White Facebook Logo
Log in with Facebook
White Facebook Logo
Sign up with Facebook
Modal close button
Success Guide

How to use your 1st-party pixel data to improve your account’s ROAS on Meta

What if your first-party data not does match the ROAS reported by Meta?

More importantly, how do you decide whether you should scale or kill a campaign when you have two different sets of data?

In this video, Karan Jassar, Socioh’s Founder and CEO goes over how to get the most value from your attribution tool:

Video link - https://www.youtube.com/embed/Fo40mGYRjtw

Transcript: What’s up guys? A very common question that we get at Socioh is what to do when my in-platform data doesn’t line up with my first-party pixel data. I wanna walk you through a real example and see how to go about making that decision. So let’s get started.

You can see in this example, that Meta’s reporting a 2.55X ROAS, and Socioh is only attributing 0.56X ROAS.

So that’s almost five times of what your first-party pixel is telling you. What do you do in this case?
The very first thing I check for in such scenarios is whether or not Meta overreporting on views-only conversions. Now you can see that in this table, in Socioh (we pulled this data from Meta).

Views-only conversions are conversions where Meta showed your ad to someone, but they never really clicked on it, and that conversion would probably take place anyway. Sometimes Meta will over-report on this stat and that is the cause of the discrepancy In this case, however, that is not the case. So let’s see what else we can look at. Next, I’ll look at Socioh’s reported revenue on just a one-day click attribution window. And this is reported on an accrual basis.

Now, if you don’t know what that means, what accrual reporting is and why it’s helpful, I highly recommend you watch this webinar that I did with Andrew Foxwell. It’s available on our blog and it explains these concepts in very, very simple terms. Back to our example. On a one-day click basis, Socioh’s reporting 0.58X ROAS, and just eight orders.

What’s worse is that our machine learning is predicting only a 9% lift in revenue 28 days out. What this means is that for this campaign, there is some delayed attribution, and some delayed convergence that may happen, and that number is just 9%, which is way below average for this account. Now typically, for this account, we’ve noticed that this number varies anywhere from 25% to 44-45%. So even on that basis, this is really, really low. That’s a negative.

The next stat I look for is your new-customer acquisition cost (nCAC) and your new-customer AOV (nAOV). And you can see in this case, we paid $210 to acquire a new customer and they’ve all on average spent only $87 with us. So this (campaign) is clearly not profitable for acquiring new customers.

And also look at your blended CAC versus your AOV. And even from that perspective, these numbers are almost equal, so we’re losing money on every transaction for this campaign. The next thing I look out for is new versus repeat customer split. Now, in this case, you see it’s about 55% new customers, and that is actually pretty good for a retargeting campaign.

Typically, for a top-of-the-funnel campaign, we want this number to be anywhere above 60-70% — actually above 70 at the top-of-the-funnel would be fantastic. Now, whereas this stat is positive for a retargeting campaign, it’s just not good enough to justify the CAC numbers that we’re seeing.

The last thing I typically pay attention to is the click-through rate (CTR) and the CPM (cost per thousand impressions) for this campaign.

Now, both these numbers are good for this campaign. What I mean is that the click-through rate (CTR) is above the account average. And the CPM is below account average and that’s what you want. And whereas that’s positive, again, we’re still losing money on every transaction. And finally, the decision just comes down to that if you are losing money on every transaction and new customer acquisition is not at all profitable, we should shut down this campaign.

And that is the right answer here.

If you have similar questions about your store or any attribution-related questions, please send them my way. I’ll be happy to make a loom for you. Have a good day.