Understand your Event Segmentation analysis

An Event Segmentation analysis shows how often actors perform events and how those results differ across groups of actors. Each module in the template controls one part of the question the analysis answers. The template consists of the following modules:

Choose the measure that the analysis calculates for the selected events and segments. The measure decides what the chart reports, such as the total number of events or the number of distinct actors.

An Event Segmentation analysis measures the following:

To see a measure change over time, click the not over time drop-down list and choose Over time. Results over time let you set up period-over-period comparison for each measure.

A metric is a custom measure derived from the same event dataset that the analysis uses, like ProductInteractions. The metric aggregates data from that dataset.

For example, a Freemium to Enterprise Ratio metric divides Freemium user events by Enterprise user events. In an Event Segmentation analysis, this metric shows a typical ratio of four to six freemium events per enterprise event. For Add Data Source events, the ratio drops to one to two. The lower ratio indicates that enterprise users rely on that feature more than freemium users.

The Uniques, Frequency, Intervals Engaged, Active %, and Average Events measures count actors instead of events. These measures add a catalog picker where you choose the actor to count, such as Users. The Active % measure requires an active event annotation on your event stream.

Create a custom measure when the prebuilt measures in Analytics do not answer your question. A custom measure combines events from more than one event stream in a single formula. Add a formula block, and then choose it as the measure in the Measure module.

The following example calculates the number of purchases for each user who viewed a product details page (PDP).

Choose one or more events to analyze. The events you choose define the behavior that the analysis measures.

The event picker displays event streams and their event types. In Analytics, event streams are datasets configured to model event flows. Not every dataset is an event stream. For example, the Users dataset lists users instead of events.

Hover over an event to see its usage stats, including the average events received per day. Use the stats to judge whether an event still receives data. See Event taxonomy.

Choose which occurrences of an event count toward the Uniques measure to separate new behavior from repeat behavior. For example, count only users who played content for the first time ever. The chart then shows new adopters instead of returning users.

Each event in the Events module has its own occurrence setting. The setting is available only when the measure is Uniques.

A non-default option displays its name as a label on the event. The label tells readers that the analysis does not count every occurrence. Click the label to change the option.

Compare groups of actors, such as users, to see how the measure differs between them. Add cohorts in the Segment by Cohort section. Add properties in the Breakdown by property section. Choose an existing cohort from the drop-down list. To define a behavioral cohort block in one click, click New Cohort.

Narrow the data in a visualization to answer a more specific question. For example, define a region filter to see results for one region only.

The Filters module also lets you choose JSON columns. When you click a JSON column, it expands to display its keys. Choose a key and click Apply. The selected key becomes the display name for that column.

Present the analysis results in the format that best answers your question. In this module, edit the visualization, examine the underlying Structured Query Language (SQL) query, and add charts to a dashboard.

Six visualization types are available: Line Chart, Bar Chart, Pie Chart, Horizontal Bar Chart, Single Value, and Table. A Pie Chart shows the distribution of events by segmentation attribute across the selected time range.

The following features are available in this module:

Switch the y-axis between absolute numbers and percentages in one click to compare segments of different sizes. Absolute numbers give a clearer breakdown of how many actors are in each segment. Click # for numbers or % for percentages.

Limit the chart to the segments with the most activity so the visualization stays readable. Click the Top drop-down list, choose or enter a number, and click Apply. The total count of selected events decides which attribute values rank highest. Beelines also let you set the number of top segments.

Configure the time range and time grain to control which events the analysis includes and how it groups them. The time range is the complete period of events that the analysis includes. For example, use the last two years or the dates between two specific days. The time grain is the granularity of the analysis, such as Daily (one day) or Weekly (seven days).

Set the time range from the drop-down list, or choose a quick option. Try different time ranges and time grains without leaving the chart. To set a lag, click Offset, and then set Ending.

Choose a sampling mode to balance query speed against result precision. Click the sampling control in the visualization toolbar. Its tooltip shows the current sampling mode. The Sampling Mode drop-down list offers three modes: Disabled, Enabled - Faster Response, and Enabled - Higher Precision.

Keep a series in the same color across charts so that dashboards are easier to read. Override the default color and display name for individual data series from the chart legend. See Override the palette for an exploration.