The Engagement Matrix is a report type that allows you to understand the overall engagement of features in your product. It can help provide a general understanding of feature usage and can aid in identifying poorly performing features to either refactor or deprecate. By allowing you to see your top and your bottom events in a matrix view together, you can compare frequent and infrequent usage. The Engagement Matrix will also aid in helping prioritize which events or features to focus on in your product.
Table of Contents
- Chart Control Panel
- Sectioned By
- Breakdown Data Table
Note: The Engagement Matrix is available for Enterprise, Growth, and Scholarship clients.
Chart Control Panel
Just like in all our other chart types, the left module is used to select from specific events you want to analyze in the chart. You can select up to 20 of your event types or you can use the default options provided by Amplitude:
- [Amplitude] Top Events: This will show you the top 50 events by event totals in your project in the matrix.
- [Amplitude] Bottom Events: This will show you the lowest 50 events by event totals in your project in the matrix.
- [Amplitude] Top and Bottom Events: This will show you the combined top 50 and lowest 50 events by event totals in your project.
Once you have selected your events, you can use the data table below the chart to deselect events that you do not wish to see in the Engagement Matrix.
Additionally, you can also add 'where' and 'grouped by' clauses in the left module to further segment your events. However, please note that the 'grouped by' clause is disabled for '[Amplitude] Top Events', '[Amplitude] Bottom Events', and '[Amplitude] Top and Bottom Events'.
You have the option to use account-level reporting for this chart type as well and you can do this by toggling the "User" dropdown if you have instrumented groups. In addition, you also have the option to apply cohorts as segments or filter out your users in other ways as well. You can read more about the segmentation module here.
..measured by: In the bottom module you have the option to view the events you have selected by "%DAU", "%WAU", "%MAU". This will be what is displayed in the X-axis. You will also see the text change after the "..measured by" depending on the interval you have selected in the date picker:
- %DAU: Percentage of your daily active users.
- %WAU: Percentage of your weekly active users.
- %MAU: Percentage of your monthly active users.
Note: For some inactive events, you may see a %DAU/%WAU/%MAU greater than 100% because the data point is including users who are only firing those inactive events and may not necessarily be active in the time interval selected.
Average Times Performed/Average Days Performed: Next, you can choose between "Average Days Performed" or "Average Times Performed" which allows you to modify what you want to see on the Y-axis. You can read more about these two different chart visualizations below.
sectioned by: You can also choose to view the chart cross-sectioned by the median or the average number of times. This will change the gray lines that section the matrix to be either the median or average %DAU/%WAU/%MAU and the median or average frequency. Read more about what these lines mean below.
Log/Linear Scale: Finally, in the bottom module you can specify if you would like to see the chart on a log or a linear scale depending on which scale better visualizes the Engagement Matrix for you.
You can modify the Y-axis by changing between the dropdown in the bottom module. This will allow you to see if you would like the average number of times an event was performed or the average number of days on which an event was performed.
Average Times Performed
If "Average Times Performed" is selected, the Y-axis will display the average number of times an event was performed per day/week/month. For example, the following report shows us on the Y-axis the log of the average number of times each event was performed in a month by our active users in the last 3 months.
Average Days Performed
If "Average Days Performed" is selected, the Y-axis will display the average number of days an event was performed per day/week/month. So, if you are looking at the Engagement Matrix in a daily interval, then you may see events at the very top of the Y-axis because you are looking at the average days performed per day. The following report shows you the average number of days each event was performed in a month for our active users in the last 3 months on a linear scale.
The Engagement Matrix is cross-sectioned by two light blue lines that will allow you to get a rough estimate on whether or not a certain event is performing better or worse than the median/average.
If you have median selected, then the vertical line will show you the median %DAU/WAU/MAU that perform each event. It is calculated by taking the median of all the %DAU/WAU/MAUs of each individual event (so each individual data point). It will also reflect appropriately if you are looking at the Engagement Matrix on a linear or a log scale. The horizontal line will show you the median frequency of your events and it is calculated by taking the median of all the individual frequencies of each event (so each individual data point). You can use our Microscope feature to hover over each individual data point and see the numbers that are going into the median calculation. For example, the following chart is telling you that the median percentage of your weekly active users performing each event in the matrix is 58.08%. The median weekly frequency of these events by those users is 1.72 days.
If you have average selected, then the vertical line will show you the average %DAU/WAU/MAU that perform each event. It is calculated by taking the average of all the %DAU/WAU/MAUs of each individual event (so each individual data point). It will also reflect appropriately if you are looking at the Engagement Matrix on a linear or a log scale. The horizontal line will show you the average frequency of your events. It is calculated by taking the average of all the individual frequencies of each event (so each individual data point). You can hover over each individual data point to see the numbers from each point that are going into the average calculation. In the following report, we can see that the average percentage of your weekly active users performing each event is 53.72%. The average frequency for those users is 1.66 days across all the events that are selected.
The light blue lines allow you to gauge the performance of your events or features (e.g. if you have instrumented each feature as an event or several events represent a certain feature). It also helps you prioritize which events or features you wish to focus on. Here is what each section of the matrix represents:
- Top left corner: These are events performed with high frequency by a low number of users. These could represent power features that a small subset of your users find are of high value. You could perhaps think of ways to improve this feature and make it more accessible to the rest of your users. You want to move these data points to the top right corner of the matrix.
- Top right corner: These are events performed with high frequency by a high number of users. These are likely core features or events in your product and represent things a lot of people are doing a lot of the time.
- Bottom left corner: These are events performed with low frequency by a low number of users. These are events or features that you may want to improve or deprecate and have low usage by a low number of your users.
- Bottom right corner: These are events performed with low frequency by a high number of users. This could be something a lot of people think is useful but only use once or twice or they could be one time events all your users must do (e.g. creating an account or finishing an onboarding tutorial).
For example, in the following matrix it seems like the 'PlaySong' event is a core event in our product while the 'PlayFromNotification' event is in the bottom left. This could indicate that your product team should focus on moving the 'PlayFromNotification' event to the right and prompting users more prominently to allow notifications from your app.
Since the Engagement Matrix allows you to look at so many events on a single chart, the chart has a zoom feature implemented. To zoom in on a certain cluster of data points, drag your mouse diagonally across the data points you would like to zoom in on. This will zoom in on that area allowing you to examine those data points more closely. For example, we can see in the following report a cluster of data points that we would like to dig deeper into to see exactly which events are falling into which section of the matrix.
To zoom in on a cluster of data points, drag your mouse so that a rectangle forms exactly as shown in the above screenshot. The Engagement Matrix will zoom in so that you can get a better view of the data points. To return to the original view, click the "Reset zoom" button.
Breakdown Data Table
Below the Engagement Matrix you will find a data table that contains all the information shown above. Like all other data tables in Amplitude, you can perform functions on the columns. The column average or median you select will be the same as the values the cross-sectional gray lines denote. You also have the option to export this data as a CSV file and display either the average number of times performed or the average number of days for each event.
You can also export the table as a CSV file by clicking the "Export CSV" button.