Using Dashboards

This guide is for BI analysts and business analysts. It explains how to save Analytics Agent Q&A results into dashboards and maintain sustainable business analysis assets. Ordinary business users primarily view dashboards, ask follow-up questions around them, and submit feedback — they do not need to directly maintain dashboard configurations.

Dashboards are not a daily work object for administrators. Administrators are responsible for users, roles, permissions, audits, and governance; BI analysts are responsible for organizing business questions, metric definitions, and charts into dashboards that can be continuously viewed by the team.

What Problems Dashboards Solve

Dashboards are suitable for locking in business questions that need to be viewed repeatedly, such as:

  • Daily business overview.
  • Sales funnel monitoring.
  • User activity and churn monitoring.
  • Financial collections monitoring.
  • Operations campaign effectiveness review.
  • Account health overview.

One-time questions can be completed directly in Q&A; high-value, periodic, team-shared questions are suitable for consolidation into dashboards.

Role Responsibilities

BI Analysts / Business Analysts

Responsible for:

  • Designing dashboard themes.
  • Selecting valuable charts and tables from Q&A results.
  • Saving charts/tables to dashboards.
  • Maintaining dashboard names, layouts, and sharing status.
  • Explaining metric changes.
  • Adjusting questions, charts, and metric definitions based on feedback.

Business Users

Responsible for:

  • Viewing shared dashboards.
  • Making business judgments around dashboards.
  • Submitting feedback on inaccurate or hard-to-understand results.
  • Raising new analysis needs with BI analysts.

Administrators

Responsible for:

  • Configuring users, roles, and permissions.
  • Managing analytics domain access scope.
  • Reviewing audit logs.
  • Handling governance boundary issues.

Administrators should not take on daily dashboard content development work.

Saving Charts/Tables from Q&A to a Dashboard

In practical validation, both charts and tables from Analytics Agent Q&A results can be saved directly to a dashboard.

Process:

  1. Ask a question on the analysis page.
  2. Wait for the system to generate a table or chart.
  3. Click the "Save to Dashboard" icon in the upper right corner of the chart or table.
  4. In the popup, choose to create a new dashboard or add to an existing one.
  5. Enter the dashboard name when creating a new one.
  6. Select visibility: visible only to me, or shared.
  7. Click confirm.

After saving successfully, the system shows a save success notification. Go to the dashboard list to see the new dashboard.

In practice, a dashboard saved from a Q&A chart shows:

  • 1 chart.
  • Associated domain as the current analytics domain.
  • Update time as the time of saving.

This confirms that dashboard components inherit the Q&A result and analytics domain context.

How to Name a Dashboard When Creating One

Dashboard names should reflect the business theme, not a one-time question.

Not recommended:

Test Q&A chart dashboard

Recommended:

Account Health Monitoring Dashboard

Sales Conversion Analysis Dashboard

Financial Collections Tracking Dashboard

Naming recommendations:

  • Include the business object.
  • Include the analysis purpose.
  • Avoid temporary, test, or personal naming.
  • If it is a topic-specific dashboard, include the topic name.

Visibility Selection

When saving to a dashboard, you can choose:

  • Visible only to me.
  • Shared.

Recommendations:

  • Use "visible only to me" during the draft phase.
  • Share only after metric definitions are confirmed.
  • Do not arbitrarily share dashboards containing sensitive data.
  • Before sharing, confirm that analytics domain permissions and row-level permissions align with business boundaries.

Sharing is not a substitute for permission governance. Whether a user can view dashboard content should still be constrained by analytics domain, resource permissions, row-level permissions, and column hiding configurations.

Dashboard List

The dashboard entry is in the "Dashboards" menu in the left navigation.

In practice, the dashboard list contains:

  • My Dashboards.
  • Shared Dashboards.
  • Create New Dashboard.

Visible fields in My Dashboards include:

  • Name.
  • Number of charts.
  • Associated domain.
  • Update time.

Shared Dashboards show the creator, so business users can know the source of the dashboard.

Row actions include:

  • Share or unshare.
  • Delete.

Delete is a destructive operation and should be used with caution.

Dashboard Detail Page

After entering the dashboard detail page, you can see:

  • Dashboard name.
  • ASK AI.
  • Version history.
  • Chart titles.
  • Chart update times.
  • Next update time hints.

In practice, charts saved from Q&A showed an update time, and indicated the next update would be approximately 24 hours later. This shows dashboards are not simple screenshots but analysis components that can continuously refresh data.

Chart Operations

Charts in a dashboard support full-screen viewing.

Practical notes:

  • Clicking the full-screen icon on a chart enters full-screen mode.
  • After entering full-screen, click the close button in the upper right corner to exit.
  • Pressing Esc does not close a full-screen chart.

