Overview
Working with Dashboards and Visualizations is worth 16% of the exam, second only to query execution, and it has seven objectives. You build AI/BI Dashboards with multiple pages, several datasets and widgets that include charts, text and images. You create visualizations directly in notebooks and the SQL editor. You define, configure and test parameters in SQL queries and in dashboards using named parameter markers such as :region.
The sharing objective has three parts: granting permissions to workspace users and groups, sharing with external users through shareable links, and embedding dashboards in external applications. You also schedule automatic refreshes, configure an alert with a condition, threshold and notification destination, and choose the visualization type that communicates an insight clearly. AI/BI Dashboards are the current product, replacing the legacy dashboards that older material still describes.
How it's tested
Questions describe a reporting requirement or a sharing constraint and ask which dashboard feature, permission setting, schedule or chart fits. Some ask what a viewer experiences, for example which data they see when credentials are embedded. Read who the audience is and where they sit, inside or outside the workspace.
-
Share it right
An audience such as workspace groups, account users without workspace access or external customers is described, and you choose the sharing or embedding method.
-
Parameter or filter
A requirement to change query logic or narrow displayed results is given, and you choose between a parameter and a field filter.
-
Pick the chart
A message such as a trend, a comparison, a distribution or a single KPI is described, and you choose the visualization type.
A sales manager wants a weekly email of the regional dashboard, and recipients without data access should still see figures. Sharing the editable draft would require each person to have table access. Publishing with embedded credentials runs queries as the publisher, and a schedule with subscribers sends a snapshot each week. That combination meets both needs, while individual credentials would show recipients empty or failing widgets.
Objectives 7
Official objectives from the Databricks outline, in outline order.
6.1 Build dashboards using AI/BI Dashboards, including multi-tabs/page layouts, multiple data sources/datasets, and widgets (visualizations, text, images).Building multi-page dashboards from datasets and widgets.
- What it means
- AI/BI Dashboards separate data from layout. On the Data tab you define datasets from tables, views, metric views or SQL queries. On the canvas you add pages and place widgets: visualizations bound to a dataset, text boxes with markdown, images, and filter widgets. Genie can be enabled so viewers ask follow-up questions.
- What to understand
- A dashboard can have many datasets and pages, and one dataset can feed several widgets, which avoids running duplicate queries. Visualizations are configured from dataset fields with aggregations and can be drafted with AI assistance. Filters apply to fields across datasets on a page or globally. Draft and published versions are separate: editors work on the draft, then publish so viewers see a stable version.
- How it shows up
- Questions ask how to add a second data source, reuse a dataset across widgets or organise content into pages.
- Common confusion
- Thinking each widget needs its own query, as in legacy dashboards.
6.2 Create visualizations in notebooks and the SQL editor.Charts from query results outside dashboards.
- What it means
- After running a query in a notebook cell or the SQL editor, you can add a visualization to the results, choosing a chart type and configuring axes, grouping and aggregation. Results can also be profiled to show summary statistics.
- What to understand
- These charts are useful for exploration and for documenting analysis next to the code. They are tied to the query result, so rerunning the cell or query refreshes them. Notebook charts can be added to notebook dashboards, while AI/BI Dashboards are the tool for shared, published reporting. Data profiles give quick counts, missing values and distributions.
- How it shows up
- A scenario describes exploratory analysis and asks where to create a quick chart without building a dashboard.
- Common confusion
- Assuming visualizations can only be created inside AI/BI Dashboards.
6.3 Work with parameters in SQL queries and dashboards, including defining, configuring, and testing parameters.Named parameters and how they differ from filters.
- What it means
- A parameter is a placeholder in SQL, written as a named marker like :start_date, whose value is supplied at run time. In the SQL editor a widget appears for it; in dashboards a parameter widget lets viewers set values that are substituted into the dataset query.
- What to understand
- Parameters change the query itself, so they can alter calculations, limits or date windows before aggregation, while filters narrow results already returned by a dataset. Parameters have a type, such as string, number, date or date range, a default value, and can offer single or multiple selections. Test parameters in the SQL editor before adding them to a dashboard. Using parameter markers rather than string concatenation also prevents SQL injection.
