How I scaled Facebook Ads profit by 128%
I had a Facebook Ads problem. Our campaign performance had plateaued. In-platform ROAS was declining as was ad efficiency and contribution margin over the course of several months. Seasonality and historical trends would have suggested the opposite. But here we were. Initially I tried to solve the problem through technical optimizations: decreasing budgets where ROAS was particularly troubling and implementing stricter bidding adjustments.
Nothing.
Next, we started to drill down into the data and go through the account, campaigns, ad groups, and ads with a fine-tooth comb. We sliced and diced the data every which way, trying to figure out why our ROAS and profitability were declining. We looked at audience performance, placements, asset types, device performance, countries, day of week, type of campaign, bidding strategy, sale vs. no-sale periods, change logs. The whole nine.
What did we discover?
It all comes down to creative. Specifically, how long a creative is remains in circulation. We found that ad ROAS declines significantly the longer the asset is live. Non-dynamic creatives stop being profitable (i.e. their ROAS dips below 3x) after 9 days. However, our current in-market average was 19 days. This means that we were not profitable for the majority of the time these ads were live and we essentially were hemorrhaging money by clinging to outdated and underperforming creatives.
Facebook weekly ROAS for non-dynamic ads
However, if we were to maintain a continual refresh every 9 days we hypothesized we would remain profitable. So that's what we set out to do: continually refresh our ads every 9 days to prevent a drop in ROAS and create a short-cycle feedback loop between the marketing and design teams to accelerate learnings and data-driven decision making.
Specifically, we embraced betatyping.
What is betatyping? It's a systematic methodology that involves crafting multiple variations of each ad creative, testing them simultaneously, measuring results across creative asset KPIs, and iterating on the winners. It allowed us to quickly test and gather insights into which creatives resonated with our target audience, which ones fell flat, and how we could ideate off of the winning creatives to refine our target messaging. Essentially, our betatyping process looked like this:
ideate 3 creative concepts to express one element of our unique value proposition
generate 3 assets for each concept (total of 9 assets)
create a single Facebook campaign and upload all 9 creative assets (I won't go into campaign specifics like budget, bidding, targeting, etc.)
run the campaign for 2 weeks and make sure it's out of the learning phase (~50 conversions)
evaluate the 9 assets using with the following KPIs:
CPM
CTR
CVR
Quality ranking
Engagement ranking
Conversion ranking
analyze trends and pick the top 3 asset performers by asking:
What's performing? Why?
Are specific products performing better than others?
Creative types?
Angles?
Offers?
ideate on those 3 assets and create 2-3 assets per winning asset
upload the new assets into the same Facebook campaign and repeat steps 4-6 for another cycle
Here’s what betatyping got us in the following months:
ROAS jumped from below 3x to over 4x
Ad spend increased by 31% (yes, we bucked the correlation between ad spend and profitability)
Contribution margin soared by 128%
Contribution margin rate ascended from 6% to 11%
Facebook MoM performance
These results underscore the power of betatyping as a data-driven approach to optimizing Facebook Ads campaigns. By continuously experimenting with creative variations in a quick feedback loop, we unlocked Facebook Ads. So remember...
Embrace experimentation: Don't shy away from testing new ideas and approaches.
Create a data-driven feedback loop: Utilize data to guide your creative decisions and optimize your campaigns.
Ruthlessly replace bad creatives: Fresh, high-impact creatives are the driving force behind success.