Explore Your Data

Use the mParticle Agent to explore configured data inputs and sources, events, user attributes, calculated attributes, aggregate event volume and trends, total user and device counts, and attribute coverage. The Agent summarizes relevant findings and helps connect them to supported audience and prediction workflows.

You can begin with an open-ended business question or even a general idea about the data you want to understand. You don’t need to know the exact names of events or attributes; just describe the behaviors or goals you want to explore in general terms.

Depending on your question and the data available in your workspace, the agent can help you explore:

• Configured data sources: Review the SDK, feed, or API inputs currently configured for your workspace.
• Events: Find events listed in the Data Catalog, their specific types, associated inputs, and custom attributes.
• User attributes: Review the user attributes listed in the Data Catalog, their types and scopes, and their data sources.
• Calculated attributes: See the different types of calculated attributes configured, the events used in their calculation, and whether they’re active.
• Aggregate data and trends: Investigate aggregate event volume, trends, total user or device counts, and event attribute coverage.

Remember that even if an event or attribute is listed in your Data Catalog, it might not have enough recent activity to be useful. Similarly, a configured data input might not be actively sending data. Data exploration with the agent helps you distinguish between data that’s merely on record, and data that’s actually useful for your goals.

Find available events and user attributes

Start by describing the user behaviors and characteristics you’re curious about. The agent can search events and attributes listed in the Data Catalog, identify event types and associated sources, review available custom event attributes, and explain user-attribute types and scopes.

For example, ask the agent:

• What commerce events do we track?
• Which user attributes could identify high-loyalty members?
• Which data sources provide this event?
• What attributes are available on this event?

Understand activity and coverage

The agent can explain trends about your data, including:

• Event volume over a selected period, including daily or hourly trends and breakdowns by data input.
• Unique user or device counts by dimensions such as time, data input, platform, channel, or country.
• Event attribute coverage, including the share of observed events where an attribute is populated.

For example, ask the agent:

• Has this event been received recently?, or
• Which data input sends this event the most?

Review calculated attributes and commerce data

The agent can explain how a calculated attribute is computed, such as whether it counts events or totals a numeric value. It can also identify any events, attributes, or time periods the calculated attribute relies on and whether the calculated attribute is active or still in a draft state.

For example, ask:

• Do we currently calculate purchase count over the last 30 days?
• What is the specific event used when calculating purchase count?
• Which data inputs contribute the most to this calculated attribute?

You can also use the agent to explore more complex commerce event types.

For example, ask Which product fields are available on purchase events? or Does our Web SDK input provide more commerce data than our mobile SDK inputs?

Refine your inquiry

Use follow-up questions to narrow your investigation by data input, environment, or date range. Activity and attribute coverage can differ across inputs and environments, even when the same event appears in the Data Catalog.

1. What commerce events and attributes are available in this workspace?
2. Which data inputs send the events you found, and are they available in Production?
3. Show me the aggregate activity for those events over the last 30 days. Are they being received consistently?
4. Which attributes are available on the events, and how often are they populated?
5. Summarize any gaps, inconsistencies, or unexpected changes in the results.

Review your findings

Before using the results to make a decision, confirm:

• Workspace and environment: The results come from the workspace and data environment you intended to examine.
• Data inputs: The query includes the SDK, feed, or API inputs relevant to your question.
• Time range: The selected period represents the activity you want to understand.
• Recent activity: The events or attributes have enough recent activity to support your conclusion.
• Coverage: The relevant attributes are populated consistently enough for your intended use.

The agent’s responses reflect the data it can access within the scope of your question. Keep these limitations in mind:

• Use aggregate results as scoped evidence: Event volume and attribute coverage reflect the selected environment and time range. They aren’t a complete historical audit, data-quality certification, or forecast of future activity.
• Keep exploration focused: The agent isn’t intended for unrestricted warehouse queries, arbitrary joins, full data lineage, or general-purpose data audits. It also doesn’t expose raw records, personal identifiers, or unrestricted warehouse rows.

For more information about the agent’s access and privacy boundaries, see Data and Privacy.

Troubleshooting data exploration issues

The agent can’t find an event or attribute

Confirm that the event or attribute exists in the workspace you’re exploring. The Agent uses workspace information and doesn’t invent an event, attribute, or value when it can’t find a match.

If you recently added the event or attribute, navigate to Data Platform > Data Catalog and confirm that it appears before trying again. For information about the available metadata, see Data Catalog.

If the missing signal is a Development user attribute sent recently, it may still be processing.

I sent user attributes using the development data environment, but the agent can’t find them

Development user attributes may appear in aggregate activity before their names become available for search and audience building. If you first sent or updated the attributes today, allow approximately one day for processing and check again the following day.

For more information, see Development user-attribute processing.

The agent can see a field but not its values or coverage

A field can be visible in the Data Catalog even when its values or coverage information aren’t available to the agent. When the agent can’t inspect the information needed to complete a criterion, provide the exact value or refine the request using information you can verify in your workspace.

Value inspection depends on your workspace configuration and permissions. The agent doesn’t infer values it can’t inspect.

A data query returns no results

Confirm that the query uses the intended environment, sources, and time range. Try a broader time range or remove an unnecessary source filter, then run the query again.

No returned rows means only that the specific query didn’t return results for its selected scope. It doesn’t prove that the workspace has no data.

Next steps

• Build Audiences: Use the data you explored to prepare a new real-time audience.
• FAQ: Find answers to common questions about the mParticle Agent.
• Data and Privacy: Understand the agent’s data access and privacy boundaries.