Facebook advertisers struggle to track sales after Apple privacy changes
For Facebook advertisers, the new world under Apple's strict data regime is making it more difficult to do even the most basic analysis, like detecting whether a consumer who saw an ad was the same person who made a purchase. And this fuzziness in the system comes from more than just one glitch that Facebook reported last week, according to advertisers.
Jeff Richards, a paid media consultant who manages digital advertising for multiple brands, says Facebook's ad platform has blind spots ever since Apple started cracking down on tracking on its iPhones this year. The issue goes beyond the “bug” Facebook reported last week, which affected app-install campaigns, Richards says. Facebook underreported the number of downloads marketers received from ads that ran on certain iPhones.
Richards breaks down the wider challenge for brands with a recent example: One of his clients, which he declined to name, recorded 20 purchases on their website. The purchases seemed like they would correlate with a Facebook ad campaign that was running at the same time, but Facebook Ads Manager reported zero conversions. Conversions are the measurement of how many people took an action from an ad, like making a purchase.
In this case, Richards says the brand saw zero return on its ad dollars, even though it was clear consumers were shopping. “There's definitely a mismatch between the conversion value reported and the actual conversion value,” Richards says.
Last week, Facebook sent a note to ad buyers about fixes to its frayed data connection to Apple iPhones. Facebook said advertisers should start to see higher conversion values. “We will be expanding modeled reporting for Website Purchases Conversion Value,” according to the note, which was obtained by Ad Age. Website Purchase Conversion Value is how Facebook calculates when a sale registers on an advertiser's own website after being referred from a Facebook ad. It's exactly the kind of data that is being lost in Apple's anti-tracking vortex.
“With this update, statistical modeling will be used to estimate associated values of additional modeled conversions where events cannot be observed directly (for example, where conversion data may be partial or missing due to data transmission limitations),” Facebook's note said. “As a result, you may see an increase in reported conversion values and other associated metrics like [Return on Ad Spend.] This update should help our bidding system learn from more representative data, which should then lead to more efficient performance.”
Even this fix might not be good enough, Richards says. Facebook is relying on “statistical” modeling in the cases where it is not getting direct data from Apple. “There is a gap, and Facebook is attempting to make up that gap with statistical modeling,” Richards says. “Basically, they're making an assumption of how many conversions there are from these users.”
What's happening is that Apple updated iOS, its iPhone software, which forced apps to receive permission from users to track their activity on devices. If a consumer withholds their permission, then when an ad leads to a website outside of the platform, Facebook loses site of the conversion. Flurry, an app analytics company, has said the opt-in rate for tracking is about 25%, but that varies depending on the app.
A Facebook spokesperson said the company has been communicating with brands and ad buyers as the platform adapts to changing standards around data. Facebook also is working on new measurement programs using what it calls “privacy-enhancing technologies,” which apply concepts like encryption and anonymized audience data to analyze ad campaigns.
“We've been clear with our advertisers that there will be a transition period as the industry, and Facebook, evolve,” a Facebook spokesperson said in a statement to Ad Age. “We will continue to partner with our clients to help them navigate this uncertainty and to share what we're building in both the short and long term, such as our work on privacy-enhancing technologies.”
Bryan Cano, senior director of media strategy at Stitcher Ads, a social media ad tech platform that works with dozens of brands, says that most sophisticated Facebook advertisers are dealing with the same shortcomings in the reporting. Cano says that all ad platforms, including Snapchat, TikTok and Pinterest, are having similar issues, but that Facebook is historically the highest-performing channel, so the signal loss feels like a bigger deal when a brand sees reported conversions fall off a cliff. “For marketers, it's completely disrupted their flow,” Cano says. “I'm seeing the exact same thing where I'm driving sales. I know I'm driving sales and Facebook is underreporting.”
Facebook's fix, using statistical modeling, is meant to restore confidence and show brands that the ads are still performing. “It's going to give a lot of brands visibility back,” Cano says, which comes just in time for the fourth-quarter holiday splurge. The measurement data is crucial to knowing which “campaigns to invest more into, which ads to pause,” Cano says. “Unless you're running a very simple marketing mix, which most brands aren't.”
In some cases, Facebook is working directly with brands to test programs like Private Lift Measurement, which analyzes pools of purchase data from a brand, not tied to one individual, to report an average return on ad spend back.
Advertisers, however, are growing frustrated as internet companies like Apple and Facebook battle it out. Last week's “bug” appeared to be a case of the two companies just not working together. Facebook's problem was with how it integrated with Apple's SKAdNetwork, a platform that controls the flow of information from iPhones back to apps. Facebook advertisers were seeing 10% fewer “conversions” than they should have. That's clearly a metrics error that goes against Facebook, says one advertising executive at a large holding company, who spoke on the condition of anonymity.
“It is in Facebook's interest to count every single conversion they generate to the extent that is possible,” the exec says. “The real problem here is that Facebook and Apple have a corporate rivalry that is downright unhealthy, which likely inhibited the diagnosis of this issue in the first place.”