Skip to main content
Engagement Matrix helps you understand how users engage with product features by comparing two dimensions:
  • Breadth - how many active users use a feature.
  • Depth - how frequently those users perform the feature.
Each selected event is plotted as a point on a two-dimensional matrix. The matrix uses the selected Median or Average values to divide the chart into four quadrants, helping you compare feature engagement patterns at a glance. Use Engagement Matrix to identify features that are widely adopted, features used frequently by a smaller group of users, features that are broadly used but infrequently used, and features with relatively low adoption and frequency.

What Engagement Matrix answers

Engagement Matrix is useful when you want to understand questions such as:
  • Which features are used by the largest share of active users?
  • Which features are used repeatedly?
  • Which features are both widely adopted and frequently used?
  • Which features are mainly used by a smaller group of highly engaged users?
  • Which features have broad adoption but relatively low usage frequency?
  • Which features have both low adoption and low frequency?
  • How has feature engagement moved between two time periods?
  • How do selected events compare with one another?
The matrix combines adoption and frequency into a single view so that you can compare engagement patterns across events.

Core concepts

How Engagement Matrix works

Engagement Matrix follows four main areas:
  1. Events Configuration - choose the events you want to analyze.
  2. Filters - narrow the analysis using available properties or cohorts.
  3. Date Range & Compare - define the analysis period, granularity, and optional comparison.
  4. Engagement Matrix - review the plotted events and supporting matrix table.

Build an Engagement Matrix

1. Select events

The Events Configuration section controls which events are included in the matrix. You can:
  • Select multiple events.
  • Search for events.
  • Remove selected events.
  • Add filters to individual events.
Each selected event appears as an event card containing:
  • Event Name
  • Filter option
  • Remove option
Each selected event is represented by a plotted point in the matrix when comparison is disabled.

Default event selection

When you first open Engagement Matrix, Top Events is selected by default. The system automatically selects the top 20 events based on overall event usage within the selected time range (by default last 7 days). These events are plotted automatically so you can begin exploring engagement without manually selecting events. You can change this selection at any time by:
  • Removing events.
  • Adding events.
  • Searching for and selecting specific events.
The matrix updates based on the selected events.

2. Filter individual events

You can apply filters to an individual event from its event card. Event-level filters support:
  • Event Properties
  • User Properties
This allows you to restrict the data represented by a particular event. For example, if you select a purchase event, an event-level filter can be applied using the properties available for that event or its users.

3. Choose how engagement is measured

The Measured As section controls the engagement metrics displayed in the matrix. Engagement is represented using:
  • % Active Users
  • Average Times Performed
The active-user metric changes according to the selected granularity.

Daily

Daily granularity uses:
  • % DAU
  • Average Times Performed per day

Weekly

Weekly granularity uses:
  • % WAU
  • Average Times Performed per week

Monthly

Monthly granularity uses:
  • % MAU
  • Average Times Performed per month
This means that changing the granularity changes the active-user population used for the breadth calculation.

4. Choose how the matrix is divided

The Sectioned By option determines how the horizontal and vertical divider lines are calculated. You can choose:
  • Median
  • Average

Median

When Median is selected, the matrix is divided using the median values of the plotted events. The vertical divider represents the median Breadth value. The horizontal divider represents the median Depth/Frequency value.

Average

When Average is selected, the matrix is divided using the average values of the plotted events. The vertical divider represents the average Breadth value. The horizontal divider represents the average Depth/Frequency value. Because the divider values depend on the selected events, changing the selected events can change the quadrant boundaries.

5. Apply filters

The main Filters section allows you to narrow the Engagement Matrix analysis. The available property categories are:
  • All Properties
  • User Properties
  • Cohorts
You can also choose the data type of the filter. Filters affect the data used for the Engagement Matrix analysis.

6. Select a date range

Engagement Matrix supports predefined time windows as well as a custom date range. Available predefined time windows are:
  • Today
  • Yesterday
  • Last 7 days
  • Last 30 days
  • 3 months
  • 12 months
You can also use the Custom Date Range Picker to define a specific analysis period.

7. Select the granularity

The supported granularity options are:
  • Daily
  • Weekly
  • Monthly
Granularity determines the active-user metric used in the matrix: It also determines the corresponding frequency period. For example, Monthly uses % MAU and Average Times Performed per month.

8. Compare two periods

You can compare engagement movement between two relative periods by enabling Compare. Comparison supports:
  • Day
  • Week
  • Month
You can enter a:
  • Start Period
  • End Period
Examples:
  • Day 0 → Day 5
  • Week 0 → Week 1
  • Month 0 → Month 1

Compare visualization

When Compare is enabled, each selected event is represented by two points:
  • One point for the start period.
  • One point for the end period.
The start-period point uses approximately 70-80% opacity. The end-period point uses 100% opacity. For example, if two events are selected, the matrix displays four plotted points:
  • Event A - start period
  • Event A - end period
  • Event B - start period
  • Event B - end period

Compare hover behavior

When you hover over either point belonging to an event:
  • Both points for that event are highlighted.
  • A faint connector line appears between the points.
  • The direction of movement is shown.
  • Other event points appear faded.
This makes it easier to see how an event’s engagement moved between the two selected periods.

Compare tooltip

The tooltip displays:
  • Event Name
  • Start Period % Active Users
  • Start Period Average Times
  • End Period % Active Users
  • End Period Average Times

Incomplete periods

Incomplete periods are excluded from calculations. For example, if the current month is not complete, that incomplete month should not be included when calculating monthly averages. This prevents a partially completed period from being treated as a complete period when comparing engagement.

Understand the Engagement Matrix

The Engagement Matrix is a two-dimensional scatter chart.

X-axis - Breadth

The X-axis represents Active Users: Low Active Users → High Active Users Moving from left to right means that a larger share of the active-user population performed the event.

