Choose the right measurement in Metric Explorer

Measured as sets how Amplitude counts a metric in Metric Explorer. Each metric has its own measurement, so one chart can show unique users of one event, totals of another, and revenue from a third. To change a metric's measurement, open Measured as on that metric and pick an option. A new metric starts with the measurement of the metric above it.

Find the measurement for your question

Event measurements

Event measurements count users or occurrences of the metric's event in each interval.

• Unique Users: The number of distinct users in the segment who triggered the event. A user who triggers the event several times in one interval counts once in that interval.
• Event Totals: The number of times users triggered the event.
• % of Active Users: The users who triggered the event, divided by the users who triggered any active event in the same interval and segment. Measuring an inactive event can return more than 100%, because more users can trigger an inactive event than any active event.
• Avg Event per User: Event totals divided by the users who triggered the event. Users who didn't trigger the event don't count toward the average. For example, if one user triggers the event once and another triggers it twice, the average is 1.5.
• Frequency: A histogram that groups users into buckets by how many times each triggered the event during the date range.
• Frequency Percentile: The number of times the user at a percentile you choose triggered the event. For example, the 90th percentile shows how often your most active 10% of users perform the action.

Count events or cart items

When an event metric groups by a cart property, such as products.category, a single event can contain several items. Choose how Amplitude counts:

• Counting Event: Counts each event once per property value, even if it contains several items with that value.
• Counting Item: Counts every item in the cart.

For example, a Checkout event with two items in the category tacos counts as 1 with Counting Event and 2 with Counting Item. To count the items in a cart property, such as products, instead of the events that contain them, measure the event as Semantic Totals.

Property measurements

Property measurements aggregate a property on the metric's events. Pick the property after you choose the measurement. Sum, average, median, minimum, maximum, and percentile need a numeric property.

• Sum of Property Value: The total of the property across events, for example order value.
• Average of Property Value: The sum of the property divided by the number of events with the property.
• Median of Property Value: The middle value. Use it instead of the average when a few large values skew results.
• Min property value and Max property value: The smallest and largest values in each interval.
• Percentile of Property Value: The value at a percentile you choose, such as the 90th percentile of page load time.
• Distinct Property Values: The number of different values of the property.
• Avg distinct property values per user: The number of different property values each user triggered, on average.
• Distribution of Property Value: A histogram of event totals by property value. Each bucket includes its minimum and excludes its maximum.

Revenue measurements

Revenue measurements are under More in Measured as. Pick the revenue property on the event, such as $revenue.

• Revenue: The total of the revenue property. Amplitude formats it as currency in your project's currency.
• Average Revenue per Active User: Revenue divided by the users who triggered any active event in the same interval and segment.

Session measurements

Session measurements count the sessions that contain the metric's event. Measure [Amplitude] Any Active Event to count all sessions. For how sessions work and how to filter them, go to Analyze sessions.

• Total Sessions: The number of valid sessions in each interval.
• Average Sessions per User: Total sessions divided by the users active in the interval.
• Total Session Duration: The sum of the lengths of all sessions in the interval.
• Average Session Duration: Total session duration divided by total sessions.
• Time Spent per User: Total session duration divided by the users active in the interval.
• Distribution of Session Length: A histogram of sessions by length.
• Events per Session: The average number of times the metric's event occurs per session.

For duration measurements, the Counting in control sets the unit, from seconds to months. Auto picks a unit that fits the values.

Distribution views

Frequency, Distribution of Property Value, and Distribution of Session Length render as histograms, so a chart that uses one of them can't include other metric types. When every metric on the chart is Distribution of Property Value or Distribution of Session Length, you can switch between two views:

• PDF: Shows the count or share of values in each bucket. Use it to compare the shape of distributions across segments of different sizes.
• CDF: Shows the count or share of values at or below each bucket. Use it to read approximate percentiles. For example, the 75th percentile is the smallest bucket where the CDF reaches 75%.