Looker for visualization in GCP - Step-by-Step Execution

Process Flow - Looker for visualization

Connect to Data Source

↓

Create LookML Model

↓

Define Explores & Views

↓

Build Visualizations

↓

Save & Share Dashboards

↓

Users Interact & Explore Data

Looker connects to data, models it with LookML, builds visualizations, and shares dashboards for users to explore.

Execution Sample

GCP

1. Connect Looker to BigQuery dataset
2. Define LookML model with views
3. Create an Explore for analysis
4. Build charts and tables
5. Save dashboard and share

This sequence shows how Looker connects data, models it, creates visualizations, and shares dashboards.

Process Table

Step	Action	Input/Config	Result/Output	Notes
1	Connect to Data Source	BigQuery project and dataset info	Connection established	Looker can query data live
2	Create LookML Model	Define views and explores in LookML	Model ready for analysis	Models describe data structure
3	Build Visualization	Select fields and chart type	Chart/table rendered	Visualizes data from model
4	Save Dashboard	Add visualizations to dashboard	Dashboard saved	Dashboard groups visuals
5	Share Dashboard	Set permissions and share link	Users can view and explore	Enables collaboration
6	User Interaction	Users filter and drill down	Data updates dynamically	Exploration without coding
7	Exit	No further actions	Session ends	User finishes data exploration

💡 User finishes exploring data and closes Looker session

Status Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	After Step 5	After Step 6	Final
Connection Status	Not connected	Connected	Connected	Connected	Connected	Connected	Connected	Connected
LookML Model	None	None	Defined	Defined	Defined	Defined	Defined	Defined
Visualization	None	None	None	Created	Created	Created	Updated by user	Final visual state
Dashboard	None	None	None	Saved	Saved	Shared	Shared	Shared
User Interaction	None	None	None	None	None	None	Active	Ended

Key Moments - 3 Insights

Why do we need to create a LookML model before building visualizations?

LookML models define the data structure and relationships so Looker knows how to query and display data correctly, as shown in step 2 of the execution_table.

Can users explore data without writing SQL in Looker?

Yes, users interact with visualizations and filters dynamically without coding, as shown in step 6 where user interaction updates data views.

What happens if the data source connection is lost?

Looker cannot query data, so visualizations won't update. Connection status in variable_tracker shows 'Connected' only after step 1.

Visual Quiz - 3 Questions

Look at the execution_table at step 3, what is the output?

A Chart or table rendered

B LookML model is defined

C Connection to data source established

D Dashboard saved

💡 Hint

Refer to the 'Result/Output' column in row for step 3 in execution_table

According to variable_tracker, when does the 'Dashboard' variable change from 'None' to 'Saved'?

A After Step 2

B After Step 5

C After Step 4

D After Step 6

💡 Hint

Check the 'Dashboard' row in variable_tracker and see the value after each step

If the connection to the data source fails at step 1, what will happen to the 'Connection Status' variable?

A It will show 'Connected'

B It will remain 'Not connected'

C It will change to 'Disconnected' after step 3

D It will show 'Active'

💡 Hint

Look at the 'Connection Status' row in variable_tracker and what it shows at start and after step 1

Concept Snapshot

Looker connects to data sources like BigQuery.
You define data models with LookML.
Build visualizations by selecting fields and chart types.
Save and share dashboards for collaboration.
Users explore data interactively without coding.

Full Transcript

Looker is a tool that connects to data sources such as BigQuery. First, you establish a connection to your data. Then, you create a LookML model that describes your data structure and relationships. After that, you build visualizations like charts and tables by selecting fields and chart types. These visualizations are saved into dashboards, which you can share with others. Users can then interact with these dashboards by filtering and drilling down into data without needing to write any code. The process ends when the user finishes exploring the data and closes the session.

Practice

1. What is the main purpose of Looker in data analysis?

easy

A. To write complex SQL queries manually

B. To create visual reports from raw data without coding

C. To store large amounts of raw data

D. To replace databases with spreadsheets

Solution

1. Step 1: Understand Looker's role

   Looker is designed to turn raw data into visual reports easily.