Y-axis - Depth

The Y-axis represents Frequency: Low Frequency → High Frequency Moving from bottom to top means that users performed the event more frequently. The selected Median or Average values create the horizontal and vertical divider lines.

Understand the four quadrants

The matrix contains four quadrants.

Core Features

High Active Users + High Frequency These events are used by a large share of active users and are performed frequently. Identifies this quadrant as representing core features.

Power Features

Low Active Users + High Frequency These events have lower breadth but higher frequency. Identifies this quadrant as representing power-user features.

Utility Features

High Active Users + Low Frequency These events are used by many active users but have lower frequency. Identifies this quadrant as representing utility features.

Underperforming Features

Low Active Users + Low Frequency These events have both lower breadth and lower frequency relative to the other plotted events. Identifies this quadrant as representing underperforming features. The quadrant is a relative classification based on the selected divider calculation; it does not by itself explain why an event has its position.

Matrix Table

The Matrix Table appears below the visualization. It provides a tabular view of the events plotted in the matrix. The table contains:
  • Checkbox
  • Event Name
  • % Active Users
  • Average Times Performed
The table also displays the overall Median or Average values used for the matrix dividers.

Select or deselect events

You can use the checkbox for each event to control whether that event is included in the matrix. When an event is deselected:
  1. The event is removed from the matrix.
  2. The Median/Average calculations are recalculated.
  3. The quadrant boundaries update automatically.
This means the matrix boundaries reflect the currently selected events.

Export the Matrix Table

You can export the matrix table as a CSV file. The downloaded CSV contains the visible table columns. This allows you to take the table data outside the Engagement Matrix for further analysis.

Save an Engagement Matrix

You can save the Engagement Matrix chart and:
  • Pin it to an existing dashboard.
  • Pin it to a new dashboard.

How the Engagement Matrix is calculated

The Engagement Matrix uses Breadth and Depth calculations for each selected event.

Breadth

Breadth represents the percentage of active users who performed the selected event. Conceptually: Breadth = Users who performed the event ÷ Active Users × 100 The active-user population depends on the selected granularity. For example:
  • Daily → DAU
  • Weekly → WAU
  • Monthly → MAU

Depth

Depth represents the average number of times the event was performed by users who performed that event. Conceptually: Average Times Performed = Total Event Occurrences ÷ Users Who Performed the Event For example, if:
  • 295,998 users performed an event.
  • The event occurred 2,669,466 times.
Then: 2,669,466 ÷ 295,998 ≈ 9.0 This means the users who performed that event performed it approximately 9 times during the relevant period.

Example: Reading a feature’s position

Suppose you are analyzing a product feature represented by an event. If the event appears:

Far to the right

A relatively large percentage of active users performed the event. This indicates higher Breadth.

High on the chart

Users who performed the event performed it frequently. This indicates higher Depth.

Top right

The event has both high Breadth and high Depth relative to the other selected events. It therefore falls into the Core Features quadrant.

Top left

The event has lower Breadth but higher Depth. It falls into the Power Features quadrant.

Bottom right

The event has higher Breadth but lower Depth. It falls into the Utility Features quadrant.

Bottom left

The event has lower Breadth and lower Depth. It falls into the Underperforming Features quadrant.

Example: Comparing engagement movement

Suppose you select a feature and compare: Day 0 → Day 5 The matrix displays two points for that feature.
  • The Day 0 point represents the earlier comparison period.
  • The Day 5 point represents the later comparison period.
When you hover over either point, both points are highlighted and a connector shows the movement between them. The tooltip lets you compare:
  • Day 0 % Active Users
  • Day 0 Average Times
  • Day 5 % Active Users
  • Day 5 Average Times
This lets you see whether the feature’s Breadth, Depth, or both have changed between the selected periods.

Typical ways to use Engagement Matrix

Identify core features

Select the relevant product events and look for events positioned in the Core Features quadrant. These events have both relatively high adoption and frequency.

Find highly frequent niche features

Look at the Power Features quadrant. These events have high frequency but lower active-user breadth.

Identify broadly used utility behavior

Look at the Utility Features quadrant. These events have high breadth but lower frequency.

Find lower-engagement events

Look at the Underperforming Features quadrant. These events have both lower breadth and lower frequency relative to the selected events.

Compare engagement over time

Enable Compare and select two relative periods. Use the movement between the two points for each event to understand how its engagement changed between the periods.

Important behavior to remember

  • Each selected event appears as one point when Compare is disabled.
  • Each selected event appears as two points when Compare is enabled.
  • The matrix divider values are based on the selected Median or Average setting.
  • Deselecting an event recalculates the divider values and updates the quadrants.
  • Changing granularity changes the active-user metric from DAU to WAU to MAU.
  • Incomplete periods are excluded from calculations.
  • The default Top Events selection contains the top 20 events based on overall event usage within the selected time range.
  • Users can replace the default events with their own event selection.
  • Event-level filters support Event Properties and User Properties.
  • Analysis-level filters support All Properties, Event Properties, User Properties, and Cohorts.
  • The matrix table can be exported as CSV.
  • The chart can be saved and pinned to dashboards.

In summary

Engagement Matrix provides a single view for comparing feature engagement across two dimensions: Breadth - how many active users use a feature. Depth - how frequently those users use it. By plotting selected events against these dimensions and dividing the matrix using Median or Average values, Engagement Matrix organizes events into four engagement patterns:
  • Core Features - high breadth, high frequency
  • Power Features - low breadth, high frequency
  • Utility Features - high breadth, low frequency
  • Underperforming Features - low breadth, low frequency
You can customize the events, apply filters, choose a time range and granularity, compare two periods, inspect the supporting table, export the results, and save the chart to dashboards.