This year, we've almost tripled our ad spend on Meta while reducing our CPA by over 15%. Here's exactly how we did it that you can steal too...
But first, a quick disclaimer. Improved performance is not going to come from 1 thing, despite what an agency or SAAS might sell you. It's the sum of all the little, 1% improvements along the way with a few 10% improvements. Here were some of ours:
1. Consolidated Account Structure
Our account was never crazy complex, but we used to do LLAs, Interests, and Retargeting. Over time, we bundled it down and are left with two primary campaigns: Creative Testing and Main.
Our account was never crazy complex, but we used to do LLAs, Interests, and Retargeting. Over time, we bundled it down and are left with two primary campaigns: Creative Testing and Main.
2. Obsession On Creative Testing
We used to spend 10% of budget on testing. In the last 90 days, we have spent exactly 47.9% of budget on our "Creative Testing" campaign. I used to be worried that even 10% would be too high because I didn't want 10% of our budget to be "wasted".
We used to spend 10% of budget on testing. In the last 90 days, we have spent exactly 47.9% of budget on our "Creative Testing" campaign. I used to be worried that even 10% would be too high because I didn't want 10% of our budget to be "wasted".
Man I was dumb. In fact, despite it being less than 50% of our budget, Creative Testing campaign has had over 54% of purchases in the account. Why is that?
I no longer see it black and white as "testing vs scaling". I think there are 4 categories of ads:
1. Green - "Winners" - Get moved to main and get a lot of budget
2. Yellow - At or below CPA targets with reasonable amount of spend, but not swallowing all the budget.
1. Green - "Winners" - Get moved to main and get a lot of budget
2. Yellow - At or below CPA targets with reasonable amount of spend, but not swallowing all the budget.
We just keep these ads running, and will often iterate on them to try to get them to spend more.
3. Yellow - Above CPA targets, but no far off. These will get paused and iterated on, then relaunched.
4. Red - Bad ads. Usually won't spend, and will get paused.
3. Yellow - Above CPA targets, but no far off. These will get paused and iterated on, then relaunched.
4. Red - Bad ads. Usually won't spend, and will get paused.
Note this is all in one CBO, with ad sets separated out by angles. This automatically allocates spend based on expected performance, so I have very little risk. This part is crucial, because spend is the main metric to understand if an ad is good or not.
I believe this has reduced our CPA by allowing us to squeeze out more incremental conversions from each ad set, which is usually a different angle of product. I used to think these ads weren't "winners" because they weren't good enough to scale, but that extra $1k-$2k per day...
...has allowed us to eek out extra conversions efficiently, while also finding "winners" to let ride.
3. Landing Page Testing
This has been huge for us. We used to send most traffic to PDPs, now we send it all to LPs. We use a variety of different and are constantly testing.
3. Landing Page Testing
This has been huge for us. We used to send most traffic to PDPs, now we send it all to LPs. We use a variety of different and are constantly testing.
The key is to look for 1% improvements with iterative testing, while getting the occasional 10% improvement from different styles. Up until now, we had no good way to do this because we required code every time we needed a change. That is until we found @replohq. Replo has....
...allowed us to create beautiful, on brand, and fast landing pages without requiring and code. This has been a game changer to allow our marketing team to test so much faster and cheaper. Try Replo here: rb.gy/lya1fk #affiliate
4. Using Our Quiz on 90% Of Traffic
90% of our traffic goes through a quiz funnel. In addition to being our best funnel from a RPS perspective, this has driven MASSIVE list growth. We're well over 1M subs and growing fast by the minute. We use @OctaneAI for our quizzes.
90% of our traffic goes through a quiz funnel. In addition to being our best funnel from a RPS perspective, this has driven MASSIVE list growth. We're well over 1M subs and growing fast by the minute. We use @OctaneAI for our quizzes.
5. Getting Accurate Data
Much is written about attribution and again I will never say 1 thing solves everything. But I would be scared sh!*less without @northbeam_io. The data is scary accurate, and we really feel isolated from any data losses. I am a proud user and advisor.
Much is written about attribution and again I will never say 1 thing solves everything. But I would be scared sh!*less without @northbeam_io. The data is scary accurate, and we really feel isolated from any data losses. I am a proud user and advisor.
It's been over a year of iOS and the game has changed, but it's still very winnable. These are some great ideas you can test to help get your performance back on track. It's just what works for us, but hopefully this gives you a lot to think about and test.
I hope you've found this thread helpful.
Like/Retweet the first tweet below to spread the word.
Follow me @codyplof for more.
Like/Retweet the first tweet below to spread the word.
Follow me @codyplof for more.
https://twitter.com/codyplof/status/1591441388289150976
If you liked this thread, you will love my free weekly newsletter where I share EVERYTHING I know about ads, landing pages, and growth.
Join over 6,000 other top Founders, CEOs, and marketers here: sparklp.co/p/08386007ee
Join over 6,000 other top Founders, CEOs, and marketers here: sparklp.co/p/08386007ee
• • •
Missing some Tweet in this thread? You can try to
force a refresh
