Facebook is one of the most popular advertising platforms. It offers great targeting options and helps businesses reach their target customers effectively. However, in order to get the most out of Facebook ads, you need to be able to analyze their performance effectively.
In this article, we will show you how to do that. We will also share the 4 important steps and some practical tips for optimizing your Facebook ad performance using our analysis framework.
So, let’s get started!
Jump to:
- Step 1: Determining Your Key Metrics
- Step 2: Understanding the Causal Relationships Between The Metrics
- Step 3: Checking Your Blind Spots (with Correlation Matrix)
- Step 4: Using Analysis Framework
Step 1: Determining Your Key Metrics
The first step in analyzing Facebook ad performance is determining and understanding the key metrics. The key metrics vary depending on your business goal, and we have identified the most common ones below for retail businesses.
In brief, we have grouped the metrics into three different layers: primary metrics, secondary metrics, and tertiary metrics.
The primary metrics are the key performance indicators that give you a high-level view of your Facebook ad campaign’s performance, target audience quality, and resonance level. They may vary depending on your business goal, but some of the most common ones include:
- ROAS: Return on ad spend, which indicates how profitable your Facebook ads are. [Calculation: Sales ÷ Adspend]
- CPM: The average cost for 1,000 ad impressions. [Calculation: (Adspend ÷ Impressions)*1000]
- Purchases: The number of purchase events tracked on your website and attributed to your ads.
- Average Order Value (AOV): The average value of a purchase made as a result of your ad. [Calculation: Sales ÷ Purchases]
- Ad Quality: It is our custom-defined quality score, which is calculated based on the average of quality ranking, engagement rate ranking, and conversion rate ranking. It is a crucial metric as it reflects your ad’s relative performance in the ad auction, which can help to lower your CPM significantly.
The secondary metrics give you insights into what factors may be affecting your Facebook Ads performance from ad engagement, onsite engagement to conversion. They can help you identify areas for improvement in your Facebook ad campaign. Below are some of the metrics we have identified or custom-defined:
- Reach: The number of people who saw your ads at least once.
- Frequency: The average number of times each person saw your ad. [Calculation: Impressions ÷ Reach]
- Unique Outbound Clicks (UOC) Rate: The percentage of people who saw your ad and performed an outbound click to visit your website. [Calculation: (Unique Outbound Clicks ÷ Reach)*100]
- Average Content Views: The number of view content events on average for each unique visitor who saw your ad. [Calculation: Content Views ÷ Unique Outbound Clicks]
- Average Adds To Cart: The number of add to cart events on average for each unique visitor who saw your ad. [Calculation: Adds to Cart ÷ Unique Outbound Clicks]
- Checkout Rate: The percentage of people who visited your website and initiated checkout events. [Calculation: (Checkouts Initiated ÷ Unique Outbound Clicks)*100]
- Conversion Rate: The percentage of people who visited your website and made a purchase. [Calculation: (Purchases ÷ Unique Outbound Clicks)*100]
While for the tertiary level, it provides the descriptive metrics that give further elaboration on how well your Facebook ads are performing, which includes:
- CPS: Cost per sale, the average cost per result/purchase from your ads. [Calculation: Adspend ÷ Purchases]
- Sales: The total purchase conversion values on your website attributed to your ads.
- Adspend: The total amount of money you’ve spent on your campaign, ad set, or ad.
Step 2: Understanding the Causal Relationships Between The Metrics
Once you have identified and understood your key metrics, the next step is to understand the causal relationships between them. It helps you to identify which factors are causing changes in your key metrics so that you can take action to improve them. For example, if ROAS is low, you might look at factors such as CPM, AOV, and ad quality score to determine why and then take steps to optimize as necessary.
By understanding the causal relationships between your metrics, you can develop a complete picture of how your ads are performing and make better decisions about optimizing them.
In brief, its key influencing factors are boiled down into six main areas, which include:
- Audience Targeting
- Ad Creative
- Product USPs
- Promotion & Offers
- Onsite Navigation
- Shopping Experience
Based on our learnings, we have identified the key influencing factors for each metric as below.
| Key Metric | Key Influencing Factors |
| ROAS |
|
| Reach |
|
| CPM |
|
| Frequency |
|
| Ad Quality Score |
|
| UOC Rate |
|
| Avg. Content Views |
|
| Avg. Adds To Cart |
|
| Checkout Rate |
|
| Purchases |
|
| Conversion Rate |
|
| CPS |
|
| AOV |
|
| Sales |
|
| Adspend |
|
We want to emphasize that these are still our subjective interpretations of the causal connectives. They may not be completely accurate, but it is a good starting point to guide our thinking during the performance analysis.
Step 3: Checking Your Blind Spots (with Correlation Matrix)
Besides understanding the causal relationships between your metrics, checking for any blind spots in your data is also essential. A correlation matrix can help you do this.
