizing I misunderstood. You're counting total image loads, not image loads with the referral. My bad :)
- epi0Bauqu 15y agoThere are two distinct groups of clicks being counted. For clicks to the search engine (from the micro-site) I can use referrer stuff. For clicks to the micro-site itself, I disregarded referrer stuff (for the reason you said and from previous experience I know Twitter clients often omit them), and looked instead at the requests of the images on the page. So what would click to the page but not request an image? Either they left immediately (though even then I'd expect the first image to be at least requested) or a lot of people have no images enabled (very unlikely for mainstream Twitter users) or it is some form of click fraud (e.g. bots, miscalculation?) or something else?
- DrJosiah 15y agoNon-https images? It is not uncommon for browsers on https sites to not load http images.
- epi0Bauqu 15y agoIt isn't an https site.
- mike-cardwell 15y agoI would say it is uncommon.