How Many Websites Have Totally Bogus Traffic Numbers Due To Facebook Bug?
from the just-wondering... dept
For a while now, we’ve heard of websites claiming that they now get a ridiculous amount of referral traffic from Facebook. We’ve certainly noticed that we get a decent amount of traffic from Facebook, but it’s rarely in the top five sites for referrals. For a while, I’ve just wondered if people just don’t like to pass around our stuff on Facebook (as opposed to Twitter, which does drive lots of traffic — or if, perhaps, we didn’t do enough to encourage people to follow our Facebook page). However, something odd happened a couple weeks ago. All of a sudden, we noticed a ton of traffic coming from Facebook. Before noon, we’d already passed a normal day’s worth of traffic, and by the afternoon, we were on track to more than triple a standard day’s page views. But, then we noticed a few oddities. First, a lot of the traffic was going to relatively old stories. Second, doing a search on Facebook didn’t turn up anyone linking to those stories. Third, and most importantly, looking at the number of unique visitors, as opposed to pageviews, showed that those were more or less in line with a standard day’s traffic. Clearly something was off.
I started chatting with a few folks about it, and Marcus Carab pointed us to information on a Facebook bug that’s been around for about a year, in which Facebook’s “like” button adds a string (fb_xd_fragment, for those who are curious) to the URL that leads to a blank page… often causing multiple pageviews. There are workarounds, though it’s stunning that Facebook — which is pushing use of the “like” button all over the place — has not implemented its own fix. By that evening we’d installed a workaround, redirecting the bogus links to legitimate links, and stopping some of the false reloads. Yet, over the next few days, we noticed that even when we fixed that “known” problem, we were still seeing a hell of a lot more traffic from Facebook than made sense or that we had seen before.
Eventually, we used the referrer URL (www.facebook.com/plugins/like.php?….) to track it down to a problem with the way Facebook’s “like” button behaves when our pages are accessed with IE7 (and possibly IE6). Something in the button basically goes into a loop and just keeps requesting the page that it is on — essentially, repeatedly “reloading.” This makes page views shoot up like crazy. Because of this, if you visit our page with IE, we no longer show you a “like” button. Since doing so, our pageview numbers have returned to expected levels. (For our IE users, now that we’ve confirmed that the problem was the like button on IE, we plan to try an alternative implementation of the like button to see if that avoids the problem.)
Apparently, we’re not alone. Other sites also just noticed that they were dealing with similar issues and put in place similar workarounds.
But here’s the question: how many people don’t realize that these Facebook bugs exist, and are happily lapping up the not-really-there pageviews and reporting them as legit? I would guess that many people who are recipients of such a traffic deluge honestly don’t realize that it’s a bug and that the traffic is phantom traffic. But it wouldn’t surprise me if a few sites are simply happy for any way to “juice” their numbers. Over the years there have been a few big cases of newspaper circulation scandals. It kind of makes you wonder when we’ll start to see something similar with websites that report fake numbers concerning their traffic?
Of course, for us, being honest probably costs us money. Since many ad campaigns work on a page view (CPM) basis, if we can report triple our existing page views, that certainly would boost our ad revenue. But it’s also incredibly dishonest, and, for those who know this is happening, potentially fraud. It seems like only a matter of time until we hear about sites purposely leveraging such things for their own advantage.