2. Step 2: Compare options

   Only To create visual reports from raw data without coding describes Looker's main function correctly; others describe unrelated tasks.

3. Final Answer:

   To create visual reports from raw data without coding -> Option B

4. Quick Check:

   Looker = Visual reports without coding [OK]

Hint: Looker = easy visuals from data, no coding needed [OK]

Common Mistakes:

• Thinking Looker stores data like a database
• Confusing Looker with manual SQL writing
• Assuming Looker replaces spreadsheets

2. Which syntax correctly defines a measure in LookML?

easy

A. measure total_sales { sum: sales }

B. measure total_sales = sum(${TABLE}.sales)

C. measure: total_sales { type: sum; sql: ${TABLE}.sales ;; }

D. measure: total_sales = sum(sales)

Solution

1. Step 1: Recall LookML measure syntax

   Measures use the format: measure: name { type: ...; sql: ... ;; }

2. Step 2: Check each option

   measure: total_sales { type: sum; sql: ${TABLE}.sales ;; } matches correct LookML syntax with colons and braces; others do not.

3. Final Answer:

   measure: total_sales { type: sum; sql: ${TABLE}.sales ;; } -> Option C

4. Quick Check:

   LookML measure syntax uses colons and braces [OK]

Hint: LookML uses colons and braces for measures [OK]

Common Mistakes:

• Using equal signs instead of colons
• Omitting braces {} around measure definition
• Incorrect SQL syntax inside measure

3. Given this LookML snippet, what will the measure average_price calculate?

measure: average_price {
  type: average
  sql: ${TABLE}.price ;; }

medium

A. Sum of all prices

B. Count of price entries

C. Maximum price value

D. Average of the price column

Solution

1. Step 1: Identify measure type

   The measure type is 'average', which calculates the mean value.

2. Step 2: Understand SQL reference

   The SQL references the price column, so it averages all prices.

3. Final Answer:

   Average of the price column -> Option D

4. Quick Check:

   Measure type 'average' = mean value [OK]

Hint: Measure type 'average' means mean value [OK]

Common Mistakes:

• Confusing average with sum or count
• Ignoring the measure type keyword
• Assuming it returns max value

4. You wrote this LookML measure but get an error:

measure: total_revenue {
  type: sum
  sql: ${revenue} ;; }

What is the likely cause?

medium

A. Missing ${TABLE} prefix before revenue

B. Incorrect measure type 'sum'

C. Extra semicolon after sql statement

D. Measure name cannot have underscores

Solution

1. Step 1: Check SQL reference format

   LookML requires ${TABLE}.column to reference table columns properly.

2. Step 2: Identify missing prefix

   The sql references ${revenue} without ${TABLE}., causing error.

3. Final Answer:

   Missing ${TABLE} prefix before revenue -> Option A

4. Quick Check:

   Column refs need ${TABLE}. prefix [OK]

Hint: Always prefix columns with ${TABLE}. in LookML SQL [OK]

Common Mistakes:

• Omitting ${TABLE}. before column names
• Thinking measure type 'sum' is invalid
• Assuming underscores are not allowed in names

5. You want to create a dashboard showing total sales and average order value per region. Which LookML approach is best?

hard

A. Define measures for total sales and average order value, then create an explore with region dimension

B. Write raw SQL queries for each metric and paste results manually into dashboard

C. Create separate dashboards for sales and orders without linking regions

D. Use only dimensions without measures to keep dashboard simple

Solution

1. Step 1: Identify needed metrics and grouping

   You need total sales and average order value as measures, grouped by region dimension.

2. Step 2: Use LookML best practice

   Defining measures and an explore with region dimension allows dynamic, reusable dashboards.

3. Step 3: Evaluate other options

   The other options do not provide integrated, maintainable visualization.

4. Final Answer:

   Define measures for total sales and average order value, then create an explore with region dimension -> Option A

5. Quick Check:

   Measures + dimension + explore = best dashboard [OK]

Hint: Use measures + dimension + explore for dashboards [OK]

Common Mistakes:

• Manually copying SQL results instead of using LookML
• Separating dashboards without common grouping
• Ignoring measures and using only dimensions