If you're reading this, your chair might feel crowded, because there's 63 percent more of you than there used to be. Or 56 percent, maybe. Who knows? Last week, Quantcast, the web-traffic-monitoring service whose numbers are the basis for Gawker's editorial decisions, announced in a vaguely worded blog post that it had performed "a major measurement update" for "even greater measurement accuracy."

This update applied retroactively to Quantcast's data, so that (for instance) this past month of June, in which Quantcast had counted 6.8 million unique United States visitors to Gawker, now accounts for 11.1 million unique U.S. visitors. That is a difference of 4.3 million. Or 4,315,090, to be precise, since we are talking about measurements of individual users who read the site.

And the revision is not a simple re-weighting of factors. Apparently Quantcast's cookie-based measuring system was missing readers on mobile devices. So when the company revised its system to find them, they appeared in irregular batches—63 percent more readers showed up in June, but only 57 percent more showed up in April.


Long ago, when news was delivered in stacks of dried and flattened wood pulp, nobody really saw where it went once it was thrown off the truck, and readers were all a phantom army. There was the old concept of "pass-along," where for every paper that went out, you assumed that multiple people were reading it (that is, encountered the ads)—that one humble copy of your free weekly, left in a coffeehouse restroom stall, might reach several dozen readers, all giving it their rapt attention, so conservatively you should guess that your real audience was two and a half—four—five—hey, why not seven!—seven times the mundane number of actual newspapers you put out. Who could say otherwise?

With news distributed online, the informational mystery goes in the other direction: not multiplication, but division. The servers now record each of the many, many times a piece is accessed—how many times the pages are flipped in all those coffeehouse restrooms—but those voluminous numbers must somehow be distilled down to actual readership figures. (Pace the commenters, nobody gives a shit about "page clicks." ) How many views are going to a single reader? How many are going to that reader's mobile device?


Everybody's distillation comes out differently. The revolution that was supposed to have linked real reading behavior to real results—remember when clickthrough was going to make advertising a precise science?—has produced whole new kinds of distraction. In June, a New York Times story informed readers that Betches Love This was getting "four million unique visitors a month," by Google Analytics numbers. Quantcast (at least the Quantcast numbers of today) puts this year's best month for Betches Love This at 551,538 people.

So there's the Google Analytics number; there's the subscribers-only ComScore number (used by Gawker's ad department). The Quantcast numbers have the advantage of being public and daily. They are great for bragging with:

(And for setting Gawker Media site-performance bonuses with.)

They have the disadvantage, it turns out, of allowing 4,315,090 people to just wink into existence overnight.

[Image by Jim Cooke]