43

I'm starting to block lots of javascript trackers using the Ghostery browser extension for reasons.

Currently Stack Exchange seems to be using the following tracking technologies across multiple sites according to my brief analysis:

Personally, I was stunned recently at how much more responsive the site was under marginal network connections with everything but gravatar blocked, but I also don't want to kneecap the site's revenue and research data. With broadband, I'm not seeing very measurable issues, but still am learning what/how these work in practice.

Would someone go out on a limb and rank these in terms of which ones are most critical to the site as currently designed?

I'm more tempted to unblock trackers when they are few and/or justified by the site owners with somewhat reasonable explanation on how they support the mission. It's clear my experience is better blocking these trackers, but I'm looking forward to learning the provider side of the story as well.

Thanks to anyone that can point me to where this is already documented or can answer below in a semi-official capacity.

  • 4
    it's probably documented in 1984 – gnat Aug 13 '15 at 21:29
  • 1
    @gnat Heh - I doubt orwell would have scarcely imagined the complexity and value of data we sling about these days. I also have pretty high expectations that someone at SE is both already way on top of this and has a really awesome reply once this finds a way to their attention. – bmike Aug 13 '15 at 21:35
  • @bmike We think hard before we add Yet Another Tracker to Stack Overflow. Having said that, I passed this along internally to see if someone who knows more about how this data is used than I do (i.e. beyond "Adzerk is what serves our ads, please don't kill that") can respond. – Adam Lear Aug 14 '15 at 0:12
  • Also, I'm curious - have you measured the performance difference? What sort of "more responsive" are we looking at for you? – Adam Lear Aug 14 '15 at 0:13
  • 3
    @AnnaLear I was basically on dial up for a week and disabling all non-native JS was the difference between 2 second page renders and 30 second "why even bother". Also, I measured battery drain over 30 minutes of "working" on the site. With a constrained CPU on my MacBook - it was strikingly more idle without all the "tracking" which translates into maybe another hour of work on one charge if I were to only browse the site. You can install network link conditioner and experience this yourself. Choose slow DNS and 2000 ms delay up/down. 😈 – bmike Aug 14 '15 at 2:07
  • 3
    @bmike Doesn't sound like as big a deal in that case (although I believe it still serves things like community ads and such), but if we're gonna give general advice, I hesitate to say that it's totally cool to block Adzerk. Joel went into a bit more detail on this, but we do our best to keep the ads relevant and inobtrusive, so unless you're stuck on a really crappy/limited connection... you know. :) – Adam Lear Aug 14 '15 at 4:14
  • IIUC ados.js isn't served from static.adzerk.net anymore, but through SE's standard CDN (or at least that's what this post suggests) – Vogel612's Shadow May 10 '17 at 18:43
57

Adzerk serves many of our ads.

There are three analytics beacons, used to gather statistics on site usage:

  • Google Analytics (which gives us lots of raw data we use internally)
  • Quantcast, which is widely used by the Internet at large to see how big we are, and
  • ComScore, aka the "ScoreCard Research Beacon," which many of our advertisers require to see how much traffic we get.

Basically those three are necessary to measure the amount of traffic we get, where it's coming from, etc., and to document that traffic for advertisers who are paying the bills here.

(There are certain stupid reasons why it's not enough just to use one tracker. Google Analytics gives us very detailed raw data but doesn't show it to the world. Quantcast publishes rankings which are important to advertising clients, investors, and journalists who only want to bother with Important Websites, but doesn't provide detailed raw data. And the biggest advertisers won't trust traffic data unless it comes from ComScore, and ComScore runs this funny business where if you don't put their tracker on your site, they "estimate" your traffic and somehow their estimate is always about 5% of your real traffic, so you need the tracker.)

The bottom line is that we really need those three to be operating as an ad-supported website. That said, we get 100,000,000 visits a month... if you're in the jungle on a 1200 baud modem, and we just can't count your page views, we'll manage.

Anything else you see (facebook, doubleclick) are not something we put there; they could be coming from ads or indirectly through Quantcast but I'd have to look into it more closely.

I'm sympathetic with Ghostery and run it myself; even on high speed connections it's astonishing how sites like The New York Times load 27 unexplained things and don't stop spinning and scrolling for half a minute. That said we have been very careful only to install the absolute minimum amount of junk, we use AdZerk because (among other reasons) it's one of the fastest ad servers around, we have never used flash, ad-networks, third-party ads, interstitials, or even animated GIF ads of any type, and our "Ghostery Index" is about as low as you'll find on any ad-supported website on the Internet.

  • 20
    Answers like these make me so glad for the community. I'm finding that it's far easier to trust beacons from sites I otherwise trust and have good explanations and context for the partners they choose and link to. – bmike Aug 14 '15 at 7:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .