This is the only in-depth tutorial on the Facebook algorithm on the internet.
In this new tutorial you’ll learn, step-by-step, how to:
- Manage your own ads
- Get better results
- Create your own strategies
Let’s dive right in.
Why Facebook Ads
With the wide variety of options when it comes to paid advertisements, you may find yourself asking why Facebook ads are so effective.
Facebook’s main competitors are Google ads, Snapchat, and Tik Tok. Other forms of competition include Instagram influencers and organic SEO (search engine optimization).
However, none of these platforms come even close to Facebook when it comes to running advertisements for your online store.
Launching in 2007, Facebook ads have been constantly improving and evolving for over a decade.
There are dozens of reasons why Facebook ads are so effective, but three main reasons are because of its massive userbase, amazing targeting options, and fast scaling ability.
Furthermore, the main reason that Facebook is so much more effective than its competitors is its machine learning algorithm.
Let’s take a closer look at each of the 3 top benefits before looking closely at Facebook’s main mechanism, which is its machine learning algorithm.
Facebook’s Massive Userbase
According to Facebook, 1.79 billion people on average log onto Facebook everyday and are considered daily active users.
People always say that Facebook’s userbase is dying and less people are using it, however, there was a 12% increase of daily active users from June 2019 to June 2020.
Facebook’s userbase is not only already massive, but it’s still growing as well.
As an advertiser, this should excite you and help you understand why Facebook ads are so powerful.
3.14 billion people actively use Facebook, Instagram, WhatsApp, or Messenger each month.
This means that you have the ability to advertise to over 40% of the global population through Facebook ads.
No other platform has an audience this large that you can effectively target.
Facebook’s Bullseye Targeting
Facebook ads offer the ability to pinpoint the exact audience you’re looking for.
For example, if you were selling athletic apparel, you can easily find audiences that are interested in fitness, working out, or Nike.
You can also specify the age, gender, country, occupation, and much more using Facebook ads.
Furthermore, Facebook offers a tool called Lookalike Audiences.
The Lookalike Audience tool works by creating a custom audience of people who already interacted with your advertisement. You then create a Lookalike Audience using your custom audience, and Facebook finds people who have similar attributes.
For this reason, Lookalike Audiences are only effective once you’ve already run advertisements and collected data.
The ability to automatically generate audiences based on data you give Facebook makes it extremely easy to run advertisements.
Rather than continuing to spend money testing certain interest groups and demographics, Facebook auto generates audiences based on the data of users who have already interacted with your advertisement and website.
This type of hands-off targeting is one of the reasons why the Facebook ads platform is so powerful.
Facebook’s Scaling Ability
Scaling your Facebook ads means increasing your ad spend while maintaining a positive return.
Businesses will scale their stores once they’ve seen consistent and profitable returns.
However, this is one of the hardest challenges that online businesses face because they can potentially lose money if the scaled advertisements don’t perform well.
Despite the potential risks, Facebook ads is the easiest platform to scale your business when compared to its competitors.
Facebook’s machine learning algorithm and ability to optimize means that Facebook will always show your ads to the best users.
Nothing compares to Facebook’s accuracy and ability to predict potential consumers.
Facebook Ads and Verum Ecom
At this point you should have a basic understanding of why the Facebook ads platform rises above its competitors.
There’s nothing inherently wrong with the other platforms, we’ve just seen the most success using Facebook.
The Verum Ecom team has been using Facebook ads for over 3 years and generated multiple 6 and 7 figure dropshipping stores. Check out our YouTube channel with 100,646 subscribers after reading this article to learn more about Facebook ads and dropshipping.
It’s always a good idea to test multiple platforms and see which one garners you the best results. However, we recommend that you take it one step at a time and learn how to use the Facebook ads platform first.
The skills you gain from learning Facebook ads will help immensely if you ever try using another platform.
That being said, let’s take a look at the Facebook ads machine learning algorithm.
The Facebook Machine Learning Algorithm
Facebook’s machine learning algorithm is the engine that helps Facebook find the best people to deliver your advertisements to.
Your goal as an advertiser is to learn how the algorithm works so that you can leverage it to run effective ad campaigns.
What is Machine Learning?
Machine learning provides systems with the ability to automatically learn and improve as it receives new data.
The key factor is data.
Data is the basis of any machine learning algorithm.
The data that the Facebook machine learning algorithm uses for your ads comes from your website visitors.
This is all possible because of the Facebook Pixel.
The Facebook Pixel is a piece of code that allows websites to track their visitors.
Once a user visits your website by clicking on your Facebook ad, the Pixel will log all the actions the user performs and then profile the visitor.
They create the profile based on the users age, sex location, likes, interests, groups, browsing history, and much more.
The algorithm will then start to find patterns in the data to make connections and ultimately generate the estimated action rate.
To find the estimated action rate, Facebook’s machine learning system predicts a particular person’s likelihood of taking the advertiser’s desired action (business objective) based on all of their data.
How the Algorithm Learns & Optimizes
Let’s go through an example so you can visualize how the process works.
For our example, let’s pretend you are an advertiser selling a golden retriever product and are looking for audiences to target.
Luckily for you, the Facebook machine learning algorithm has already found all the people interested in golden retrievers.
Based on the people who have already interacted with golden retriever posts, Facebook grouped them into 1 or more interests such as dog lovers, golden retriever, puppy love, and dogs.
You can type these interests in at the ad set level of your Facebook ads manager.
However, you can see that the total audience size is massive with a potential reach of 150,000,000 people.
And some of these people might not be interested in your product.
Targeting each individual person would be extremely expensive and unrealistic.
Luckily for you, this is where the machine learning algorithm comes in.
Whenever you target a specific audience, Facebook will pick smaller groups of people within the audience to target.
Facebook will use these small samples to find common data points and then expand the audience based on those data points.
Let’s say you launch an ad targeting the interest group ‘golden retriever’ that consists of 13,000,000 people.
Although the audience is big, Facebook will only show it to a small group of people on the first day.
Let’s say Facebook showed the advertisement to 1000 people on the first day.
After the first day is over, Facebook will determine which users reacted positively to the advertisement.
If 50 people reacted positively, Facebook will take the 50 people and try to find people in the 13,000,000 audience who have similar attributes.
The next day, the algorithm uses the data and decides to show your advertisement to 10,000 people aged 20-30 that are actual dog owners.
The algorithm will continue to evaluate the performance of the ads and expand the audience.
If an audience produces poor results, Facebook will restart the learning process and look for a smaller audience again.
The Facebook ads machine learning algorithm is in a constant process of testing data, analyzing results, and making adjustments. The machine is always trying to find the best people to show your ads to.
This example was just a brief overview so you have a clear understanding of how the Facebook algorithm works and why it’s so powerful.
The actual in-depth mechanics and formulas are confidential to Facebook.
However, knowing the general concepts will tremendously benefit you.
The Key Takeaway of the Facebook Algorithm
Rather than blindly showing your advertisement to people, the machine learning algorithm shows your ad to people with a high estimated action rate.
In summary, the algorithm automatically optimizes your ad campaigns to help you achieve better results.
Now that we’ve covered Facebook’s machine learning algorithm, let’s discuss the Facebook ads structure and how the algorithm plays a role in your ads manager.
Facebook Ads Structure
The Facebook ads hierarchy refers to the way in which Facebook structures your campaigns.
Just like in any type of paid marketing, campaigns are a way to interface with Facebook’s machine learning algorithm.
Campaigns will tell the algorithm how much money to spend, who to target, and much more. Campaigns are made up of many components, each providing different pieces of information it will use.
Facebook ads consist of 3 components which are campaigns, ad sets, and ads.
We will break down each component and tell you what role the component plays when interfacing with the machine learning algorithm.
Campaigns
The primary purpose of a campaign in the context of machine learning is to provide the machine learning algorithm with the objective it will be optimizing for.
The objective tells Facebook to find people who are most likely to take your desired action, and show them your ad.
There are 11 marketing objectives you can choose from which include:
Brand awareness, reach, traffic, engagement, app installs, video views, lead generation, messages, conversions, catalog sales, and store traffic.
For example, choosing video views will show the ad to people most likely to watch your video.
On the other hand, choosing conversion will show the advertisement to people who are most likely to convert.
Although there are many options, as a dropshipper you’re almost exclusively going to use the ‘Conversions’ objective.
Conversions are the same thing as sales.
By choosing conversions, you are telling the Facebook machine learning algorithm to find people most likely to buy products from your website.
Furthermore, you have the option of using ‘Campaign Budget Optimization’ or ‘CBO’. We will discuss this later on in the article.
Ad Set
The ad set is where you will be defining what audience you want your advertisement to be shown to.
To define your audience, you must specify the following settings: targeting, delivery, schedule, bidding, and placements.
Targeting
Allows you to specify various details about who you want to serve your ad to. These include age, gender, location, and specific interests.
The potential reach in the upper-right hand corner is the total amount of people in your specified audience.
Furthermore, you can again specify which action you want the machine learning algorithm to optimize for.
For example: specifying optimize to purchase optimizes to finding people most likely to purchase.
We recommend dropshippers choose purchase, which is the same thing as conversions at the campaign level, making Facebook work even more to target buyers.
Scheduling
Scheduling allows you to specify when the algorithm should begin displaying your ads on Facebook.
In other words, when you want your ads to start running. You can choose the day of the week as well as the time.
When choosing your time, we recommend starting the ads at midnight. This will give Facebook enough time to optimize throughout the day.
Starting the ads too late in the day will cause Facebook to rush and potentially show the ads to nonoptimal people.
Bidding
Bidding is how much you want to spend in the ad auction.
The ad auction is essentially a mechanism that Facebook uses to determine which advertisers competing for the same audience gets to show their ad.
You can specify your own bids, which is manual bidding, or let Facebook handle things, which is automatic bidding.
We highly recommend advertisers choose automatic bidding.
Manual bidding is much more complex and is useful for more advanced strategies.
We mainly use automatic bidding because it’s shown successful results.
Placements
The various areas within Facebook’s platform that your advertisement can be displayed on.
For example: Facebook newsfeed, Instagram stories, Facebook marketplace, and much more.
You have the option to manually select each placement or choose automatic placements.
Automatic placements allows the Facebook algorithm to choose the placement it thinks will perform best given your advertisement.
Automatic placements tend to work best because it utilizes the machine learning algorithm.
It’s become so advanced that it not only knows the best type of placement given your ad and audience, but also knows that some placements are better than others at particular times throughout the day.
As such, it will actually adjust the budget for each placement throughout the day to get you the best results possible.
Ad
This is where you will provide Facebook with the visual content you wish to promote.
In other words, what people will see when the advertisement is being served. A few different options include carousel ads, images, videos, and collections.
Depending on which option you pick, you will have to provide Facebook with pictures, videos, written text, or both.
The ad will ultimately link the user to your website in the form of a button or URL in the text.
The types of content and settings you should use depend on your particular product and audience.
Facebook Ads Manager 101 – How to Gage Ad Performance
Now that we’ve gone over all the components that make up Facebook ads, let’s discuss how to use the ads manager.
The ads manager can be extremely confusing because of all the options, buttons, and tabs on the interface.
However, once you have a grasp of the interface, it will become a powerful tool to gage the performance of your campaigns.
Once you load into your ads manager it will look like this.
Starting from the top, let’s go over each of the menu items.
The top left will show which ad account you are using and if you click it, it will then display all the ad accounts you own or have been added to.
The date selector on the top right will give you the option to choose the date range of data you want to display in your ads manager.
The refresh button will allow you to refresh the data to give you the most up-to-date information.
The 3 dots icon next to it will have 2 options which are ‘Reset Ads Manager’ and ‘Keyboard shortcuts’.
You can reset your ads manager if you encounter any glitches.
Furthermore, you can view the various shortcuts if you want to learn them.
The shortcuts are completely optional because everything can be performed by clicking your mouse.
Next, in the main window you will see campaigns, ad sets, and ads.
You can view all the ad sets in a particular campaign by checking the campaign box and then clicking ad sets.
You can view the ad sets for multiple campaigns by selecting multiple campaigns and then clicking ad sets.
Furthermore, you can see all the ads within an ad set by checking the box next to the ad set and then clicking ads.
Next, let’s look at the filtering options. The ‘Columns’ button allows you to select from different premade options to gage ad performance.
Furthermore, you can click ‘Customize Columns’.
Here, you can manually select each metric you want the ads manager to display.
If you are a beginner, copy the metrics in the image below.
These are the keys statistics you will be looking for as a dropshipper.
Save the preset and name it “Verum Ecom Ads Checklist”.
Your ads manager will then display the selected information in columns.
Next, let’s look at the ‘Breakdown’ dropdown. This allows you to segment your campaign results by whichever attribute you pick.
You can breakdown your campaign by age, gender, country, and so forth.
If you want to break down all of your ad campaigns combined, you can use the ‘Reports’ dropdown and click ‘Create custom report’.
This feature allows you to see the overall breakdown by segment and allows you to get a clearer view of your entire ads manager.
There is a lot more that goes on within Facebook ads, but these are the key features you will be using to monitor your ads performance on a day to day basis.
The next thing you need to learn about as an advertiser is Facebook’s CBO, or Campaign Budget Optimization tool.
CBO Mechanics & Optimization
There are 2 ways to set your budget in the Facebook ads manager.
You can either set a budget for each individual ad set or set the budget of the entire campaign using the CBO feature.
Campaign Budget Optimization is setting the budget of your advertisement on the campaign level, which means you don’t set any budgets for your individual ad sets.
Facebook uses the budget given in the CBO and automatically distributes it to each of your ad sets based on the algorithm’s data.
This feature was first launched in November 2017.
At first, the performance of CBO’s were unreliable, and advertisers preferred setting the budget at the ad set level.
However, CBO’s are performing extremely well now and we recommend that advertisers test them out themselves.
Why did Facebook create CBO’s?
Their goal was to make the Facebook ads platform easier to use, so anyone can run ads without any previous experience.
Currently, the Facebook Ads learning curve is high because you have to be familiar with the machine learning algorithm and various nuances of the platform.
Furthermore, the experience needed to understand the results of your ads and what to change takes years to fully develop.
When should I turn off my ads? How much money should I spend before moving on to another product? How do I scale?
These are common questions that beginners have that are hard to answer.
The answers come down to experience and intuition.
Small businesses have trouble investing money into learning the platform or hiring an expert, and so Facebook misses out on a lot of potential revenue from small businesses that choose to not run Facebook Ads.
Facebook realized this and decided to do something about it by introducing Power5.
The Power5 is a collection of features that make the platform easy to use. In addition to CBO’s, you have automatic placements, dynamic ads, account simplification, and auto-advanced matching.
Some of these features are still being developed by Facebook, but in the next few years their vision will become reality.
Any business will be able to easily run Facebook Ads yielding great results.
How CBO’s work
Rather than manually specifying the daily budget for each ad set, the algorithm decides how to allocate your daily campaign budget between your ad sets.
The goal is to allocate the most budget towards the ad sets that are yielding the best results.
Let’s look at an example comparing a campaign with CBO and without CBO.
A normal campaign without a CBO is called ‘Ad Set Budget Optimization’ or ‘ABO’.
The CBO budget is $30, whereas the ABO ad sets are each budgeted at $10.
In summary, both of them have 3 ad sets and will spend $30 a day.
For campaign without CBO
- Each ad set has $10 budgeted per day
- 1st ad set has 3 purchases at $3.33 per purchase
- 2nd ad set has 5 purchases at $2.00 per purchase
- 3rd ad set had 2 purchases at $5.00 per purchase
- Overall cost per purchase (CPP) = $3.00
CBO Campaign
- 1st ad set had 3 purchases at $7 budget for $2.33 CPP
- 2nd ad set had 10 purchases at $18 budget for $1.80 CPP
- 3rd ad set had 2 purchases at $5 budget for $2.50
- Overall CPP = $2.00
Why did the CBO do better?
The campaign started off by spending a little bit on each ad set.
After some amount of time, the Facebook algorithm determined that ad set 2 was performing better.