- How it shows up
- A requirement such as letting viewers choose the comparison window points to a parameter rather than a filter.
- Common confusion
- Using a field filter when the value must change the SQL logic.
6.4 Configure permissions through the UI to share dashboards with workspace users/groups, external users through shareable links, and embed dashboards in external apps.Sharing, publishing credentials and embedding.
- What it means
- Dashboard permissions are Can View, Can Run, Can Edit and Can Manage, granted to users and groups through the Share dialog. Published dashboards can embed the publisher's credentials or require viewers' own credentials. Dashboards can be shared with account users outside the workspace, through shareable links, and embedded in external websites or applications.
- What to understand
- With embedded credentials, viewers see data through the publisher's access and do not need table permissions, but they see the same data. With individual credentials, each viewer needs data and warehouse access, and row filters apply to them personally. External embedding uses a service principal and tokens so customers who are not Databricks users can view dashboards, with row-level restrictions passed by the host app. Draft access and published access are managed separately.
- How it shows up
- An audience is described and you choose the permission, credential mode or embedding approach that shows the right data.
- Common confusion
- Assuming viewers of an embedded-credentials dashboard need their own SELECT grants.
6.5 Schedule an automatic dashboard refresh.Refresh schedules and subscriptions.
- What it means
- A published dashboard can have one or more schedules that rerun its datasets at set intervals and update cached results. Subscribers to a schedule receive a snapshot by email or other destinations after each run.
- What to understand
- Schedules run with the credentials of the schedule owner, so data access is evaluated for that identity. They keep results warm so viewers load fast, but each run uses warehouse compute, so align frequency with how often data changes. Subscribers can be workspace users or notification destinations. A dashboard must be published before schedules apply.
- How it shows up
- A team wants a dashboard refreshed and emailed every morning; the answer is a schedule with subscribers.
- Common confusion
- Expecting a schedule on the draft to update what viewers see.
6.6 Configure an alert with a desired threshold and destination.Databricks SQL alerts on query results.
- What it means
- An alert runs a SQL query on a schedule and checks a condition on a result column, such as a value above or below a threshold. When the condition is met, it sends notifications to recipients or to notification destinations like email, Slack, Microsoft Teams, PagerDuty or webhooks.
- What to understand
- Configuration covers the query, the column and aggregation to evaluate, the operator and threshold, the schedule, and the notify behaviour, such as only when the state changes or on every evaluation. Alert states move between OK, triggered and unknown. Notification destinations are configured by admins and reused. Custom templates can include the value that fired the alert.
- How it shows up
- A monitoring requirement such as notify Slack when daily refunds exceed 500 asks you to set threshold, schedule and destination.
- Common confusion
- Expecting an alert to fire without a schedule; it only evaluates when it runs.
6.7 Identify the effective visualization type to communicate insights clearly.Matching chart type to the message.
- What it means
- Choosing a visualization means matching the message to the chart: line charts for trends over time, bar charts for comparing categories, scatter plots for relationships, histograms for distributions, counters for single KPIs, tables for detail, maps for geography, and heatmaps or pivots for two dimensions.
- What to understand
- Pie charts work only for a few parts of a whole; with many categories a sorted bar chart is clearer. Stacked bars show composition but make inner segments hard to compare. Dual axes can mislead unless clearly labelled. Counters can compare a value to a target. Keep consistent colours across widgets for the same category, and sort bars by value unless order is meaningful, such as months.
- How it shows up
- A communication goal is described and you choose the clearest chart type.
- Common confusion
- Choosing a pie chart for twelve or more categories.
High-yield concepts
The terms, rules and formulas this domain turns into questions most often.
-
AI/BI Dashboard
Current Databricks dashboard product.
- Remember
- Data tab for datasets, canvas for widgets.
- Tested as
- Building questions.
- Trap
- Answering with legacy dashboard concepts.
-
Dataset
Query or table feeding dashboard widgets.
- Remember
- One dataset can serve many widgets.
- Tested as
- Dashboard design.
- Trap
- Duplicating queries per widget.
-
Draft and published versions
Editors change the draft; viewers see the published one.
- Remember
- Republish to show changes.
- Tested as
- Sharing questions.
- Trap
- Expecting edits to appear immediately.
