Facebook Advertisers Are Itching For Change As Bugs Infest Its Attribution Tech
Facebook bugs put attribution and agency companies in a tough spot, though. Marketers are used to judging campaigns by Facebook reports. Nowadays, those reports may show dramatic fluctuations in targeting or return on ad spend. A couple years ago, those fluctuations would be a sign that something strange was happening with the campaign targeting or creative, or perhaps the landing page being linked to.
In other words, Facebook campaign reports pointed the marketer to something that they, not Facebook, should address.
But wild fluctuations are now the norm for Facebook attribution reports and may not say anything about the campaign, other than that Facebook is tinkering with its own machinery as it figures out a new attribution system.
“The biggest issue is communication,” Hladnik said.
Facebook is heads-down trying to fix the ad platform as gears and springs fly out of it like a cartoon pocket watch.
To be fair, Facebook did warn advertisers and ad tech partners that this year would be a mess and that it’s using more modeled attribution, rather than attributing actual conversions.
But Facebook can do more to help its partners through this difficult course correction, said David Herrmann, a social advertising consultant.
For one thing, it’s clear that when Facebook works on platform updates, their own tinkering causes a cascade of weird outputs in ad campaign reports, Herrmann said. Facebook can warn agencies or ad buyers which specific days it will work on platform upgrades, for instance.
“It makes it very hard to manage expectations [with clients],” Hladnik said, echoing Herrmann’s concerns.
Marketers are focused on meeting their goals and often aren’t hands-on practitioners aware of the nuances of different ad platforms.
If the ad buyer can alert the brand that Facebook is working on the backend and to anticipate strange results these days, it can relieve the tensions on some difficult, urgent calls when the reports come in with bizarre numbers.
Another big problem is the delay on reporting data. With its third-party pixel and SDK network, Facebook used to pick up practically every visit or conversion to another site or app. That data was fed in real time to campaign reports.
Now data comes in large batches that upload for thousands of accounts simultaneously every two or three days, Hladnik said.
Again, Facebook did warn advertisers to expect slow reporting and not to optimize campaigns without a few days to build up a level of data so that the platform can model results.
And some advertisers are less keenly affected. Larger brands, particularly with higher-priced items, are less affected by Facebook’s upheaval, Hladnik said.
Toyota’s profit margin does not hinge on the Facebook ad platform, and a massive brand with a diversified ad budget, like McDonald’s, is relatively unaffected by a two or three-day delay on reporting.
But the mechanics of Facebook’s ad platform are built into many ecommerce businesses. Startup or smaller niche brands succeeded on Facebook partially because they are big tinkerers themselves. They target carefully, change spend levels and optimize more often, because they need to slice and dice platform audiences to find those perfect prospects.
Facebook’s consistent response has been to be patient and, well, to slow down. In February, Facebook acknowledged that it was still underreporting attribution, but said it had cut the error margin from 15% to 8%.
If every time the Facebook ad platform butterfly flaps its wings, a tsunami hits the attribution reports, then ad buyers are going to stop trusting Facebook data or promises of performance improvement.
The tools Facebook is building to fix the black box ad platform are an even darker box, Herrmann said. “What are they doing to earn the trust of media buyers or publishers?” Because Facebook doesn’t provide details on what’s being worked on or fixed in the background, even as that work swings ad campaign performance.
Even more than Google, Facebook has built its ad dominance on the idea that its data and algorithms are the best on the market, according to Hladnik. You didn’t have to trust Mark Zuckerberg; Advertisers trusted Facebook results.
“But that’s not the case anymore. You cannot trust the data,” he said. “It makes me question the data from before too.”