Mobile retargeting: how to measure performance

    More or less have learned to measure the effectiveness of advertising campaigns for the purchase of traffic. Many people know how to calculate ROI and know how to apply cohort analysis. But when you add retargeting, things get more complicated.

    For several months, we at Getloyal have been doing mobile retargeting for customers from all over the world. In this post, we’ll show you how to measure the effectiveness of retargeting companies with simple examples.




    Pure User Acquisition


    Let's start with a simple example when we do only User Acquisition, without retargeting. Let's say you have a food delivery application. In January, you raised 10,000 units for $ 10,000. In total, over the course of the year, 600 users out of these 10,000 made their first order (conversion 6%). Some of these users made only 1 order per year, some 20, but on average it turned out that each of these 600 users made 4 orders per year. The average earnings per order was $ 5. The first-year advertising campaign ROI was 120%.



    In reality, it rarely happens that an economy converges “head on”, but, in general, this is possible if we have a cool application and targeted traffic.

    A ROI of 120% is good, but you want more, and you decide to apply retargeting. Strictly speaking, you have two options for targeting:

    1. For those who did not even convert to the first order (we will grow C1).
    2. Those who made the first order (we will grow Orders per user).

    Take the first case - we will target only those who have not yet made the first order. As soon as the user made the first order, he disappears from our audience.

    Net retargeting


    To begin with, let's look at an example when we target users who are definitely not converting themselves. For example, we attracted 10,000 users a year ago. 600 of them were converted to first orders. The remaining 9,400 have never made a first order since then. Most likely, they will never make this order if we do nothing, because a lot of time has passed. In this case, we are ready to pay for the return of these users almost as much as for a new user.

    Here is what our model will look like in this case:



    We divided the 10,000 users we brought in a year ago into two groups: 600 and 9,400. In the second group, we consider the cost-taking cost equal to zero - because before that we had already spent $ 10,000 and received 600 paying users. The remaining 9,400 users are a side effect of this campaign.

    This model shows that with the help of retargeting we were able to convert 133 additional users, and the retargeting ROI turned out to be higher than from the usual User Acquisition campaign. In real life, this will not always be the case, by no means all retargeting campaigns work efficiently, but there are cases when retargeting really helps return users cheaper than bring new ones.

    User Acquisition + Retargeting


    In practice, it rarely happens that we know for sure that users will never be able to convert themselves for sure. It makes sense to catch up with user retargeting right after he installed the application. In this case, his interest is still lively, and retargeting will be most effective. However, we won’t know who we really attracted by retargeting, and who was going to place an order anyway. The user could just click on our link, but in fact he would have converted without any retargeting. This effect is called cannibalization.

    In this case, the notorious A / B test will help us. We’ll take half of the users (group A) and launch retargeting for them. The second half of users (group B) will be left as is, without paid retargeting. Over time, we compare the cost of the first order and the ROI in one and the other group. If the ROI in the retargeting group is no lower, then retargeting is justified.

    Here is what our model will look like in this case:



    Here we can compare and see that as a result, CAC (Customer Acquisition Cost) in the group with retargeting turned out to be lower than in the group without retargeting, and the ROI is higher. In this case, we can say that retargeting is effective.

    If the final results in the group with retargeting are lower than in the control group, it is necessary to optimize the retargeting campaign.

    Of course, this model is simplified, and in reality we are dealing with a much larger number of different metrics, with more complex funnels, with a large number of different channels with different traffic costs. However, the basic principles will always remain the same.

    What else is important to consider


    1. In a good way, for each type of retargeting and for each advertising channel, you need to conduct your A / B test. This requires a fairly large amount of time and resources. But then the picture becomes clear, and you can carry out retargeting campaigns, already knowing the general numbers.

    2. If the costs of retargeting are within 1-2% of the total budget, then there is no point in conducting an A / B test - you will not see anything on such small numbers. However, with such volumes, in most cases cannibalization can be neglected.

    3.The cost of attracting through a retargeting channel cannot be considered separately - this information will not work. This cost will always be lower than the cost of attracting new traffic! Because this cost includes both those users whom we returned by retargeting, and those who would return already. Therefore, it makes sense to summarize the bones for attraction with the bones for retargeting.

    4. If we do retargeting for organic traffic, which in fact came to us for free, then the results of retargeting will always be worse if you look at the cost of attraction. Because even the lowest cost is higher than zero. So you can not compare. If we retarget organic traffic, we should consider the final ROI already.

    5.It is dangerous to mix different user cohorts. Those who came to us this month and those who came 12 months ago are different audiences. For those who came a long time ago, and still have not made a single order, we are ready to pay as for a new user. For those who have come recently, we are willing to pay less, because they are likely to convert and so.

    Conclusion


    Careful and painstaking work with retargeting requires a lot of time and resources - it is necessary to test different audiences and conduct many measurements. Doing it manually is quite expensive, so automated tools specially designed for mobile applications come to the rescue. We already talked about them in the article “Mobile retargeting: setting up in trackers and traffic sources” .

    If you want to start returning users to the app and make more money with mobile retargeting, contact us at Getloyal .

    Also popular now: