In today's competitive business landscape, it's more important than ever to have a deep understanding of how customers behave over time, which can be done with the help of cohort analysis. In this article, we will explore the importance of tracking custom conversion cohorts and the insights that can be gained from their analysis and show how you can use it all in automation.
In SplitMetrics Acquire, cohorts are groups of users who share a common feature within a selected time frame. Cohort analysis is a dynamic measurement because it doesn't simply add up your actions during a particular time period. Instead, it examines the actions of different groups of users over time. So, you can get an idea of how their behavior changes in the short or long run and use this information to reduce churn and increase conversions, and in a variety of other use cases:
See the full picture of profitability and determine how effectively you are using your advertising budget.
Obtain valuable insights into the performance of your marketing campaigns and find out whether your campaign strategy needs some changes.
Identify patterns in user behavior, such as the most common paths to conversion or the most popular in-app purchases.
Let's consider what types of cohorts we have in SplitMetrics Acquire and how they can be useful for your campaign management 👇.
There are two main optimization models in advertising, cost per action (CPA) and return on advertising spend (ROAS).
For ROAS-based optimization, you can leverage ROAS and Revenue cohorts. Read more about these metrics in Cohort ROAS and Revenue in Automated Rules.
For CPA-based optimization, we offer you to track the following types of cohort metrics:
Cohort cost per custom conversion D0 – D360
Cohort revenue per custom conversion D0 – D360
Cohort number of custom conversions D0 – D360
How do we calculate these types of metrics?
— In our cohorts, users are grouped based on the install date, which represents the day when the app is initially opened after being downloaded and installed on their device.
— We use a day as a cohort size, meaning all users who opened the app on one day are to be included in one cohort.
— We consider the moment of installation as "0". So the first day of installation is D0 and the following day is D1.
— You can use cohorts to trace changes in user behavior from the time of installation up to 360 days. We use a daily granularity, so you can view the following cohorts on the platform: D0, D1, D3, D7, D14, D30, D60, D90, D120, D180, and D360.
Note. Cohorts disregard the time of the event in hours and use only the event day. This trend is typically used to evaluate user performance relative to a campaign.
Example:
Users who opened the app on January 1, 2023, make up one cohort. Users who opened the app on January 2, 2023, make up the second cohort, and so on.
Cohort metrics are available in the:
- Ads Manager
- Total Summary
- Charts and Detailed Charts
- Reports
- Automation
With custom conversion cohorts, you can track full-funnel analytics and run automated rules based on the custom conversion cohorts data, enabling you to optimize bids and therefore avoid overspending.
As mentioned above, there are three types of cohorts.
Cohort cost per custom conversion
The cohort cost per custom conversion metric measures the cost of acquiring a single conversion, for example, a purchase, in a specific group of users (cohort) who opened the app in a certain period of time.
Optimizing ads by cost per conversion cohort is crucial when you have a deep understanding of your customers' behavior over time. By analyzing how long it takes for customers to make a purchase, how often they buy, and how much they spend, you can use this information to improve your cost per conversion. For instance, if you know that most of your customers buy within the first three days of interacting with your ad, you can prioritize your ad efforts during this period to increase conversions.
Additionally, optimizing ads by the cost per conversion cohort can be essential when you have a limited advertising budget. By focusing your ad spend on the most promising cohorts with the highest likelihood of converting, you can allocate your budget to more effective campaigns and keywords.
Cohort revenue per custom conversion
Revenue per custom conversion cohort is a metric that allows you to analyze how much revenue is generated by each cohort of users who converted within a specific time period.
It helps:
Understand the long-term value of your customers and identify trends in customers' behavior over time.
See how the value of your customers changes as they continue to interact with an application.
Identify which customer groups are the most profitable over time, which groups tend to make larger purchases, and which groups tend to have the highest retention rates.
Cohort number of conversions
Tracking the cohort number of conversions allows you to understand how different groups of users behave over time, which can provide valuable insights for improving user retention and increasing revenue.
For example, you may find that users who opened the app in a certain period of time have a higher conversion rate in a particular custom conversion than users who opened the app in another one. This information can give you an understanding of what factors may be contributing to higher conversion rates and how you can replicate that success in other cohorts. Analyzing custom conversions in cohorts can also help you identify the impact of changes you make to your app over time. By comparing the conversion rates of different cohorts before and after a change, you can see if the change had a positive or negative impact on user behavior.
Cost per conversion can be used in the rules to detect and "cut off" the keywords that don't bring conversions. Suppose you have already determined that you want ROAS D3 to be 15% and cost per conversion — $25. Then you can set up a rule to decrease the bid for those keywords that:
1. Didn't bring conversions but spent more than $25:
2. Brought conversions for the 3rd day but spent more than $25:
Conversely, if you fall into your KPI, you can increase the bids with automated rules to get more installs that convert into targeted actions within the cost-per-conversion KPI. In this case, it's necessary to estimate the cost per conversion for each cohort in advance.
💡 Tip. In the automated rules, it's better to use the value of Cost Per Conversion DX – 5% as a buffer.
Let's consider an example of the following set of data for January 1, 2023 – March 1, 2023:
Based on the estimations, you can set up the following rules:
Increase bid by x% if Installs > 0 and Purchase D1 > 0, and Cost per Purchase D1 < $10.05
Increase bid by x% if Installs > 0 and Purchase D3 > 0, and Cost per Purchase D3 < $7.48.
Increase bid by x%, if Installs > 0 and Purchase D7 > 0, and Cost per Purchase D7 < $5.61.
Increase bid by x% if Installs > 0 and Purchase D14 > 0, and Cost per Purchase D14 < $4.74.
Increase bid by x% if Installs > 0 and Purchase D30 > 0, and Cost per Purchase D30 < $3.97.
You can also add rules to further verify the average cost per conversion :
Increase bid by x% if Installs > 0 and Purchase D1 > 0, and Cost per Purchase D1 < $10.05, and Cost per Purchase [for the previous 7 days] < 3.74.
Increase bid by x% if Installs > 0 and Purchase D3 > 0, and Cost per Purchase D3 < $7.48, and Cost per Purchase [for the previous 7 days] < 3.74.
And so on.
💡 Tip. We recommend you set the following minimum date ranges for the data analysis:
For D0 and D1 — previous 3 days.
For D3 — previous 7 days.
For D7 —previous 14 days.
For D14 and D30 — previous 30 days.
For more information or further assistance, access our Support team
or your Customer Success Manager.