Associate Data Practitioner · Domain 2: Data Analysis and Presentation · 27% of the exam
Task 2.2: Visualize data and create dashboards in Looker given business requirements.
Creating, changing and sharing Looker dashboards, choosing between Looker and Data Studio (Looker Studio in the guide), and editing simple LookML parameters.
Study it
Looker dashboards: creating, changing, sharing and scheduling
Lesson coming
Looker or Data Studio (formerly Looker Studio) for an analytics use case
Lesson coming
LookML basics: views, explores, dimensions and measures
Lesson coming
Sample questions
Try each one before opening the answer. Every option is explained, with the Google Cloud documentation page that proves it.
Question 1 · choose 1
A four-person marketing team wants, by Friday, one table that lines up each campaign's planned budget from a Google Sheets file with its actual ad spend from a BigQuery table, matched on campaign ID. They build reports with drag-and-drop tools, the report will be shared view-only with an outside agency, and there is no budget for new licenses or subscriptions. What should the team use?
AA Looker user-defined dashboard built on an Explore over the BigQuery table and the sheet
BA Data Studio report on the BigQuery data source, with a calculated field that looks up each campaign's budget in the Sheets data source
CA Data Studio blend of the Sheets and BigQuery data sources that uses the union operator, shared with the agency as viewers
DA Data Studio blend of the Sheets and BigQuery data sources joined on campaign ID, shared with the agency as viewers
Show the answer and why
AA Looker user-defined dashboard built on an Explore over the BigQuery table and the sheet
Incorrect
Looker dashboards rest on Explores defined in a LookML model, which a developer must create in a Looker instance, a paid service; this does not fit a no-license, drag-and-drop team by Friday.
BA Data Studio report on the BigQuery data source, with a calculated field that looks up each campaign's budget in the Sheets data source
Incorrect
A calculated field is a formula over fields of its own data source; it cannot fetch values from a different data source, so the budget cannot be looked up this way.
CA Data Studio blend of the Sheets and BigQuery data sources that uses the union operator, shared with the agency as viewers
Incorrect
A union stacks the rows of one table under the other, so budget rows and spend rows stay on separate lines instead of being paired by campaign ID.
DA Data Studio blend of the Sheets and BigQuery data sources joined on campaign ID, shared with the agency as viewers
Correct
Data Studio is a no-cost, drag-and-drop tool that connects to both Google Sheets and BigQuery; a blend that joins the sources on campaign ID puts budget and spend on one row, and reports can be shared view-only.
Data Studio suits fast, self-service reporting at no cost: it connects to sources such as Sheets and BigQuery without programming, blends up to five data sources with a join or union operator, and shares reports with view or edit access. Joins combine records side by side on matching keys; a union appends rows. Looker is the choice when a governed LookML model and a managed instance are worth the setup and cost.
In a Looker view based on an orders table (one row per order, primary key order_id), a developer must add a field that shows how many different customers placed orders, so that a dashboard tile can show it for any date range. Which LookML should the developer add?
A count_distinct measure counts the distinct values of the expression in its sql parameter, here each customer once, whatever filters the query applies.
A count measure counts rows based on the table's primary key and does not support the sql parameter, so it cannot count distinct customers.
C`measure: buyers { type: sum sql: ${customer_id} ;; }`
Incorrect
A sum measure adds the values together; adding customer IDs produces a meaningless number, not a count of customers.
D`dimension: buyers { type: number sql: ${TABLE}.customer_id ;; }`
Incorrect
A dimension returns a value for each row and is used for grouping; it does not aggregate, so it cannot give a count across the date range.
In LookML, dimensions describe row-level attributes and measures aggregate them. type: count counts rows by the primary key and takes no sql parameter; to count distinct values of another field, such as customer_id, use type: count_distinct, which uses SQL's COUNT DISTINCT.
A Data Studio report reads the BigQuery table `ops.shipments`, about 40 million rows that a pipeline updates every 15 minutes. Operations managers notice that the charts can show figures up to 12 hours old, and they want changes to appear within about 15 minutes without anyone pressing refresh, while every chart keeps covering all of the rows. What should the owner of the report change?
