How do you calculate the average bar?

The average bar provides a quick baseline to compare the performance of one ad against the rest of the set. Learn how we calculate it.

Written by Elliott Brand
Updated over a week ago

The average bar is an exciting new Motion feature that allows users to quickly compare the performance of one ad group against the rest of the set. This article delves into how we calculate the averages and best practices for introducing it into your workflow.

# Calculating spend

Spend is calculated by taking a simple mean of the spend of all groups.

# Calculating other KPIs

To calculate the other KPIs, we take a weighted average of the KPI against ad spend, eliminating all KPI values of zero or unknown KPI values from the equation.

## Weighted average

When calculating the average of the KPI, we first weigh each group based on how much they have spent. Take the example below, which would be calculated to show a weighted average of 2.6 for the KPI.

 Ad KPI Value Total Spend Weighting Final Value Ad 1 2 \$8 0.8 1.6 Ad 2 5 \$2 0.2 1

By using weighted averages, we get a final result that is more reflective of the performance of the set as a whole, and eliminate potential outliers caused by data that is high due to limited data size (for example, an ad that has very high ROAS due to low ad spend).

## Eliminating KPI values of zero

Not all ads will be optimized for the same outcomes. For example, in one report you might have some ads that are video ads and other ads that are image ads. Consequently, when calculating the average of a KPI like video plays, including all ads in the calculation will result in an average that skews significantly lower because the image ads will have data values of zero for the selected metric.

As a result, we made the decision to not factor in any ads that have an unknown value or value of zero in any KPI when calculating the average of that KPI. While this will lead to more accurate averages, it is important to keep in mind that the consequence of this decision is that your averages will always be more precise the longer the data range you use and the more ads you compare in a set.

# Rounding

When calculating averages, we rounded the final average digits to the nearest 100 decimal points or the nearest integer, based upon whether the KPI is decimal based (such as ROAS) or whole-number based (such as comments).

# Making accurate decisions with the average bar

The average bar provides a quick baseline for you to compare the performance of one ad group against the rest of the set. This said, we suggest that you keep the following principles in mind when leveraging averages as a metric:

1. The larger the date range, the more precise the average will be. This is because the more time that an ad has to collect data, the less likely outliers will be present in the results of the set (such as really high KPI figures or KPI values of zero)

2. The more ads you have in your report, the more precise the average will be. A report that contains more ads has a larger sample size of data to calculate the averages from.