While Facebook may have $19 billion to spare on WhatsApp, so far the social network hasn’t ponied up the resources to address a real threat to its advertising cash mine: click fraud.
In the last quarter of 2013, Facebook sold more than $2.3 billion in ads on a cost-per-thousand impressions and pay-per-click basis, but advertisers who track and measure clicks are detecting considerable fraud on the site.
I’ve learned this firsthand from dozens of our clients who buy Facebook placements. Facebook ads are attracting both people and robots, according to these advertisers, but the bots are diluting performance.
On Facebook, our clients are seeing high rates of fraud because companies have mobilized click-happy bots to raise the cost per lead for their competitors. This practice is frowned upon, but it’s easy to get away with. Even with click forensics, our clients can’t figure out exactly where these fake clicks come from, but their reports on Facebook campaigns make it clear that the source isn’t human.
When advertisers can prove that their clicks are fraudulent, Facebook has been known to issue credits.
A viewable impressions model could help. To avoid getting burned by bots, advertisers should pay for ads based on viewable impressions, not just cost-per-thousand impressions and pay per click. This model would ensure that advertisers pay only when users are actually capable of engaging with the ad. With viewable impressions, advertisers would pay for time slots -- 10 seconds, 15 seconds or maybe even 30 -- and only pay for the time if the ad appears on a user’s screen for that duration.
Bots cannot create fraudulent viewable impressions because they are basically computer programs; they cannot perform the actions that distinguish a genuine user view from a false one.
Mere “impressions” are registered when a web page loads and the pixels fire; viewable impressions are registered when the underlying technology confirms that ad has been displayed for a period of time. This is critical because many advertisers now pay for bottom-of-the-page ads that technically load but could not have possibly been seen by users because they never scroll down far enough on the page. The publisher or ad network charges a fee for ads that never were displayed.
In contrast, if advertisers paid only for viewable impressions, they would not get charged in the above scenario. And if ad-serving technology correlated viewable impressions with clicks, advertisers would be certain that only real people are clicking. Regardless of whether the creative work were a dynamic banner, native ad or video, it would have to load and be in view before it counts.
The viewable impressions model is not only compelling because it would eliminate click fraud, but also because it would help publishers monetize their inventory more effectively. If a publisher could guarantee that advertisers paid only for viewable impressions and clicks from human beings, his or her ad inventory suddenly would become more valuable. So far, no bots can counterfeit viewable impressions and correlated clicks.
So far the viewable impessions model is available on some websites but not on Facebook.
Facebook provides advantages to advertisers. As I’ve discussed on Adotas, advertisers should not desert Facebook. Lead generation on the social network is strong despite rampant click fraud. Advertisers should shift some budget to ad space that affords viewable impressions to reduce their bleed rate. As a fourth quarter study by Adobe Systems found, Facebook’s cost per thousand impressions has also increased 437 percent over the past year.
Facebook’s increasing popularity for advertisers has freed up premium ad space across the web. So, advertisers can save on their budget without sacrificing quality when they bid on this inventory. Real-time bidding on demand-side platforms now allows advertisers to appraise premium inventory and bid on space; this is an arrangement whereby companies can buy space to place their own ads, a service offered by my company along with running a full-service ad network. Advertisers can purchase ads by the amount of viewable impressions, clicks or standard impressions. Moreover, demand-side platforms can target audiences that do not overlap with the Facebook crowd. This is a smart way to diversify an ad strategy.
In turn, publishers determined to win business back from Facebook could use a supply-side platform to offer ad inventory based on viewable impressions. This would be a quick way to increase the value of ad inventory.
If Facebook were to begin correlating clicks with viewable impressions, it could eliminate click fraud. Everyone in the advertising world would welcome this change since the global ad marketplace is nothing without confidence that real, living consumers are seeing and clicking on content. Until Facebook learns how to clean up click fraud, advertisers can bypass the bots by shifting resources to premium ad space, purchased according to the number of viewable impressions. Likewise, publishers can gain an edge over Facebook and social media upstarts by offering this fraud-free option.