AReplace the data source with an extracted data source that has Auto update turned on
BAdd the shipments table to a BigQuery BI Engine reservation as a preferred table
CSet the data freshness of the BigQuery data source to every 15 minutes
DTurn on the report's auto refresh with an interval of 15 minutes
Show the answer and why
AReplace the data source with an extracted data source that has Auto update turned on
Incorrect
Extracted data sources hold static information, updated only on their Auto update schedule, and are limited to 100 MB and 750,000 rows, so a 40-million-row table would not fit.
BAdd the shipments table to a BigQuery BI Engine reservation as a preferred table
Incorrect
BI Engine makes the queries faster, but the report still reuses earlier results until its data freshness threshold passes, so figures stay as old as before.
CSet the data freshness of the BigQuery data source to every 15 minutes
Correct
A BigQuery data source has a data freshness setting, adjustable from every 1 to 50 minutes up to every 12 hours, which decides how long the report can reuse earlier results before querying again.
DTurn on the report's auto refresh with an interval of 15 minutes
Incorrect
Auto refresh settings on a report don't affect the data source's data freshness; refreshes that come sooner than the data source's threshold are answered from memory, so the figures stay up to 12 hours old.
Most Data Studio data sources have a data freshness option that trades up-to-date figures against performance and query cost; for BigQuery it can be set from every 1 to 50 minutes up to every 12 hours. A report's auto refresh only re-requests data and is answered from memory until the data source's freshness threshold expires. Extracted data sources are static snapshots with size and row limits, updated only on an Auto update schedule.
Regional directors want the Looker "Weekly Sales" dashboard, laid out as it appears on screen, attached to an email every Monday morning without anyone sending it by hand. What should the dashboard owner set up?
AA Google Sheets delivery of the dashboard
BA recurring email schedule for the dashboard in PDF format
CA shared copy of the dashboard's browser URL
DA one-time send of the dashboard as a CSV zip file attachment
Show the answer and why
AA Google Sheets delivery of the dashboard
Incorrect
The Google Sheets destination is not supported for dashboards, because it streams the results of a single query as plain CSV.
BA recurring email schedule for the dashboard in PDF format
Correct
Looker can schedule a dashboard for recurring delivery to email, and the PDF format sends an image of the dashboard with tiles as arranged, as an attachment.
CA shared copy of the dashboard's browser URL
Incorrect
A URL shares the query for Looker users to open; it does not send the dashboard to anyone on a schedule.
DA one-time send of the dashboard as a CSV zip file attachment
Incorrect
A one-time send happens once, and a CSV zip file holds unformatted data rather than the dashboard's layout.
The Looker Scheduler delivers dashboards one time or on a recurring schedule to destinations such as email, webhooks, Amazon S3 and SFTP. For dashboards, formats include PDF and PNG images of the layout and a CSV zip file of the tile data.
A LookML model has an `orders` Explore. Users must also be able to pick fields from the `customers` view, where many orders belong to one customer and orders.customer_id matches customers.id. Which addition to the orders Explore does this?
one_to_many would say that one order has many customers, the reverse of the data, which can make Looker compute joined measures incorrectly.
C`extends: [customers]`
Incorrect
extends copies the definition of another Explore or view to build on it; it does not join a view's rows to orders.
D`sql_table_name: customers ;;`
Incorrect
sql_table_name is a view parameter that names the database table behind a view; it does not add a second view to an Explore.
An Explore is the starting point for queries, and join adds views to it. Each join needs sql_on (or foreign_key) for the ON clause, and its relationship (default many_to_one) tells Looker how rows relate so that measures are calculated correctly across the join.
In Data Studio, a data team wants one BigQuery data source with agreed field names and calculated fields that a dozen different reports will all use, with only the data team able to edit it. How should the data source be created?
AInside the first report, as an embedded data source
BAs a blended data source inside each report
CFrom the Data Studio home page, as a reusable data source
DAs an extracted data source copied into each of the reports
Show the answer and why
AInside the first report, as an embedded data source
Incorrect
A data source created while editing a report is embedded in that report, and anyone who can edit the report can edit it and its connection.
BAs a blended data source inside each report
Incorrect
A blend combines several data sources in one report; it does not give a dozen reports one shared definition.
