As Facebook gears up for its big IPO day tomorrow, its hallmark metrics -- advertising and engagement -- have started to look as murky as ever. Just yesterday, for example, due to fears about the utility of Facebook ads, General Motors pulled its $10 million dollar ad campaign from the social network. And just as advertisers have started questioning the value of "engagement," the very idea of engagement doesn't look so stable these days. Last week, Buzzfeed's John Herrman questioned Social Reader's value, showing a marked decline in engagement -- possibly because of some changes Facebook made in how they displayed headlines to readers or possibly because people were losing interest in the products. Now video apps are getting the same treatment, with GigaOm's Om Malik charting the stark decline in interest of these apps. Again, it's not clear if the decrease comes from Facebook or from user fatigue, but either way, something's not working. And, as engagement is one of those buzz-wordy metrics Facebook uses to attract advertisers and prove its value, this comes at a pretty awful time.

Like social reader, both SocialCam and Viddy, two up-and-coming apps, saw the bad kind of spikes in their engagement, as Malik shows via some scary charts. They got a big rush of users all at once and then the numbers started to drop. Over at AllThingsD, Mike Isaac outlines the possible theories for these spikes:

  • It has to do with Open Graph integration. These apps saw spikes when seamless sharing kicked in on Open Graph. The apps started showing up in the stream, so more people decided to join. 
  • Facebook tweaked Open Graph to benefit these two apps, which it happens to own or users got over the shiny allure of a video app. This is possibly what happened with Social Reader, too. Facebook decided to turn on some sort of switch to make these apps show up more in Open Graph and then it turned off that switch.
  • These apps gamed the system. This idea comes from tech entrepreneur Matt Galligan, who discovered SocialCam forced users to share on Facebook, even when they had opted out. SocialCam has since said they've fixed this bug. 

Though some of those reasons rank more nefarious on an evil scale, each is problematic for Facebook as it tries to prove the value of engagement. If Facebook has that much control over what its users see on Open Graph, it could face the same type of anti-trust concerns as Google. Facebook owns both SocialCam and Viddy. Making these apps show up more often in the stream looks like favoritism. If it's just that users have started losing interest in these apps, Facebook has a different kind of problem, as Malik explains. "I had pointed out that this was going to be a short-lived bump and the focus for apps cannot be on raw numbers but on retention and usage. That is, and will remain, the hard part of these apps. Finding sustainable growth is not an easy challenge," he writes. And if the apps are forcing engagement, that won't sit well with users over time. 

At its best, app "engagement" means Facebookers have found an app they like and are heading to the site to use it. That means more time spent on Facebook, which should attract advertisers. But so far that term doesn't quite mean that. Rather, various factors inflate the numbers and users drop off for one reason or another. In the current model, engagement is an ephemeral state. 

These concerns have poor timing, as Facebook embarks on the public market. But not even the GM news has phased investors, as Facebook announced it would sell 25 percent more shares less than 24 hours after the big advertiser pulled out. Social media bubble, anyone?