What's working in Facebook Ads: August 2021

It’s rough out there in the Facebook ads land.

Not all accounts are bad, but we continue to see instability, inconsistency, rollercoaster ups and downs, and a bigger range of what is and isn’t working across different accounts.

With that being said, here’s a look at what we’re seeing trend upwards for the month across accounts. Should you test any of these please let us know, so we can add even more data to our findings.

Our whole purpose here is to share ideas. If you like this, join us here.

What’s working

• Try new conversion optimizations. If hot retargeting (like 7-10 day VC/ATC/IC) is getting conversions that are too expensive, or CPMs are too high, try optimizing for View Content over purchase. Sometimes Facebook gets a bit stuck on “who” it thinks is likely to purchase so moving up the funnel can be a great way to help Facebook look at these warmer audiences who likely already have purchase intent.

• Grouped warm retargeting (middle funnel) audiences. While our analytical brains love to know exactly which audience(s) is converting, grouping all middle funnel audiences (think WCA, video views, engagers, email engagers/openers, etc.) together has been quite successful as a whole.

  • We’re currently segmenting by date range, so actively we have both a 30 day warm retargeting group and a 31-90 day warm retargeting group running.

• ABO. We’re still #TeamCBO, but lately, CBO has not been as effective in spending across ad sets to find quality traffic that is going to convert. So at this time we’re putting more control back in the ad buyer’s hands and dedicating smaller consistent budgets to specific ad sets/audiences.

• CPA. We continue to look at our CPA as a guiding metric on the platform. Yes, we know that conversion data is not as accurate and thus “ghost” conversions are making it past Facebook’s Attribution abilities and thus causing CPA to look much higher. That said, in working with our clients, we are testing new CPA goals that are continuing to help us see MER on the back side level up.

• Continue to test new things. What was working in August of last year or even March of 2021 just isn’t working right now. That’s way we’ve continued to test something new every few days. We’re taking a closer look at our existing audiences to find greater similarities or even greater differences for that matter and how we may be able to combine them in new & unique ways.

  • For example, we’ve been looking at the top converting videos over the last 45 days and then creating lookalikes from these specific video viewers.

  • Or lookalikes of first time purchasers from this year

    • We’ve even been testing how to retarget this audience to encourage a second purchase if they have not all ready done so.

  • We’re also testing a bit crazy stacks like a 3%ll or 5%ll engagers audience combined with a 1% Add To Cart or Purchase LL (maybe this is not so off the wall, but you get where we’re going with this).

• Looking at Google Analytics. Have you seen this blog on how to get started with UTMs, and what to look at once you’re diving into the data on the backend? You can also look at time on site data from UTMs in GA to determine which audiences/ads are most qualified for driving length of time on site, which oftentimes can allude to high-quality audiences who are interested in converting

• Less exclusions. Yes, you read that right. While we are/used to be sticklers for strong exclusions, we’ve found using less prospecting exclusions can be helpful. For example, instead of excluding the whole list of custom audiences, maybe just exclude 180 day purchasers, or even 30 day purchasers, or no exclusions at all and see where it takes you. At the end of the day, a purchase is a purchase right now.

• Creating custom metrics that work for your account. With the loss of unique ATCs with ATT/iOS14, it’s hard to rely on uATC to PUR %, since ATC fires with each ATC and IC/PUR only fires once per checkout. With one client, we are testing out mATC (modeled ATC), where we determined through Shop data that the average number of products per cart was 10.83 over the last 90 days. We then created mATC (total ATCs / 10.83) to create mATC, then created LPV to mATC % and mATC to PUR % to better understand drop-off rate and where customers are exiting the conversion process. So far, it’s looking almost identical to what used to be unique ATCs.

Don't want to miss out on this 🔥 content for another minute?

What’s not working

• Optimizing same-day, or looking at anything less than a 3-4 day window. We currently are optimizing accounts every three days (checking every day), but not making decisions or changes too quickly.

• Expecting in-platform ROAS to be above a 2-3X. We’ve been considering success above a 1X (breakeven) in cold prospecting, and a 2.5-3X in retargeting over a 7/14/30-day window. We’re also looking at MER and Google Analytics to determine how to optimize. Have you read this blog on why MER is more important than ever right now?

• Too many creatives/ads per ad set. Similar to our reasoning for sticking with ABO, ad sets have not been able to spread out spend across ads very well. We’ve been putting no more than 2-3 (4 MAX) ads per ad set to determine success. If you want to do creative testing right now, manually turn ads on/off to make sure they get spend and gather equal data appropriately.