Blog


Scaling Facebook Ads – We tested 7 methods, here’s what’s working best….

Split Test: The Best Way to Scale According to 7 Methods

So your campaign is running great and it’s time to scale – or maybe your client wants to scale fast because they have a budget or ad spend quota they need to hit. However, what’s important to remember when you’re trying to scale, is that there are a number of factors that should change your approach. What are you optimizing for, for example? How fast do you want to scale?

What’s the difference?

There are a number of scaling methodologies, here’s the ones we’d like to test:

  1. Horizontal, no bid: duplicate ad multiple times, stay at the same bid amount/automatic bid
  2. Horizontal, manual bid: duplicate ad set multiple times, change manual bid
  3. Vertical, slow: scale ad set slowly by 16% per day
  4. Vertical, xtra slow: scale ad set by 10% every three days
  5. Vertical, fast: scale ad set by 40% per day
  6. Vertical, xtra fast: scale ad set by 100% per day
  7. Random, campaigns: duplicate entire campaign (pause ad sets that aren’t needed for test in duplicated campaign)

The Test

We wanted to see if there was really some “best” way to scale, even though scaling is heavily dependant on several factors.

Hypothesis

In this case, coming up with an exact hypothesis is hard, because as mentioned before, there are a few things you might want to optimize for when scaling.

For example, what is the priority here? Is it to scale as fast as possible? Or is it more important to maintain a healthy CPA/CPL? Which means there may not be a “best” way to scale, because there are definitely multiple interpretations of “best” in this case.

So why not run each methodology and see what happens?

Test Rules

  1. We are limiting the scaling tests to methods that keep the creative the same. This means that our test does not consider the possibility of introducing new offers in order to scale.
  2. We are not scaling out to new audiences in this test.
  3. The original ad set will contain 2 live ads each. For the course of the test, we won’t be allowed to pause or create any ads within any ad set that is part of the test.
  4. We will run this test on 1% lookalike audiences only, to keep the audience sizes the same across all tested ad sets.
  5. We will NOT select expanded interests in any ad set that is part of the test.
  6. We will not do any form of optimization over the course of the test.

Strategy

Here’s how we ran our test:

  • We created 7 separate campaigns (1 for each test)
  • Optimised everything for the ‘Lead’ conversion
  • 7 Days Click 1 Day View
  • Automatic Placements

Some possible problems that could affect the outcome:

  • Overlapping audiences
  • Differing auction conditions across the difference audiences

Results

Method 1

Pre-scale:

Total Spend: $130.48
Daily Spend: $20
CPL: $4.50
CPC (Link): $1.09
CTR (Link): 0.92%
Leads: 29

7 Days Post-Scale:

Total Spend: $540.02
Daily Spend: $80 (in one ad set ads were disapproved and never ran)
CPL: $7.17
CPC (Link): $1.43
CTR (Link): 0.91%
Leads: 76

This resulted in a 59.33% rise in cost per opt-in, in the 7 days post reaching desired scale (and we didn’t even quite reach desired scale because one ad set was disapproved)

 

Method 2

Pre-Scale:

Total Spend: $128.56
Daily Spend: $20
CPL: $5.84
CPC (Link): $1.08
CTR (Link): 1.42%
Leads: 22

7 Days Post Scale:

Total Spend: $638.52
Daily Spend ~$100
CPL: $13.59
CPC (Link): $2.31
CTR (Link): 1.15%
Leads: 47

This resulted in a 132% rise in cost per lead in the 7 days post reaching desired scale

 

Method 3

Pre-Scale:

Total Spend: $128.35
Daily Spend: $20
CPL: $4.14
CPC (Link): $1.38
CTR (Link): 1.36%
Leads: 31

7 Days Post Scale:

Total Spend: $666.92
Daily Spend: ~$100
CPL: $9.39
CPC (Link): $1.67
CTR (Link): 0.98%
Leads: $9.39

This resulted in a 126% rise in cost per lead in the 7 days post reaching desired scale

Method 4

*This is the super slow scaling method. We still haven’t reached our daily budget goal of $100/day

But the cost per lead has shot up and it seems this is too slow to be practical.

Method 5

Pre-Scale:

Total Spend: $128.35
Daily Spend: $20
CPL: $6.11
CPC (Link): $1.13
CTR (Link): 1.55%
Leads: 21

7 Days Post Scale:

Total Spend: $701.49
Daily Spend: ~$100
CPL: $6.68
CPC (Link): $1.28
CTR (Link): 1.15%
Leads: 105

This resulted in a 9.33% rise in cost per lead in the 7 days post reaching desired scale

 

Method 6

Pre-Scale:

Total Spend: $128.47
Daily Spend: $20
CPL: $6.76
CPC (Link): $1.61
CTR (Link): 1.10%
Leads: 19

7 Days Post Scale:

Total Spend: $701.53
Daily Spend ~$100
CPL: $6.56
CPC (Link): $1.37
CTR (Link): 1.12%
Leads: 107

This resulted in a 3% DROP in cost per lead in the 7 days post reaching desired scale

 

Method 7

Pre-Scale:

Total Spend: $128.08
Daily Spend: $20
CPL: $4.42
CPC (Link): $0.95
CTR (Link): 1.41%
Leads: 29

5.5 Days Post Scale (These got cut because people weren’t moving through funnel at all on the back-end):

Total Spend: $525.74
Daily Spend ~$100
CPL: $7.30
CPC (Link): $1.38
CTR (Link): 1.26%
Leads: 72

This resulted in a 65% increase in cost per lead in the 5.5 days post reaching desired scale

 

 

Scaling 40% or 100% daily performed the best in our test.

 

Final Conclusions

Scaling 40% or 100% daily performed the best in our test. The 100% daily scaling actually resulted in a lower cost per conversion at scale. It’s also worth noting that for both of these scaling methodologies the CPL didn’t really shoot up even as we were still scaling.

None of the scaling methods where we duplicated ad sets or campaigns seemed to work that well for us. Though it’s possible that by getting the right configuration of manual bids you might have better luck.

All that said, please consider that there are other scaling methods that we didn’t test here. For example: scaling out to new audiences has traditionally worked really well for us. Also campaign budget optimization is looking like it’s going to be a really really good tool for scaling (results to come). And finally, introducing new offers to your audience can be another great way to scale.

Usually it’s a matter of combining a few of these methods in order to achieve the best results!