Insights

An insight charts one or more metrics from the markers your experience sends: how many players logged in each day, how many purchases they made, how much they spent, how many players are online right now. It's the default query type in the Query Builder and the one most dashboard charts use.

Every metric answers two questions: what to count (an event, or a stat like concurrent users) and how to measure it (total events, unique users, the sum of a property, and so on). Add up to 15 metrics to one insight to compare them, and combine them with formulas for ratios and rates.

Building an insightCopy Anchor Link

1. Open the builder. On a dashboard, click Edit, then Add panel → Query. Insight is selected at the top of the query setup.
2. Add a metric. Click the + beside Metrics and pick an event, such as session_start.
3. Choose how to measure it. Click the measurement under the metric (it starts as Total Events) and pick another, such as Unique Users or Aggregate Property → Sum of price.
4. Narrow it down (optional). Add a filter to one metric or to all of them, and a breakdown to split the results, for example by country.
5. Pick a time range and chart type above the chart.
6. Save. Give it a name and click Create. See Saving a query.

MetricsCopy Anchor Link

Click the + beside Metrics to add one. The picker has three tabs:

Each metric gets a letter badge (A, B, C …) that formulas use to refer to it.

Working with metricsCopy Anchor Link

Hover a metric for quick actions, or open its ⋯ menu (or right-click it) for everything:

• Duplicate: copy the metric with its measurement and filters. A quick way to chart the same event two ways, for example with and without a filter.
• Filter: add a filter that applies to this metric only. See Local filters.
• Hide metric / Show metric: take the metric off the chart without removing it. Hidden metrics are still calculated, so a formula can still use them.
• Edit display name: rename the metric on the chart, legend and table. The original event name stays in the tooltip.
• Remove: delete the metric.

Drag a metric by its handle to reorder it. The order sets the legend and series order, and formulas follow their metrics when you reorder.

MeasurementsCopy Anchor Link

The measurement decides what number a metric produces. Click it under the metric to change it.

Every measurement is worked out separately for each point on the chart: each day (or hour, or week) on a time chart, and each breakdown value.

A worked exampleCopy Anchor Link

Three players, and the purchase events they sent in one day. Each event has a price argument.

For the purchase event, that day:

Frequencies and per-player values only include players (and sessions) that sent the event. Cleo is in none of them. To compare against every active player, divide by a separate Unique Users metric of session_start with a formula.

Unique Users and active usersCopy Anchor Link

Click the gear beside Unique Users to choose the window players are counted over:

• Interval (Default): unique players in each point on the chart. On a daily chart that's each day; on a weekly chart, each week.
• Daily Active Users, Weekly Active Users, Monthly Active Users: unique players over the last 1, 7 or 30 days, rolling. A daily chart of Monthly Active Users shows, for each day, the players active in the 30 days up to that day. On a chart coarser than the window (DAU on a weekly chart), the values are averaged over each point.

Use Unique Users of session_start (or any event every player sends) for active players, and pair DAU and MAU in a formula for stickiness.

FrequenciesCopy Anchor Link

Frequency per User and Frequency per Session count events for each player or session, then summarize those counts with Average, Median, Minimum, Maximum, or a Percentile (25th, 75th, 90th or 99th). Use them for questions like "how many rounds does a typical player play a day?" (median) or "how many do our heaviest players play?" (90th percentile).

Aggregating a propertyCopy Anchor Link

Aggregate Property works on a value carried by the event, such as a price, amount or duration argument. Pick a function, then the property:

• Sum, Average, Median, Minimum, Maximum, Percentile (25th, 75th, 90th, 99th), First Value, Last Value: for numeric properties.
• Distinct Count: the number of different values, for any property. For example, how many different items were bought.

Aggregate Property per User and per Session aggregate twice. First each player's (or session's) values are combined with Sum, Average, Distinct Count, Minimum, Maximum, First Value or Last Value. Then those per-player results are combined with the outer function. The chart shows it as, for example, Average of (Sum price per User): the average total spend per paying player.

Cumulative SumCopy Anchor Link

Open Advanced in the measurement menu and choose Cumulative Sum to turn a metric into a running total: each point is the sum of every point up to and including it. Use it for growth curves like total revenue or all-time sign-ups. A Cumulative Sum chip shows on the metric; click its × to turn it off.

Cumulative sums apply to time charts (Line, Stacked Line, Column, Stacked Column and Table). Bar, Pie and Metric charts show the total over the range instead.

Chart typesCopy Anchor Link

Pick the chart from the controls above the chart.

Line, Stacked Line, Column, Stacked Column and Table are time charts: they show a value for each point in time, at the granularity you pick. Bar, Stacked Bar, Pie, Metric and World Map measure the whole range at once. A whole-range Unique Users counts each player once for the range, so it isn't the sum of the daily values.

Chart optionsCopy Anchor Link

• Show or hide a series by clicking it in the legend.
• Zoom into a time chart by dragging across it. Click Reset Zoom to go back.
• Annotations: on Line and Stacked Line charts, hover the chart and click the button in its top-right corner to switch from zooming to Create annotation. Then drag across the time the annotation covers, such as a release or a live event, and save it. Annotations belong to the project, so they show on every Line and Stacked Line insight that covers that time, along with your jobs.
• In the ⚙ tab at the top of the query setup, Invert colors order reverses the series colors, and Chart configuration sets a fixed Min and Max for the value axis.

Units and formattingCopy Anchor Link

Click the badge to the left of a metric to choose how its values are shown. Formatting only changes how numbers look, never what's calculated.

• Number, Percentage (adds %), time units from Milliseconds to Years, currencies (Robux (R$), USD ($), Euro (€), Pounds (£)), or Custom with your own prefix and suffix.
• Compact abbreviates large numbers (1,200 shows as 1.2K). On by default.
• Decimals sets the number of decimal places, from 0 to 9.

Time units say what the value is already in, not what to convert it to. If duration is sent in seconds, choose Seconds: Gamebeast shows longer values in minutes or hours on its own.

Counts have a fixed unit (Unique Users counts users, Total Events counts events, CCUs count users), so the unit is set for you and the badge shows it. You choose the unit when you aggregate a property, and for formulas.

Reading the resultsCopy Anchor Link

While you're building, a table under the chart lists each series with its Average, Min, Max and Sum over the range. Hover any point on the chart for its exact value.

With a breakdown, each breakdown value is its own series. Values beyond the breakdown's limit are grouped into an Other series.

Experiment coverage. When you filter or break down by an experiment, a note under the builder says if some markers in the range are missing experiment data. Those markers are left out of every experiment series, so treat results with low coverage carefully.

No data. If the chart says No data found, check the time range, then remove filters one at a time to find the one that excludes everything.

RecipesCopy Anchor Link

Event and argument names here are examples. Use the markers your experience sends.

With the assistantCopy Anchor Link

Describe the chart you want to the assistant, such as "daily revenue from new players over the last month, by platform", and it builds the insight, runs it, and lets you open it in the builder or add it to a dashboard. Agents connected over MCP can build and save insights the same way.