A correlation matrix is a table that shows the Pearson’s coefficient of correlation (r) between all pairs of metrics. The closer the r-value is to +/- 1, the stronger the correlation between the two metrics.
From the matrix, you can identify any pairs of metrics that have high correlations (r-values close to +/- 1), as these could indicate potential blind spots if they are not identified yet with a causal connection between them. Or you may investigate these further to determine if any factors are causing erroneous readings for these metrics.
As you see from the diagram above, we have built a correlation matrix based on the historical data set we collected so far, and below are some examples of insights that we gleaned from it.
For example, the matrix shows that Average Content Views and Average Adds to Cart have a high positive correlation (r-value of 0.74), which suggests that increases in Average Content Views are likely to cause increases in Average Adds to Cart as well. That’s pretty normal, as the more pages people view, the more inclined they are to buy.
However, what we want to emphasize here is not the causal relationships we already know. This correlation matrix is mainly for us to uncover correlations and causal relationships beyond our perception.
For example, from the matrix above, we can see a positive correlation between CPM and Frequency (r = 0.46). The correlation coefficient is not high, but this could indicate a potential blind spot if you’re not tracking these two metrics together. So this example triggers a question to ourselves on why the same people tend to view our ads more when the CPM is higher.
If we further infer the causal relationship between these two key metrics, there are two possibilities for our preliminary conclusion:
- Niche Audience Targeting: This causes the FB algorithm to increase CPM to boost reach, while at the same time, a smaller audience size leads to higher ad frequency.
- Ad Auction Competition: A lower ad quality score leads the FB algorithm to increase CPM to improve reach. If our ads’ relative performance is weaker with a limited ad budget, this could narrow our ads’ reach and tend to deliver ads to the same audience repeatedly.
Again, this is just our subjective interpretation, and we could be wrong. However, by understanding the causal relationships between your key metrics and checking for any blind spots in your data, you can develop a more comprehensive understanding of how your Facebook ads are performing and make better decisions about optimizing them.
Step 4: Using Analysis Framework
Now that we understand the key metrics and the relationship between them, let’s look at the framework we designed that helps us analyze the Facebook ad performance effectively.
In brief, our analysis framework revolves around three primary dimensions:
- Period-over-Period Comparison (Last week/ Last 4 weeks/ Last 13 weeks): measures the ads’ performance in the present and compares them to a comparable period in the past.
- Benchmarking (Target/ Industry Benchmark): compares your ads performance against the industry benchmark or your internal target.
- Momentum (Last 3 Days Average vs. Last 7 Days Average): provides an additional signal to show the most recent rate of change in your ads performance.
By analyzing your Facebook ads performance using the above 3 dimensions, you are more likely to understand what has happened in the past (descriptive), how your ads perform against your target or market average (comparative), and what may happen in the next 3 days (predictive).
From the example above, we can see that the Facebook ad performance is performing well, but the momentum seems to slow down in the last 3 days in terms of its ROAS, AOV, and Purchases. Still, the ROAS is way above our target/benchmark, which we can maintain the ads as it is and closely monitor their performance for the next 3 days.
Based on this data, we may want to consider adding a new ad visual if the ROAS and Purchases are getting poorer within the next 3 days, as we can see the UOC rate is dropping over time which is most likely due to ad fatigue. The onsite engagement (Avg. Content Views, Avg. Adds to Cart) and conversion (Checkout rate, Conversion rate) are generally healthy, which means the onsite visual merchandising still resonates well with the audiences.
This is just one example of how you can use our analysis framework to gain practical insights and make decisions about optimizing your Facebook ads.
Conclusion
Facebook ads performance can be complex and challenging to understand without the proper analysis framework, especially when different people tend to have different subjective interpretations of the data.
Please make no mistake, and we still need to rely on our subjective interpretations to analyze and reveal the story behind the data. However, by applying the right analysis framework, you can develop a common language for your team to communicate faster and more precisely in understanding how the Facebook ads perform and make more informed decisions about optimizing them while minimizing subjective biases.
Our analysis framework is one way of doing this, but there might be many other ways to slice and dice the data. The most important thing is to find the analysis approach that works best for you and helps you make better decisions about your Facebook ads.
For more advanced tips on how to get better Facebook Ads results, you might want to check out the following articles:
- 13 Things You Must Do Before Running Your Facebook Ads
- 7 Pro-Tips on Building Effective Facebook Ads in 2022
- How to Set Up Effective Audience Targeting on Facebook Ads?
- Facebook Automated Rules: What is It & How It Can Help Optimize Your Ads?
Having difficulty in analyzing and optimizing your Facebook ads? Talk to our digital expert today! We offer Facebook Ads managed services, and we’re always happy to help! 🙂
