False hopes
“We’re excited to report that your campaigns are producing awesome results - with a ROAS of 8!”
An 8x return on advertising spend is salivating for advertisers and business owners alike. When we reported this to one of our clients, we were pretty damn proud of ourselves, too. One night, as I was scrolling through Reddit, however, I stumbled across this post about ROAS numbers being artificially inflated by Meta: https://www.reddit.com/r/PPC/comments/vubqq9/facebook_ads_overreporting/
The top comment was even more damning: “This is completely normal.”
As advertisers, we take the word of billion-dollar companies as gospel. I mean, why not, right? They have thousands of developers building and maintaining these platforms worldwide - why wouldn’t the stats be accurate? How do I validate our data??
Enter, WhatConverts
At the beginning of the 2nd quarter this year, we began to pilot a new conversion and lead tracking tool, called WhatConverts. The first batch of clients we implemented WhatConverts on were a select few eCommerce businesses. Since eCommerce sales signify binary success (either there’s a sale or no sale), they were the best choice for us to validate our new tool.
Originally, we implemented WhatConverts to help us fill in the entire customer journey and understand where our clients’ customers were engaging with their brands before their purchase. Since GA4, Google Ads, and the other advertising platforms primarily use last-click attribution as default, we wanted to get a full picture of the buying experience.
During this journey, we also discovered a huge discrepancy between what Meta Ads was reporting in terms of sales and what was actually happening.
Below is a screenshot of what Meta Ads was showing us for the same client, referenced above. Our Meta Ads attribution setting was currently set to 7-day click and 1-day view, meaning sales are attributed to the ad if the user clicks on the ad and purchases within 7 days, or if they simply see the ad and purchase within 1 day.
On the other hand, here’s what WhatConverts reported in terms of last-click attribution. Meta Ads was nowhere to be found!
When we swapped our attribution method to first-click, Meta Ads did show up on the report, but with only ~$500 in attributed revenue:
Troubleshooting a five-figure reporting discrepancy
We’re used to seeing 10% to 20% reporting discrepancies due to technology and privacy limitations, but overreporting almost $15,000 seemed outrageous. Turns out, it’s a very common issue with Meta Ads. A Google search for the term “meta ads overreporting” reveals thousands of other people who are experiencing the same issue.
Apparently, Meta Ads is extra greedy when it comes to attributing performance to their ads. Even though we have our ad view attribution window set to 1 day, Meta is still attributing purchases to a much longer period than that.
As a part of our troubleshooting process, our development team double-checked our pixel implementation and could not find any issues. We thought it might have been due to the recent upgrade to the Facebook CAPI (server-side tracking) for this particular account, but our testing concluded that there were no issues with the pixel implementation.
Changing your perspective on data analysis
This story doesn’t end with a clear resolution, unfortunately. After comparing Meta Ads performance via the Ads Manager versus looking at raw data in WhatConverts as well as GA4, we’ve consistently found the same overreporting behavior. The moral of the story here is that you should always verify the data from the advertising platforms with tools like WhatConverts.
It’s a well-known fact that there is no perfect solution for tracking conversions and purchases cross-channel, although server-side tracking can help improve the accuracy of your reporting. Cross-channel tools like TripleWhale can also add more visibility into your campaign performance, especially if you’re advertising in more than one channel.
Instead of looking at your marketing data in absolute terms, change your perspective by viewing it as a trend over time. Use your marketing data as proxies for performance, and instead, prioritize revenue growth, matching your marketing spend to your growth goals, and your margins as the true indicators of performance.