-
Embedded credentials
Published dashboard runs as the publisher.
- Remember
- Viewers need no table access.
- Tested as
- Data visibility questions.
- Trap
- Expecting per-viewer row filters to apply.
-
Parameter
Named marker like :region substituted into SQL.
- Remember
- Changes query logic.
- Tested as
- Parameter versus filter.
- Trap
- Using a filter instead.
-
Filter
Narrows results of a dataset by field values.
- Remember
- Applies after the dataset query.
- Tested as
- Interactivity questions.
- Trap
- Expecting it to change calculations.
-
Can Run permission
View and refresh with own credentials.
- Remember
- Between Can View and Can Edit.
- Tested as
- Permission level questions.
- Trap
- Assuming it allows editing.
-
External embedding
Showing dashboards in apps for non-Databricks users.
- Remember
- Uses a service principal and tokens.
- Tested as
- Customer-facing scenarios.
- Trap
- Requiring each customer to have a Databricks account.
-
Schedule and subscription
Periodic refresh plus delivery of snapshots.
- Remember
- Runs as the schedule owner.
- Tested as
- Automation questions.
- Trap
- Scheduling a draft.
-
SQL alert
Scheduled query with condition and notification.
- Remember
- Threshold, operator, destination.
- Tested as
- Monitoring questions.
- Trap
- Forgetting the schedule.
-
Notification destination
Admin-defined target such as Slack or a webhook.
- Remember
- Reused across alerts and schedules.
- Tested as
- Delivery questions.
- Trap
- Assuming only email exists.
-
Counter visualization
Displays a single KPI value.
- Remember
- Can compare against a target.
- Tested as
- Chart choice.
- Trap
- Using a bar chart for one number.
Side by side
Concepts candidates mix up, compared on the point that decides the answer.
Parameters versus filters
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Parameter | Value substituted into the dataset SQL. | Changes what the query computes. | Used only to hide rows. |
| Field filter | Narrows dataset results by values. | Query logic unchanged. | Used to change a calculation window. |
| Date range parameter | Start and end values in SQL. | Filters before aggregation. | Replaced by a filter that cannot change aggregates. |
Ways to share a dashboard
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Workspace users and groups | Share dialog permissions. | Can View, Can Run, Can Edit, Can Manage. | Granting Can Edit for viewing. |
| Account users and shareable links | Viewers outside the workspace. | Rely on published, embedded credentials. | Expecting workspace access to be needed. |
| Embedding in external apps | Dashboard inside another website or app. | Service principal and tokens for external users. | Assuming every viewer needs a Databricks login. |
| Schedule subscription | Snapshot delivered after refresh. | Push, not interactive. | Expecting recipients to drill into data. |
Choosing a chart
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Line chart | Values over time. | Shows trend and seasonality. | Used for unordered categories. |
| Bar chart | Compare categories. | Easy ranking when sorted. | Replaced by pie for many categories. |
| Histogram | Distribution of one numeric variable. | Shows spread and skew. | Confused with a bar chart of categories. |
| Scatter plot | Relationship between two measures. | Reveals correlation and outliers. | Used for time trends. |
Exam traps
Each trap is an answer that looks right. Learn the clue that should change your mind.
-
Filter for logic changes
- What candidates pick
- Add a filter so viewers choose the lookback window.
- Why it looks right
- Filters are the obvious interactivity tool.
- Why it's wrong
- A filter only narrows rows already returned; changing how far back the SQL aggregates requires a parameter in the dataset query.
Clue to watch for The value must change a calculation.
-
Viewer permissions with embedded credentials
- What candidates pick
- Grant every viewer SELECT on the tables.
- Why it looks right
- Data access usually needs grants.
- Why it's wrong
- When published with embedded credentials, queries run as the publisher, so viewers need only dashboard access.
Clue to watch for Viewers lack table access but must see figures.
-
Per-user rows with embedded credentials
- What candidates pick
- Rely on row filters while embedding credentials.
- Why it looks right
- Row filters protect data.
- Why it's wrong
- Row filters evaluate the publisher's identity, so every viewer sees the publisher's rows; per-user restriction needs individual credentials.
Clue to watch for Each manager should see only their region.
