This is what happens to your ROAS when you’re trying to scale…
At the very beginning, results are slow. You’re simply testing out the waters and figuring out how everything works.
After some experimentation and little trial and error, you see some level of success.
You spend $10/day and make $20/day. Then they scale up to $20/day, $30day, and $50/day. And your ads continue to generate a decent-to-good return on ad spend.
Now you think “It’s easy! I got it!”
You're excited to see what happens if you will ramp up the budget a bit more.
When you do it this time, the opposite happens.
Only your ad spend goes up. Sales stay the same. CPA goes up.
Then you scale back down while trying to figure out what went wrong.
After another trial and error, you feel stuck not knowing how to fix the problem.
Typical things I hear store owners say: “our budget is unlimited if we hit our ROAS/CPA goals” or “we will be able to increase our spend if CPA stays low, etc.”
I get it.
Scaling isn’t as difficult as many people think it is...
...but it also isn't as straightforward as simply doubling down on your ad budget all the time.
So here are some root causes of the problem and solutions to how you can fix them.
1. "MY DEAR, YOUR STRUCTURE IS …”
Audit after audit we usually see ad accounts being structured either:
a- as if someone did them on their phone while learning how to ice-skate at the same time
b- overly complex with multiple campaigns and ad sets for every objective and interest possible.
Scaling horizontally doesn’t equal duplicating your ad sets or expanding your interests.
Most of you have too many ad sets which are difficult to manage.
Account simplification is your first step to making things right.
At the simplest level, we build TOF (cold) MOF (warm) BOF (hot) funnel set up as your step 1.
TOF (prospecting) > 5-6 ads sets > 3-4 ads per ad set
MOF (retargeting) > 5-6 ads sets > 3-4 ads per ad set
BOF (retargeting) > 5-6 ads sets > 3-4 ads per ad set
In simple terms, you want to build a funnel within your ad account.
A good start is to let your team learn about Facebook Power 5. It will significantly help you reduce your advertising efforts whilst maximizing its output.
Big brands like HelloFresh and Smile Direct Club have shown better reach and lower CPA after taking advantage of this framework.
Also, don’t forget to about having a Clear Testing Phase -> Prospecting Phase -> Scaling Phase
Basically, you should not be taking a campaign/ad set to the prospecting or scaling phase unless it meets the KPIs of testing (avg CPA, ATC, etc…).
The goal here is to figure out what ads and messages work best with which segments of your customer base.
2. NOT TESTING CREATIVES FAST ENOUGH
Creatives are your greatest lever.
A common problem is not having a system that allows you to pump up a set of free creatives on a regular basis (every 2 weeks or every month if you’re a smaller brand).
Please, please, please don’t feel discouraged because the creative that you’ve put in so much work ended up being a loser.
Not testing enough = ad fatigue = ROAS dip.
Number two is falling in love with your own creatives.
Trust Facebook’s AI, let it test as many creatives as possible, you won’t overthrow it anyway.
You may debate how much you want but most of the time it’s as simple as that.
Now thanks to Depesh Mandalia … for the Graduation Framework — small $10–50/d ad set tests where you ALWAYS be testing new creatives and/or audiences.
But for your own sake to save time and mind space, automate the process with rules so you can just drop in your new creative, and FB will tell whether it’s ready to scale or should be killed.
3. FREQUENCY BUMPS.
Frequency is also something that causes your ROAS dip.
You may have very profitable assets but the frequency is high so you try to mess with the ads e.g.: changing creatives, angles, or copy which results in a dip in the ROAS.
Based on your budgets and creative department availability, your frequency number may be different – from 2 to 4. Be aware that anything over 4 could be deadly in terms of CPC increase.
*As with many aspects of Facebook, there’s not a one-size-fits-all approach.
Variation in the ad creative > you use the same or similar angles or variations of angles in your creatives > people get tired of seeing the same ads
Placements > they are simply fatigued
Negative feedback > how many times do people dislike to leave negative feedback on your ad
Having a wider variety of creatives and their angles
Manual monitoring > simply make changes manually when the frequency exceeds your comfort level
Adjusting your ads, offer, and copy
4. AUDIENCE OVERLAPS AND ADVERTISING AGAINST YOURSELF
Another mistake is not paying attention to the type of audience you work with, such as mixing interests with LALs.
Personally, I've seen bad results when they are mixed together, especially in the testing phase.
The size of the audiences matter.
So always try to have similar size audiences (2m to 8m together etc..)
A thing to remember…
>Overlap with no exclusions > higher CPC!
>Overlap with exclusions > lower CPC!
How to check your overlaps:
Go to Audience on Facebook
Select the audiences you want to compare in Facebook (up to 5)
Click Three Little Dots next to the Delete button and then “show audience overlap”
Figure out if your audiences overlap
The overlap tool will show your “Overlap” and the “% Overlap of Selected Audience” for each of your comparison audiences.
Found a high percentage of audience overlap?
You got 2 options:
Regroup overlapping ads into fewer ad sets – if ads under different ad sets are targeting the same audience, keep them from competing against one another by compiling them into a single ad set.
Make exclusions > refine your targeting – You can use a more detailed targeting approach to exclude certain interests/behaviours/demographics and refine different aspects of your targeting for each ad set.
Initial wins don’t mean you can scale effectively.
Always simplify, whatever you’re doing.
Put in the right ad account structure.
Test more creatives. Never stop.
Watch your frequency. Adjust when needed.
Check your audience overlaps.