Because of this, it allocated more of the budget to ad set 2. As a result, you end up with a better cost per purchase while using the CBO.
This is how the CBO is ultimately supposed to work.
However, you might see your CBO’s behave differently. Sometimes, a CBO allocates most of its budget towards an ad set that ends up yielding the most expensive CPP.
Alternatively, you may see the results of the CBO vary on a day to day basis. It’s important to test both the ABO and CBO, so you can determine the results on your own store.
When creating your CBO’s, here are some tips we’ve compiled to help you achieve better results.
CBO Best Practices
- Use maximum of 1 ad set per $15 of spend
- If CBO budget is $50, you can have a maximum of 3 ad sets
- The point of this rule is to ensure that each of your ad sets gets enough spend to optimize properly
- You can have a maximum of 1 ad per $5 per spend
- If your CBO budget is $50, you can have a maximum of 10 ads total
- Don’t be too quick to turn off ad sets within a CBO
- Give Facebook at least 2 days to optimize
- If you scale the budget up, and realize the CBO performance is going down drastically, then just set the budget back to its original for the next day
- Performance will return to normal
- Avoid making big changes to your CBO during the learning phase
What is the learning phase and how do I deal with inconsistencies? Let’s talk about it.
Facebook Ad Inconsistencies & Learning Period
You may notice that your ad sets say ‘Learning’ when you launch new ads on Facebook. A common question we get is what this learning period means and how to deal with Facebook ad inconsistencies.
The learning phase is when the machine algorithm is figuring out which settings deliver the most optimal and stable results.
The ad set is learning who the target demographic is, where the optimal placements are, budget delivery time, ad budget allocation, and much more.
There’s a lot of things happening in the backend of Facebook. For this reason, your ad performance can be very unstable during the initial learning phase.
To exit the learning phase, Facebook suggests at least 50 conversions per week. Based on our experience, our ad performance is extremely consistent once we’ve exited the learning phase. There are times when we don’t need to adjust our CBO’s for weeks or even months because everything has already been optimized.
You will notice more inconsistencies at lower ad budgets. A low ad budget is around $100 a day or less. The reason for inconsistencies at lower budgets is because the Facebooks ads learning machine algorithm has less data to utilize.
Since the optimization happens through analyzing and connecting data, the algorithm has a more difficult time accurately optimizing your ad sets. This is why higher ad budgets optimize quicker and are more stable.
So how can you maintain a stable performance on lower budget CBO’s?
It all depends on your product, creative, funnel and targeting. The better your setup is, the higher chances you have of achieving a successful campaign.
For this reason, it’s important to make sure that every step of your funnel is perfect, from the ad setup to your website checkout process.
You can look at various metrics in the Facebook ads manager to determine which steps you need to modify. If the cost per clicks are high, you will want to modify your advertisement or targeting.
If the add to carts are high but purchases are low, you may want to adjust your checkout process or offer additional discount codes. It’s up to you to constantly test and analyze results.
If your lower budget CBO starts off performing well, you have luckily stumbled into something called an optimization pocket.
The best way to take advantage of this success is to duplicate the campaign, rather than increasing the budget on the initial campaign.
However, if your campaign isn’t doing well, you must perfect the set up as well cut the losing ad sets and keep the winning ones.
Closing Thoughts On the Facebook Machine Learning Algorithm
At this point you should fully understand how powerful the Facebook machine learning algorithm is and why Facebook has such a big advantage over its competitors.
No other platform has the capabilities and features that Facebook has. The ability to automatically take data, analyze it, and optimize your ad campaigns is the main reason why every major company is utilizing Facebooks ads.
Although Facebook ads are hard to learn and require spending a considerable amount of money before becoming proficient, you can check out our videos on YouTube before running your own ads.
We have case studies where we show one our successful stores and how we ran our Facebook ads. Furthermore, we have complete Facebook ad strategies you can use as well.
We recommend that you truly take the time to learn the Facebooks ads platform to the best of your ability. This means you should fully understand each of the functions in the ad manager, know what metrics to look at, keep up with changes in the platform, and learn when to modify your ads.