CFrom the Data Studio home page, as a reusable data source
Correct
Data sources created from the home page are reusable across reports, let you share a consistent data model, and can be edited only by people with whom they are shared.
DAs an extracted data source copied into each of the reports
Incorrect
An extract is a static snapshot for faster performance; copying one into each report creates separate copies to maintain.
Data Studio data sources are embedded (created inside a report, shared and copied with it, editable by its editors) or reusable (created from the home page, usable by many reports, editable only by the people they are shared with). Reusable data sources keep a consistent model across an organization.
A BI governance team is choosing between Looker and Data Studio. It needs to define a role of its own, made from exactly the permissions it picks, and grant that role to a group of business users. Which statement about the two products is accurate?
AData Studio, with a custom IAM role granted on the report assets
BLooker, where admins can create custom roles from permissions
CEither product, since both offer only predefined roles
DData Studio Pro, whose team workspaces support custom roles
Show the answer and why
AData Studio, with a custom IAM role granted on the report assets
Incorrect
Access to Data Studio assets uses predefined IAM roles; Data Studio and Data Studio Pro don't support custom roles.
BLooker, where admins can create custom roles from permissions
Correct
Looker allows custom roles: an admin combines a set of permissions with the models they apply to and grants the role to users or groups.
CEither product, since both offer only predefined roles
Incorrect
Only Data Studio is limited to predefined roles; Looker lets admins create custom roles.
DData Studio Pro, whose team workspaces support custom roles
Incorrect
Team workspaces come with their own set of roles, but Data Studio Pro, like the standard edition, doesn't support custom roles.
Looker manages access with roles and content access that belong to the Looker instance, and admins can build custom roles from permissions and model sets. Data Studio grants predefined IAM roles on assets such as reports and data sources, and neither Data Studio nor Data Studio Pro supports custom roles.
A company wants its official executive dashboard in Looker to be changed only by LookML developers, with every change tracked in the project's Git repository like the rest of the model. Which type of dashboard fits?
AA user-defined dashboard in a shared folder
BA LookML dashboard in a dashboard file of the project
CA user-defined dashboard in the owner's personal folder
DA Look saved from an Explore and pinned to a board
Show the answer and why
AA user-defined dashboard in a shared folder
Incorrect
User-defined dashboards are built with drag-and-drop in the interface by business users and developers and stored in folders, not as files in Git.
BA LookML dashboard in a dashboard file of the project
Correct
LookML dashboards are written in YAML-based dashboard files by LookML developers and stored as version-controlled files in the project's Git repository.
CA user-defined dashboard in the owner's personal folder
Incorrect
A personal folder limits who finds the dashboard, but changes still are not tracked in Git.
DA Look saved from an Explore and pinned to a board
Incorrect
A Look is a single saved query and visualization, not a dashboard, and it is not stored in the Git repository.
Looker has user-defined dashboards, created in the interface and stored in folders, and LookML dashboards, defined in dashboard files and versioned with the project in Git. Either type can be converted to the other.
A Looker developer needs a derived table of customer-level figures (order count, first order date) built from the existing `orders` Explore. She wants to define it with LookML that reuses the Explore's fields rather than with a hand-written SQL query. What should the view's derived_table use?
AA `sql` parameter holding a SELECT statement over the orders table
BA `sql_table_name` parameter that names the orders table
CAn `explore_source` parameter based on the orders Explore
DA `dimension_group` of type time on the first order date
Show the answer and why
AA `sql` parameter holding a SELECT statement over the orders table
Incorrect
A sql parameter defines a SQL-based derived table, which is the hand-written SQL that the developer wants to avoid.
BA `sql_table_name` parameter that names the orders table
Incorrect
sql_table_name points a view at an existing database table; it does not define a derived table.
CAn `explore_source` parameter based on the orders Explore
Correct
A native derived table defines its query in LookML with explore_source, selecting columns from the fields of an existing Explore.
DA `dimension_group` of type time on the first order date
Incorrect
A dimension group adds a set of time-based dimensions to an existing view; it does not create a separate derived table.
Looker derived tables are defined under a view's derived_table parameter. A native derived table uses explore_source to build its query from LookML fields, while a SQL-based derived table uses the sql parameter with a SQL query.