Brands Using ‘Incrementality Experiments’ For Facebook Measurement
Brands’ advertising performance on Facebook might not be failing, but their performance measurement is.
Since Apple’s deployment of iOS 14.5, agencies and media platforms have had trouble explaining what has happened to ad performance. But a new method utilizing “incrementality” experiments is showing promising results and could be shifting some power from Facebook back to brands and agencies that are using it.
“Apple’s policy updates are just the latest in a series of privacy-driven changes across the industry causing widespread performance reporting headaches for platforms and forcing brands to adapt how they track and measure,” said Madan Bharadwaj, CTO and co-founder of Measured, an analytics firm that has focused on developing incrementality experiments as an alternative to conventional attribution models.
“We built Measured in anticipation of this moment. We continuously refine our independent experiment designs and adapt to inevitable changes at platforms to ensure brands can always trust the insights they get from us.”
Attribution reporting from platforms like Facebook use last-click metrics, which means they take 100% of the credit for any conversion they were in the path of.
The ad industry’s consensus is that this is a flawed method for understanding the overall advertising contribution, as many consumers will come into contact with a brand multiple times in multiple places before they make a purchase.
In the past, last-click metrics caused significant duplication: major media platforms were taking far more credit for transactions than the brand was seeing.
However, since iOS 14.5 made it possible for users to opt-out of being tracked, Facebook has begun underreporting.
“They're wondering -- what is the true performance of Facebook,” says Trevor Testwuide, Measured’s co-founder and CEO. “When they go into Facebook, it's not looking so good, but they can feel it -- they know it's really driving influence.”
Bharadwaj presumes that Facebook’s underreporting may also be a strategy to dodge any future media backlash.
Measured had formerly relied on a sophisticated methodology that tracked users across platforms and devices, but in 2016, that became difficult.
“We no longer had the luxury to sync IDs across these platforms,” Testwuide explains. “I couldn't identify you from Facebook to Google to Pinterest to a conversion. The walls were going up, those IDs were getting blocked, and now you end up with these walled gardens, and I can no longer build these paths.”
These privacy restrictions and iOS 14.5 inspired Measured to create a full suite of experiment designs and actionable reporting tools that identify the true causality between Facebook media investments and actual business transaction data reported by the brand:
Audience split experiments utilize the Facebook audience split API (applications protocol interface) to divide target audiences into test and control cohorts, applying systematic lift studies to retain program-level and granular adset-level incrementality measurements.
Geo-matched market experiments use advanced matched-market data science to identify statistically significant sibling markets, at the state, DMA (designated market areas) or zip code level, that are similar to but much cheaper to test in than larger markets. Geo tests run independent of platform reporting, so they are the only near-universal system of measurement that cannot be changed by current and future platform and policy changes.
Scale testing experiments utilize the audience split API or geo-matched market method to split target audiences into 3 cohorts: test, control, and scale. The ad sets used for the test group are cloned and served to the scale group, but at 2x-3x-4x the spend to measure the point of audience saturation.
With user-level path-building measurement no longer effective, incrementality experiments and geo-testing by third-party partners like Measured can help brands and marketers stay on track of their ad performance across media channels.
Ettitude, a Los Angeles-based direct-to-consumer marketer that makes high-end, sustainable bedding and sleepwear from bamboo fibers, began working with Measured in March of this year.
“We were trying to get a better understanding of our cross-channel attribution and measurement across our full marketing funnel and also anticipating the iOS14 rollout and the implications of that,” says Marlee Rosenblum, vice president of global marketing at Ettitude. “At the highest level, that's what led to our partnership and it's been going well and proven very insightful since then."
Before working with Measured, Rosenblum says Ettitude was looking at platform attribution and Google Analytics attribution, but that these methods were limited.
“With Measured's methodologies, we're able to look at incrementality to understand, especially for those upper funnel channels, which are helping to drive incremental purchases, which to us, as a fast-growth and profitable-growth business, is really important.”
Ettitude recently ran a geo experiment with Measured that focused specifically on its Facebook prospecting campaigns. The study helped them determine that these campaigns were around 50% more incremental than its executives originally thought.
“Given the reporting adjustments that have rolled out on Facebook after iOS 14, we were frankly just nervous about continuing our Facebook spend at the same levels,” adds Rosenblum. “But this new measurement coming back from Measured helped us feel more comfortable that we could be continuing to invest in Facebook prospecting, as well as just upper-funnel tactics overall.”
Operating on a CPO (cost-per-order) or CPA (cost-per-acquisition) basis, Ettitude is now able to understand which product orders are incremental to a specific channel.
Focused on driving more upper-funnel brand awareness, Ettitude’s team now feels more confident with adding new channels to their media mix. Over the past year, Ettitude has advertised on TikTok, Pinterest, and podcasts.
“You’ll see the word incrementality a lot more now than you did a year ago,” says Jen Cadmus, Measured’s director of marketing communications.
Due to an increase in restrictions to third-party data, ID resolution, and user-level tracking, this is only the beginning.
Experimentation using first-party CRM data for retention, or geo-matched market testing for prospecting, will likely become the most accurate and effective options for measurement across platforms.