If Your Meta Ad Performance Tanked in Q1 2024, READ THIS! (Bonus Content!)

I've identified what I think is a possible source of a perceived DROP in platform "performance" over the past 1.5 months.

If Your Meta Ad Performance Tanked in Q1 2024, READ THIS!

I've identified what I think is a possible source of a perceived DROP in platform "performance" over the past 1.5 months.

The situation:

Meta ads has a clear issue with tracking/attribution over the past 1.5 months with iPhone devices.

Before rushing to judgment and acting, you need to determine if this attribution issue is immediately impacting your business.

💥 It seems that Meta is NOT processing as many iOS purchase events as it should be, even though the events are occurring & being sent back to Meta. 💥

PROCESSING = Meta using the event you sent it.

Season 2 Omg GIF by The Office

Gif by theoffice on Giphy

Typically Meta processes ~80% of purchases that you send, but somehow this got flipped and we've seen as little as 20% events being processed.


▶️ 100 purchases yesterday got sent to Meta (as evidenced in Events Manager); Meta normally "accepts" 80 of these events.

▶️ But NOW, Meta is only "accepting" 20 of these 100 purchase events on iOS devices.

You can track this through custom conversion setup (Meta will show you how many usable events you had).

The possible outcomes:

1. Brand performance is not impacted (despite Meta saying that performance is worse). Meta performance is actually not affected despite the algorithm optimizing off inaccurate data. The instance that I describe below falls in this bucket.

2. Brand performance is impacted. Theoretically, since Meta is processing less conversions, it is feeding Meta's ML false negatives (saying ppl didn't purchase, when in reality they did- this is exactly what happened following iOS14.5). This can confuse the algo bc it's optimizing on false data. This could in turn increase your CPMs as you'd be bidding for a shrinking audience of potential conversions. Brands that are in the "first outcome" scenario above may very well start experiencing this "second outcome" scenario.

An example of a brand that's impacted👇

Looking at Cost Per Purchase (iPhone vs Android), each device had roughly the same CPP until on or around February 18. In the image, it appears this may have actually started in mid January.

During this timeframe, we had an inflection point where CPP on iPhone skyrocketed much higher than Android (as attributed by Meta). This is clearly visible.

So, are more purchases actually happening on site but not getting attributed in Meta b/c of the drop in processed purchase events?

This presents the question: is this an ATTRIBUTION OR PERFORMANCE ISSUE?

Let's take a look. This is where first party DATA becomes EVERYTHING! Let's see if blended CPA and new customer acquisition were impacted at the store level.

So from Feb 18 - March 28, blended CPA for this brand actually DECREASED 5% as compared to the previous 40 days. Store performance is looking IMPROVED if anything!

Now lets look at new customer orders. If CPA "skyrocketed" on Meta (b/c of iPhone conversion issues), we would expect new customer acquisition to suffer, right? Let's find out.

First, lets look at ad spend (which obviously is a driver of new customer acquisition).

Between Feb 18 - March 28, ad spend actually dropped by 16%.

💥 If there was a customer acquisition issue present, we'd expect new customer orders to drop by at least 16%💥

New customer orders only decreased by 11%... this means that customer acquisition actually got EVEN BETTER during this perceived time of doom.

For this reason, we can conclude that in this instance, this is an attribution issue on Meta's platform...this is NOT a Meta performance issue impacting a brand's performance.

⚡️ This story may not be true for all brands ⚡️

You need to go through the same exercise with your own data to uncover the cause and ramifications of this known "issue."


Who knows...

Either a Meta glitch OR Meta testing a doomsday scenario in a world where 0 iOS events would be processed if Apple gave an ultimatum to shutdown CAPI?

In conclusion:

YOU as the brand owner or marketer need to decide if you fall in bucket 1 or bucket 2.

You need your own first party data to determine this before making any decision. So many brands are quick to pull budget without even uncovering the facts of the situation. The last thing you want to do is pull budget from a system that is ACTUALLY working (even if the platform is telling you a different story). Then your business WOULD be negatively impacted.

Make decisions on business impact and the real numbers. If you rely on deterministic attribution 100% of the time without understanding the big picture, you'll be in trouble.

BAD DATA is worse than NO DATA.

If you use bad data to inform your decision-making, you're in a much worse place.

Big shoutout to Mandar Shinde for sharing brainpower with me on this topic.

If you need help interpreting or building your own first party data set, you know how to reach me.

I welcome all feedback. Good, bad, everything in between.

Hit reply, and let’s hear it! 👂

📧 Share your thoughts or what you want me to cover next!

Yours truly,

Jonathan Snow

If you’d like to see more content of mine, be sure to give me a follow: 

📆 Schedule 1-on-1 time w/ me here.

Join the conversation

or to participate.