When conducting a brand-lift study that's testing multiple advertising strategies, Facebook uses the metric cost per incremental person (CPIP) to represent the amount you spent to get one additional person to respond favourably to a poll about your ad. CPIP can then be used to help determine which strategy is the most impactful and cost-effective. We typically assess the strategy with the lowest CPIP as the winning strategy, and you can use these results to make more informed decisions about your ads in the future.

For example, let's say that you have two advertising campaigns and you want to use a brand-lift study to understand which one resonates more with your audience. Facebook will divide your target audience into four groups: two test groups and two control groups. Within each test group, there will be an exposed group that actually saw your campaigns. For this example, let's call these two exposed groups "group A" and "group B".

After we poll people in group A, group B and the control groups, we analyse the results to determine the CPIP for the exposed groups. CPIP is one of two criteria for determining the winning strategy, and if the CPIP for group A is GBP 1 versus GBP 0.20 for group B, then the advertising strategy used for group B is most likely the winning strategy. With this advertising strategy, the cost for your brand to favourably influence one person is less than the strategy used for group A.

Note: For group B to be the definitive winning strategy, it must have a lower CPIP, and the results for the metric you're testing (for example, brand awareness or ad recall) must show a statistically significant difference between group B and its control group. Learn more about statistical significance in brand-lift studies.

How is cost per incremental person calculated?

CPIP is calculated by taking your total spend and dividing it by the number of influenced people. Using the example above, let's say that your total spend for the group B campaign is GBP 100,000 and the campaign influenced 500,000 people. We calculate CPIP by dividing 100,000 by 500,000, and the result is a CPIP of GBP 0.20.

Why do we use cost per incremental person?

When viewing the results of a brand-lift study, you'll also see the metrics percentage point lift and influenced people. Percentage point lift is the difference between the proportion of people in the exposed group and the people in the control group who took the desired action. Influenced people is the estimated number of people who responded favourably to your ad (for example, people who remembered your brand after seeing your ad).

While percentage point lift, influenced people and CPIP are all important metrics, we believe that CPIP is the most important because it is directly tied to what you spent and is calculated on a per-person basis.

Why do we not use percentage point lift?

While percentage point lift is valuable because it determines the amount of lift that occurred as a result of your campaigns, it doesn't measure the reach of your campaigns. When considering different campaign strategies, you'll want to select the strategy that will have the greatest impact with the largest audience. Percentage point lift sometimes only shows lift based on a narrow audience and may not accurately assess if the advertising campaign would be successful on a larger scale.

Using the group A and group B example, let's say that the percentage point lift for group A was 10%, while the percentage point lift for group B was 5%. If you were only using this metric, you'd conclude that group A's strategy was better.

If the strategy for group A actually reached fewer people, the number of influenced people would be lower than the influenced people for group B, which would mean that the results for group A weren't scalable for a large audience. In the graph below, the top line represents group A and the lower line represents group B. For group A, when we calculate influenced people by multiplying point lift (10%) by total reach (1 million), the result is 100,000 people. While the point lift for group B was smaller (5%), the total reach for group B was actually larger (10 million), so the resulting number of 500,000 influenced people was also larger.

We believe that CPIP is the better metric for determining success because it considers total spend and influenced people. Using this metric, we can select a winning strategy that will have the greatest impact when scaled to a large audience. In the graph below, you can see that group B (represented by the lower line) has the winning strategy, as its CPIP is lower.

If you have more questions about brand-lift studies and how the results are calculated, please contact your Facebook account representative.

* Nguồn: Facebook