#paid-media
Thread
Hey guys,
We are an Amazon native brand, but have been scaling up Shopify since January with Meta and Google Ads w/ ~90% of spend going to Meta.
Our here product that we're pushing is our 25 x pack of dough balls. We've sold ~4,000 boxes on Amazon so spillover from that platform is concerning me some.
If you look at our our total MER for Shopify, it's always hovering around 3x, but Meta is consistently reporting around 2x...
How do you guys figure out what's moving the needle? Is there a way to dissect the traffic sources to understand if it's a matter of Meta underreporting, if it's Amazon spillover, etc...?
I don't want to continue scaling Meta ads on the assumption that that's yielding us a 3x ROAS when in actual fact, it's Amazon spillover that's inflating our MER and our Meta ads actually are running at 2x ROAS.
How do I get to the bottom of this?
Through UTM parameters
Has the CAPI and server-side tracking setup done on Meta?
Also, an extreme step but turning off ads on Meta for 7 days and see what the spillover from Amazon is and then considering the average to understand how much Meta contributes
@John Dyall - Are you using any attribution system? GA4, Triple Whale, Northbeam, or anything of the sort?
You need to measure Meta off of whats its driving to the site. BUT its important to be mindful that customers will buy off of Amazon if they are aware its available there.
Kanish is also correct. One of the best ways you can do this is through either a GEO based hold out lift test or a black out test.
CAPI was set up, but it wasn't on the maximum setting... I believe Enhanced is the max? In any case, I swapped that over so hoping that helps some. I've also installed a post-purchase survey to ask customers how they heard about us. No one has said Amazon yet, so problem might be solved...
@Christopher Parrett – I am on the free version of Triple Whale. No GA4 or Northbeam. Do these tools give the ad platforms better data so that we can serve better ads, get better ROAS, etc... or are they purely for getting a better sense on attribution?
You can look at click only attribution and do a rudimentary correlation analysis to see if sales on either platform move when spend on either platform moves.
From there to get a really precise answer you need to run incrementality tests, probably Geo Holdout style to see the causal net effect on each channel.
Attribution modeling only goes so far.