This should be noted in user training to prevent users from not knowing how to return after entering full-screen.

ASK AI

The dashboard detail page has an ASK AI button.

In practice, clicking it opens a question input area on the right side of the page associated with the current dashboard, showing:

Associated dashboard: dashboard name

ASK AI also shows popular question searches and question history.

This means BI analysts and business users can continue asking follow-up questions around the current dashboard. For example:

Why is the active rate lowest for Premium plan?

Show account health overview for Google source only

Break down the metrics in the current dashboard by country

Usage recommendations:

  • Viewers can continue exploring around the dashboard.
  • BI analysts can discover new dashboard components from follow-up questions.
  • If follow-up questions produce stable value, they can continue to be saved to the dashboard.

Version History

The dashboard detail page has a version history entry.

In practice, version history shows:

  • Version numbers, e.g., V1, V2.
  • Current version marker.
  • Operator.
  • Operation time.
  • Change description, e.g., dashboard layout change.

Version history answers:

  • When the dashboard was changed.
  • Who changed the dashboard.
  • Whether the current version is the latest.
  • Whether a layout change occurred.

If a business user reports "the dashboard looks different from yesterday", the BI analyst should first check the version history before judging whether the difference is a data refresh, a layout adjustment, or a chart configuration change.

Advanced Edit

The more menu in the upper right corner of the dashboard detail page has "Advanced Edit".

In practice, clicking it opens a "Configure Dashboard" dialog containing:

  • Dashboard name.
  • JSON configuration editor.
  • Cancel.
  • Confirm.

Visible fields in the JSON configuration include:

{ "charts": ["48"], "chartTabs": null, "layout": { "chartPositions": [ { "x": 0, "y": 0, "w": 6, "h": 9, "i": "48" } ] }, "globalParams": [], "crossFilters": null }

The fields can be understood as:

  • charts: Which chart components the dashboard contains.
  • layout.chartPositions: Position and size of charts in the dashboard.
  • globalParams: Global parameters.
  • crossFilters: Cross-filter configuration for interactions between charts.

Advanced edit is suitable for BI analysts familiar with the configuration structure. Ordinary business users should not modify it directly.

Usage recommendations:

  • Confirm the current version before making changes.
  • Do not save if you are unsure about the JSON meaning.
  • Test complex layout adjustments in a test dashboard first.
  • Check version history after making changes.
  • Have business users re-verify key questions after changes.

Dashboard Development Methodology

Start from Questions, Not from Charts

First clarify the questions the dashboard should answer:

Is account health normal?

Then break it down into metrics:

  • Total account count.
  • Active account count.
  • Active rate.
  • Cancellation rate.
  • Trial conversion rate.
  • Average seat count.

Then select charts:

  • Overview metric cards.
  • Bar chart comparing by plan.
  • Line chart showing trend over time.
  • Anomaly detail table.

Explore First, Then Consolidate

Recommended process:

  1. Explore using natural language on the analysis page.
  2. Review answers, charts, tables, and business definitions.
  3. When needed for sharing or decision-making, have BI analysts review SQL statements and logs.
  4. For stable metrics, use the "Exploration" page to validate baseline and comparison period.
  5. Check change rates and auto-drill-down fields to confirm which breakdown dimensions are worth displaying.
  6. Confirm reusable questions.
  7. Save charts/tables to a private dashboard.
  8. Adjust the dashboard organization.
  9. Have business stakeholders validate.
  10. Share with the team.

Do not share unvalidated one-time charts directly with the business team.

Control Dashboard Complexity

A dashboard should not try to cover all questions.

Recommendations:

  • One dashboard for one stable business topic.
  • Place the most important metrics on the first screen.
  • Do not include too many charts.
  • Keep metrics with the same definition together.
  • Split different topics into different dashboards.

When Not to Build a Dashboard

The following situations are not suitable for immediately building a dashboard:

  • Metric definitions not yet confirmed.
  • Q&A results still unstable.
  • Data source still being adjusted.
  • Only one-time analysis value.
  • Sensitive data involved but permissions not confirmed.
  • Business users do not yet have a clear use case.

Pre-Launch Checklist

Before sharing a dashboard, check:

  • Dashboard name is clear.
  • Associated domain is correct.
  • Core charts can refresh correctly.
  • Chart titles are understandable by business users.
  • Metric definitions are confirmed.
  • Important charts have been reviewed by BI analysts (SQL statements and logs).
  • Typical business questions have been validated.
  • Visibility is confirmed.
  • Sensitive fields are confirmed to not be exposed.
  • Version history is preserved.
Reading Analysis Results
Table Rendering
Singdata © 2024 Singdata, Inc. All Rights Reserved.