How to apply insights?
5 use cases

Captain Growth gives you insights, but decision making is creative job and it’s up to a marketer to define what actions to take. Meanwhile, we extracted from our experience 5 typical actions that you can possibly apply based on Captain’s insights either to save your money on Facebook ads or to leverage results and get more sales from social media ads.

5 TOP actions on Captain’s insights:

  1. Stop ads
  2. Extract growthpoint and make a split test
  3. Exclude bottleneck
  4. Decrease budget
  5. Increase budget

Stop ads

Killing ads could be a quite abrupt decision. To stop it, you should be sure that this ad or adset doesn’t work at all and just burns your money. Captain is able to help you to make such decisions. The only thing you need to check is ‘Segment details’ of your insights. If you see that a problem lies exactly on an ad or an adset level, it makes sense to get rid of this ad or adset.

Here is an example:

In this case, a particular AD in one the adsets is a bad performer. It has burned $809 during last 7 days and it makes sense to stop it to save resources and spend them more effectively.

Extract growthpoint and make a split test

Quite often you can see insights on an intersection of different factors. Actually, these insights are the most interesting because they are hidden deeply in your data and such knowledge gives you a competitive advantage. To use this advantage, you should apply insights correctly.
If an insight is green, you should try to extract audience from ‘segment details’ and create a duplicate of an existing adset that is targeted only on this people.
Let’s consider this insights.

Here is an example:

In this case you should not change the existing adset. Just create a duplicate and target it on a growthpoint segment only: adset + male + 35-45. Put a test budget to this new adset and see how it performs versus the original one. It is called split testing. If the performance is nice, you can put more money in a new adset and increase budget step by step.

Exclude bottleneck

Bottlenecks on an intersection of different factors can be applied by excluding bad performing audience from adset. There are 2 options to cope with it:
1) Exclude bad performing segment from the adset.
Apply this option to adset that has already collected a significant amount of actions (leads, conversions, clicks, etc), so Facebook’s algorithm already learned about them and you don’t want to lose this optimization. Duplicate it, exclude the bad segment in the copy and keep track on its performance versus the original one.


2)Create new adset without bad performer and run a split-test.
Apply this option to adsets that didn’t collect a lot of actions so far or has poor performance. In these cases, you do not risk to kill Facebook learning as soon as the adset is not well learned yet. A different case for this action is if you collect 50 conversions per adset quite fast (1-2 days), so you can go through learning phase quickly.


Decrease budget

If adset performs worse that others, but it still brings you some leads, try to apply soft power. Decreasing budget will help you to save resources on bad performers and allows to relocate this money to ads or segments with better results. The percentage of your budget that should be decreased is up to you and totally depends on your strategy, but Captain advises to decrease the budget for adset by at least 20% to see some result.


Increase budget

If you see green insight with adset in segment detalization, it means that you have an opportunity to put more money in this adset. We advise to increase budget carefully and always track your performance. You should keep in mind that scaling usually decreases performance because ads cover more people with higher frequency and promotion may become not so accurate as it was on the smaller scale.
So first, you can put 10% more in an adset. If it works well after 1-2 days, try to increase the budget by 10% once again. If performance is still satisfying, try to repeat this action 3-5 times.

( No Ratings Yet )

Would like to know how
we create world's first AI marketer?

Subscribe to our blog. Mistakes, machine learning and rock-n-roll - all truth here.