"On the internet, nobody knows you’re a dog."
I’m somewhat fascinated by the recent report from a startup that roughly 80% of its Facebook clicks were coming from bots. Since of course advertising dollars are priced per impression, that translates to paying 5x more than they should be, which of course they’re rather distressed about. The company, Limited Run, is not pointing fingers at either Facebook or any malicious agents, but one has to wonder about the potential of this sort of spoofing.
One assumes that it is rather widespread, due to the classic principal-agent problem – many of these corporate Facebook advertising accounts are managed by agencies. These agencies don’t have any particular reason to verify the accounts or identities behind the clicks coming into corporate Facebook pages. They’re not on the hook for the bills, and actually look better when click-through numbers are massively inflated. The corporations paying the bills would presumably care, but have outsourced the actual management and are just happy to see the inflated click-through numbers coming from their agency partners. Basically a fraud like this has a lot of room to run.
For a small startup like Limited Run, I would guess the most likely source is a rival startup, but not necessarily. I think it’s hard not to wonder if this is ultimately coming from Facebook itself. Obviously none of these bots are originating on Facebook servers and there is not a shred of evidence to tie them to Facebook. Nobody here is stupid.* But Facebook is well-acquainted with the principal-agent problem that could make this sort of fraud effective, and is feeling some pressure from Wall Street right about now. It would be trivially easy for them to pay some teenagers under the table to make this happen.
In fact, you don’t even have to posit a very large conspiracy. Maybe the manager responsible for the Limited Run account needs to juice his numbers and dabbles in some programming on the side. It’s not like it’d be hard to write the bots and run them off of a disguised server. The fraud question is one that poses a very serious obstacle to monetization of social media, and one that I’m shocked isn’t blowing up today. Let’s also take the story today that Mitt Romney bought large numbers of Twitter followers. I believe fully that those “followers” aren’t real people. But who decides, especially when metered billing is involved? In a pay-per-click or pay-per-tweet scheme, this sort of fraud has the potential to wreck everything.
I see a large opportunity here for identification verification services that can be sold to marketers.
*: Though of course very smart people can sometimes be just mindblowingly idiotic.