Stop me if you’ve heard this before:
“Well, we spend a lot of money on Facebook ads, but we’re not 100% sure where it goes or if it works.”
A year ago, this was a niche problem for high-performance marketing agencies. For sites like Indiegogo and Kickstarter, eking out even a single-digit percentage of advantage could be the difference between being a top agency and becoming a footnote in a fast-moving industry.
Pictured: a global crowdfunding campaign on Aphrodite, which helps users select the right market for advertising.
In truth, ROAS (Return on Ad Spend) has always been hard to measure. Often, it’s found by digging deep into Facebook and Google’s interfaces, finding the numbers manually, and then transferring them to Excel to get a clearer picture. With iOS 14 and Facebook’s new “statistical modeling” feature, however, making sense of advertising data just got that much harder.
But there’s an answer. Businesses spend $50B annually on Facebook ads and $150B annually on Google Ads, and it was a matter of time before someone decided to measure ROAS precisely and make those numbers accessible. Enter Aphrodite, a San Francisco-based startup.
Pictured: an eCommerce advertising-driven revenue as part of organic revenue
In fall 2018, a team of Silicon Valley software and hardware enthusiasts got into crowdfunding, a fast-paced version of e-commerce in which every campaign is “make-or-break.” What leads to the success of a crowdfunding campaign is often a carefully orchestrated and diligently measured advertising campaign behind the scenes.
Software folks stuck in Excel: inspiring Aphrodite’s solution
In that industry, rapid and specific deployment of advertising dollars is key. Because of the speed of crowdfunding — spending $100-500,000 advertising dollars in 30 days for a chance to reach a funding goal — Aphrodite built its own internal tools to track the ROI (Return on Investment) in a way that instantly plugs into the data.
Pictured: an international sneaker platform optimizing ROI around the world
When Aphrodite was asked by companies to expand their analytics to incorporate e-commerce, it was a natural next step. The team knew this was a problem from reading articles about startups going out of business by mismanaging their ad spend. Still, e-commerce was quite different.
Image courtesy of BBC.com
“E-commerce moves about 10x slower than crowdfunding, but the scale is about 10x larger, so it works a similar load on the server,” Aphrodite CEO Daniel Dietzel said. “You’re spending $100,000 a month, but instead of in a 30-day window, you have to do that every 30 days.”
Because their platform was built for speed, the Aphrodite team was able to provide key solutions to this spring’s iOS 14 update. While Apple allowed users to decide whether they want to be tracked by apps, igniting a feud with Facebook earlier this year, Aphrodite relies on old-school UTM tags for tracking eCommerce on websites, where the majority of purchases happen.
Apple can’t change URLs, which are the basis of UTM tags, meaning that for Aphrodite, the analytics business continued as usual. While this is a low-tech workaround, another big shift meant there needed to be a high-tech workaround: Facebook’s new method of “statistical modeling.”
Facebook’s new “statistical modeling”: what is it?
Apple tore a 20-30% hole in the internet’s analytics: both iPhones and Safari browsers (the default Mac browser) block all trackers. Consequently, marketers are blindly pushing out Facebook ads without knowing what’s working, and Facebook also faces its own problem: less revenue is attributed to its platform.
Facebook introduced “statistical modeling” to round up the numbers and make up for an estimated 20-30% gap. Aphrodite went through a number of accounts with their clients and found that statistical modeling was inflating values to theoretically impossible heights.
Facebook revenue (blue) as part of overall revenue (orange), before and after “statistical modeling”
In most cases, ad-driven revenue will account for 10-50% of overall revenue. Before “statistical modeling,” this client’s results (seen above) were pretty clear: only 30% of revenue was driven by Facebook Ads.
“Data Scientists hate him! Learn how one platform became 150% of your revenue overnight using one weird trick.”
One morning, out of the blue, Facebook reported that it was accounting for up to 150% of revenue. That’s when the client and the Aphrodite team started digging together, discovering Facebook’s new system of inflating numbers. As for individual ads, statistical modeling resulted in ridiculous ROAS numbers (seen below). Achieving 20x ROAS is impressive, but 180x is highly suspect.
The biggest problem is not that Facebook is doing this. It’s that Facebook is A/B testing the new model on unsuspecting businesses. Who gets the new model is a black box.
For transparency purposes, the Aphrodite team decided to make its servers sweat and pull both models for businesses and let users switch back and forth. An “iOS 14” toggle on Aphrodite allows users to see data both before and after Facebook’s unified attribution.
Is Facebook fudging your numbers? The only way to find out is to sign up for a 7-day free trial of Aphrodite.
At $300/mo, Aphrodite is on the pricier end of the spectrum. But we’ve included them as a recommended partner because they solve problems that no one else wants to solve. And what’s more costly: paying for good analytics, or spending thousands on a platform giving you incorrect numbers?