-
Pie for many categories
- What candidates pick
- Pie chart for revenue across twenty products.
- Why it looks right
- Pie charts show share.
- Why it's wrong
- Many thin slices are impossible to compare; a sorted bar chart communicates ranking and share more clearly.
Clue to watch for More than a handful of categories.
-
Scheduling the draft
- What candidates pick
- Set a refresh on the draft dashboard.
- Why it looks right
- The draft is where you edit.
- Why it's wrong
- Schedules and subscriptions apply to the published dashboard, which is what viewers open.
Clue to watch for Viewers still see stale data.
-
Alert without destination
- What candidates pick
- Create the alert and expect a Slack message.
- Why it looks right
- Slack is a common target.
- Why it's wrong
- Slack, Teams or webhooks must be set up as notification destinations and selected on the alert.
Clue to watch for Message must reach a channel.
Answer framework
Follow these steps for dashboard building, sharing, automation and chart questions.
- Identify the task type: building content, adding interactivity, sharing, automating refresh or alerting, or choosing a chart.
- For interactivity, use a parameter when the value changes SQL logic and a filter when it narrows returned rows. Parameters appear in the dataset query with a colon prefix, such as :region.
- For sharing, decide who the audience is, then choose permissions, embedded or individual credentials, account sharing or external embedding.
- For automation, publish first, then add schedules with subscribers, or create an alert with threshold, schedule and destination.
- For charts, name the message as trend, comparison, distribution, relationship, KPI or composition and pick the matching type. A line chart suits trends over time, and a bar chart suits comparisons across categories.
Practice question
Want more like this? The Databricks DA Assoc practice test has questions for every domain.
Checklist
0 of 11
Tick each one you can do without notes. Your ticks stay in this browser.
Study strategy
Build one real dashboard from start to finish, because this section is about product behaviour. Use the samples catalog or your own uploaded data in a workspace with a SQL warehouse.
In the first session, create an AI/BI Dashboard with two pages and three datasets, one from a table, one from a SQL query and one from a metric view if available. Add a line chart, a sorted bar chart, a counter, a text box and an image, plus a filter widget. Then add a date range parameter to one dataset and notice how it differs from the filter.
In the second session, publish twice, once with embedded credentials and once without, and open each as another user if you can. Review the Share dialog permissions and read the documentation on sharing with account users and embedding for external users. Add a schedule with yourself as subscriber.
In the third session, write a query that returns a daily metric, create an alert with a threshold and an email destination, and trigger it by lowering the threshold. Finish by reviewing chart-choice rules and practising a dozen scenario questions on sharing and credentials, the area where most marks are lost.
FAQ
How heavily is Working with Dashboards and Visualizations in Databricks weighted?
This section is 16% of the exam, about seven of the 45 scored questions. It is the second-largest section after query execution.
Are legacy dashboards still tested?
The objectives name AI/BI Dashboards, which have replaced legacy dashboards. Study the current product's datasets, canvas, publishing and sharing model.
What is the difference between Can View and Can Run?
Can View lets a user see the published dashboard. Can Run also lets them rerun its queries, using their own credentials when credentials are not embedded.
Can people without a Databricks account see a dashboard?
Yes, through external embedding, where the host application authenticates with a service principal and passes viewer context. Account users outside the workspace can also be given access.
Which notification destinations can alerts use?
Email plus admin-configured destinations such as Slack, Microsoft Teams, PagerDuty and generic webhooks.
How do I add parameters to a dashboard?
Add a named parameter marker such as :region to a dataset query, then add a parameter widget on the canvas bound to it.
Can Genie be used inside a dashboard?
Yes. An AI/BI Dashboard can have Genie enabled so viewers ask follow-up questions about the dashboard's data.
Other Data & Analytics certifications
Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.
Test your Domain 6 knowledge
Check your understanding of AI/BI Dashboard sharing, credentials, parameters, schedules, alerts and chart choice with exam-style practice questions.
Keep going
- Training Training Built on your Weak areas
- Practice test Questions by domain, with explanations
- Mock exam Full length, timed like the real exam
- Readiness test Find your weakest domain in minutes
- Cheat sheet Key terms and formulas on one page
- Ask the Databricks DA Assoc AI tutor Get any concept explained your way