Genetic Algorithm for Failure Proof Facebook Ads Campaigns Strategy.

This article is not going to be about a new gimmick, tactic or a strategy for Facebook Ads, my aim is to share fundamentals that will enable you to have a clear understanding on how to manage your campaigns and drastically improve your odds of success.

First thing first, you need to understand what’s a Genetic Algorithm.

A genetic algorithm is an optimization heuristic inspired by Darwin’s theory of evolution where reproduction of the fittest and random mutations will play a role in defining the properties of the next generation.

From there I want you to start thinking of Facebook Ads as an organism and accommodate to the fact that an organism is not something isolated and stand-alone but a system with layers and hierarchies.

A natural organism is not a single, final unit; it is composed of subunits and itself may be the subunit of some larger collective.

In nature a cell has a population of intercellular molecules; in turn the organism has a population of cells, and the species has a population of organisms. This is the exact same process for Facebook Ads, an ad has a population of variables (creative, copy, targeting etc..); in turn the ad set has a population of ads, and the campaign has a population of ad sets.

Now I want to share with you the concept of antifragility which is the main reason behind evolution.

When harmed a fragile organism die, a robust organism handles the shock and an antifragile organism benefit from it.

Your goal should be to build antifragile Facebook ads campaign that benefit from randomness, volatility, variety and errors – let’s learn how to do that.

A strengthening campaign comes at the expense of some Ad sets; in turn the Ad sets strengthens at the expense of some Ads and your Ads strengthens at the expense of their own DNA (Copy, Creative, Targeting etc..).

Now that you understand that a Facebook Campaign is simply an organism let’s learn how to build the minimal version of your population.

Here’s my default (but adapt it to your budget) version of a population: 4 copy angles * 4 creatives (image/video) * 10 audience interests. This is a total of 160 unique combinations that are going to be the first generation of your population. From there you are going to be able to optimize and drive insane results. Now it’s time to use a Genetic Algorithm to optimize the population.

In this case our GA will work on a population consisting of some ads where the population size (popsize) is the number of ads. Each ad is called an individual. Each individual has a DNA. The DNA is represented as a set of parameters (features) that defines the individual. The DNA is composed of genes (Copy, Creative, Targeting, Placement etc..).

Also, each ad has a fitness value calculated this way, given a set of x metrics, each metric can hold one of the binary values 0 and 1, for each pre-established minimal metric goal the ad beat it take the value 1. To select the best individuals, this function is used, and the result of the fitness function is the fitness value representing the quality of the ad (number of successful metrics/x). The higher the fitness value the higher the quality of the ad. Selection of the best individuals based on their quality is applied to generate what is called a mating pool where the higher quality individual has higher probability of being selected in the mating pool.

Now your goal is to accelerate the evolutionary process – Thanks to the close to perfect fractality of Facebook Ads – if after 4 days an ad doesn’t reach a satisfying enough fitness value, kill it. Increase the budget of the Ads with a high fitness value and start the mating process. Generate two offspring from two successful ads. By continuously selecting and mating high-quality individuals, there will be higher chances to just keep good properties of the individuals and leave out bad ones.

Now, the offspring currently generated using the selected parents will only have the characteristics of its parents and we want to avoid that to happen too often, otherwise after a few generation members of the same “family” are going to be mixed together and the same way this type of behavior increase the odds of development disorder in humans being, it’s going to make your ad inefficient – this is where we want to introduce mutation.

I’m using a modified DEAP framework (Distributed Evolutionary Algorithms in Python) to shuffle the different variables during the mating process and introduce new ones, but you can simply use your common-sense and get far better results than 98% of Facebook Advertisers.

Once this system is installed, your job as a Facebook Advertiser is mainly to manage your ads by killing, scaling and mating.

Obviously making money with Facebook Ads is way more complex than that but as I said my goal with this article is only to provide you with a clear understanding of the fundamentals!

I hope you are ready to take your ads to the next level

-

Click here to book a Free Discovery Call with me!

To your success.

Links 
Contact Us

pacome@owedigital.com

London, UK ; Rouen, FR