Here's what we are going to cover in this thread: > numbers > ad account structure > mistakes > opportunities for improvement > results

Basic numbers for the last 30d CPM: $11.2 CPC: $2.23 CTR: 0.50% Immediately you can spot that CPC is way too high for this CPM. What does this tell us? → The problem is creative.

The next thing I noticed is they are running on multiple attributions without proper structure Let’s compare them For context, the target ROAS is 0.85 in NB Right away we can spot that 7dc has the best ROAS while not much is spent on it The next step >push more budget on 7d

Interestingly 7dc1dv has the highest average watch time and thumb-stop ratio but the lowest AOV and CR After getting back into the ad account I’m noticing there's a VV campaign running, which is causing such a discrepancy in numbers The next step is to turn off the VV campaign

Let's dive into the campaigns: There is a dedicated Testing campaign which is ✅ All testing is done with a Broad audience - again ✅ Buyers are excluded ✅ Good naming convention ✅ Let’s see what can be improved

1. Few ad sets in a testing campaign were strong performers but turned off ❌ WHY? There’s no point in turning off ANYTHING as long as it’s working, no matter if it’s just a testing campaign. You can still transfer the ad into a Scale campaign and keep it live in testing.

2. Optimization on the ad level could be better ❌ All the bad ad sets were turned off, BUT On the ad level, there were clearly some underperformers that should have been turned off. Optimization should be done both on the ad set and ad level

Maximizing the efficiency is not just scaling, it’s also cutting down on the losers, especially if some testing ads have 2x your average CPC. There’s a small chance they would be profitable. Turn them off earlier.

3. NO thumbnail testing ❌ I already talked about the importance of a thumbnail. https://x.com/IstvanicMarin/st... Here there was no thumbnail testing at all. Even worse, all the thumbnails looked automatically chosen by FB, not even manually picked.

The VV campaign had some even issues as well It was a CBO with 2 ad sets running One on auto-placement, another on IG reels/stories, which got 90% of the spend Why? Because it’s cheaper BUT it had a lower avg watch time which makes it less effective considering the objective

ABO TOF Scaling campaign with manual bids Basically, taking your best audience & ads combo and trying to squeeze them more I like the approach here, but IMO the range of bids was super big There were both CC and BC from $125-$227 I guess I needed more details about the setup

CBO TOF Scaling campaign with manual bids Okay, so here I saw a big 🚩 Ad sets were again with a big range of CC and in CBO FB spent most of the budget on the ad set with the highest CC While NOT delivering good results Tip: Test different CC in an ABO, you have more control

MOF CBO Strange thing noticed here There was a “Broad” ad set here. It doesn’t belong in MOF All MOF audiences were stacked - engagers, visitors and video watchers BUT none of the audience was excluded from prospecting Again something that could have been handled differently

The biggest mistake in the whole ad account was > Not using post ID Basically, there were the same ads, but with different post IDs, so each variation was gathering its own social proof. No wonder CPC was so high, as social proof was spread among different identical ads

So, what was the summary of the audit: 1/3 > Optimization on the ad level based on NB okay Slight improvement possible > Optimization on the ad level Should be better > Prospecting Only a Broad audience was used, and only buyers excluded

Summary 2/3 > Retargeting Pretty bad, in some campaigns “broad” running under retargeting All audiences stacked together, no exclusions in prospecting DPA active with bad results No objection-handling copies

Summary 3/3 > Bids Makes no sense to use different bids in CBO with different ads → Either standardize the bids with different ads or use ABO or standardize ads but test bids

Additional improvements 1/2 → No interest or LAL tested at all Could be an easy win → No catalog for Broad audiences (DABA) Could be great considering a lot of SKUs

Additional improvements 2/2 → no testing with dynamic creative at all Great for similar variations which you have, extract winners and run as a regular ad → standardize attribution setting → separate campaign for loyalty

Results 1/2 We ended up working with the client, fixed all the leaks in their ad account and implemented our own methods. As expected, using post ID was a massive change as it dropped CPC from $2.2 to around $1.7, but there’s still room for improvement.

Results 2/2 Interest targeting managed us to squeeze a bit extra ROAS, although was not scalable as Broad DABA had ups and downs We got rid of 7dc1dv and run everything on 1d or 7d It looks worse on FB, but it looks way better on NB, which is only what the client cares about

If you are a 7 fig ecom brand and you want us to audit your ad account in the same way Then, shoot us a DM or apply for our FREE audit here: https://inspirebrandsgroup.typ...

@IstvanicMarin Amazing thread G

@wizofecom appreciate it legend

@IstvanicMarin This is massive, thanks for giving away all your process dude. A tone of value here

@HrvojeKaralic glad I could help bro appreciate the kind words

@IstvanicMarin Marin's a philanthropist.

@IstvanicMarin @threadreaderapp unroll

@IstvanicMarin @SaveToNotion #Thread

@IstvanicMarin @SaveToBookmarks Great thread!

@IstvanicMarin Top tier thread🌟🏆

@IstvanicMarin @threadreaderapp unroll @readwise save thread @memdotai mem it @rattibha @SaveToNotion #thread

@IstvanicMarin What is your thoughts on FB pushing shop on platform instead of website? Do you think website shopping will die?