Prepare and test your skills

Prepare and test your skills

Trace the Path from Raw Data to Insight is the five-stage flow that turns raw records into business decisions on Google Cloud. Raw data is captured from source systems, stored reliably, prepared for analysis, then analyzed and interpreted. BigQuery (opens in a new tab) is the fully managed, AI-ready data platform at the center of this path, with serverless architecture and zero infrastructure management.
A five-step left-to-right process flow from capture through store, prepare, and analyze to decide, centered on BigQuery, showing how raw data becomes business decisions and new value.
Raw data capture is the entry stage where data flows from files, applications, databases, or event streams into Google Cloud. BigQuery ingests data through batch loading of local files or Cloud Storage files in Avro, Parquet, ORC, CSV, JSON, Datastore, and Firestore formats. Continuous ingestion uses the Storage Write API (gRPC) for streaming writes and Pub/Sub for ingesting event streams from anywhere at any scale. BigQuery Data Transfer Service automates recurring ingestion, while Dataflow builds batch and streaming processing pipelines and Dataproc runs batch processing, querying, and streaming with managed Apache Spark and Apache Hadoop. After capture, data lands in BigQuery storage or in a pipeline that prepares it for storage.
Data storage for analysis is the stage where raw and prepared data is kept in a reliable, query-ready form. The BigQuery storage layer ingests, stores, and optimizes data in a columnar format built for analytical queries, presents data as tables with rows and columns, supports ACID transactions, and automatically replicates data across multiple locations for high availability. Datasets are the top-level containers for tables and views, created in a regional or multi-region location that defines where data is stored. Cloud Storage holds files that BigQuery can batch load or query directly as external tables, and Dataproc Metastore provides a fully managed Apache Hive metastore for metadata. BigQuery supports structured and unstructured data and open table formats such as Apache Iceberg, Delta Lake, and Apache Hudi.
Data preparation for analysis is the stage where raw data is cleaned, transformed, and organized so queries return trusted answers. BigQuery separates its storage layer from its compute layer, and the two layers communicate over Google's petabit-scale network and scale independently without competing for the same resources. Preparation in the compute layer uses ANSI-standard SQL with joins, nested and repeated fields, aggregation and analytic functions, multi-statement queries, and spatial functions. Data can also be transformed with DML, DDL, and user-defined functions (UDFs), with BigQuery DataFrames, a pandas-compatible Python API, and with BigQuery Studio notebooks that add Python and version control. Federated queries and external tables let BigQuery query data where it lives, including Cloud Storage, Bigtable, Spanner, or Google Sheets in Drive. Knowledge Catalog adds governance with metadata management, data quality, semantic search, and data lineage, while Identity and Access Management (IAM) secures BigQuery resources with the same access model used across Google Cloud.
Analyzing and interpreting data is the stage where prepared data becomes descriptive answers, forecasts, and visualizations. The BigQuery distributed analysis engine can query terabytes in seconds and petabytes in minutes for business intelligence, ad hoc analysis, geospatial analytics, and machine learning. BigQuery ML builds and trains machine learning models directly with SQL queries, and Gemini in BigQuery provides AI assistance including data insights that generate queries from table metadata to uncover patterns. Results are interpreted through analysis and business intelligence tools (opens in a new tab) that connect to BigQuery:
| Tool | Primary use | Key capability |
|---|---|---|
| Connected Sheets | Spreadsheet analysis and sharing | Access, analyze, visualize, and share billions of rows from Google Sheets with no extra exports |
| Looker | Enterprise visualization and business intelligence | Advanced analytics, predictive analysis, and custom data applications; can connect to transactional databases in other clouds |
| Looker Studio | Simple visualization and reporting | Build simple graphics, reports, and dashboards directly from the Google Cloud console |
| Programmatic tools | Code-based analysis | Run Python, Java, or R with Jupyter Notebooks, Apache Zeppelin, or Vertex AI Workbench |
| Third-party tools | Existing tooling | Connect Tableau and Microsoft Power BI, plus existing infrastructure through ODBC and JDBC drivers |
Exam tip: BigQuery separates its storage layer from its compute layer, so each can scale independently; legacy databases often share resources between read/write and analytical operations.
How the Path Creates Business Value describes how each data stage connects to a business decision. Raw data flows from files, applications, databases, or streams into BigQuery or Cloud Storage, then Dataflow or Dataproc and BigQuery compute clean and organize it under Knowledge Catalog governance. Analysts, engineers, warehouse administrators, and data scientists then query it with SQL or Python, visualize it with business intelligence tools, or model it with machine learning. The final insight informs critical decisions, so data generates insight, drives decision-making, and creates new value through innovation and improved operations.
Evaluate How Insights Guide Business Decisions explains how governed data becomes trends, patterns, and forecasts that leaders act on. Data flows from storage into analysis and then into shared tools, so the same data supports daily operations, long-term planning, and measurement. BigQuery (opens in a new tab) finds the insight at scale, Looker (opens in a new tab) shares it as trusted dashboards, and Vertex AI extends it into predictions and generative AI.
BigQuery Surfaces Trends and Patterns at Scale is the analysis capability that lets teams query very large datasets with SQL and Python and no infrastructure to manage. BigQuery is a fully managed, AI-ready data platform with built-in machine learning, search, geospatial analysis, and business intelligence, and its scalable engine can query terabytes in seconds and petabytes in minutes. The storage layer ingests, stores, and optimizes data in a columnar format with full ACID support and automatic replication across locations for high availability. The compute layer runs analytics independently over Google's petabit-scale network, so queries, permission changes, and writes do not compete for resources as they do in legacy databases. Analysts run ANSI-standard SQL with joins, aggregations, analytic and spatial functions, or use BigQuery DataFrames for pandas-compatible Python, query external data where it lives in Cloud Storage, Bigtable, Spanner, or Google Sheets, and share results as views under Knowledge Catalog governance for lineage and discovery.
Looker Delivers Trusted Dashboards for Shared Decisions is the sharing layer that turns analysis into one governed view for every team. Looker is Google Cloud's platform for business intelligence, data applications, and embedded analytics that lets users explore, analyze, and visualize data. Data flows from sources such as BigQuery into Looker, where users retrieve and chart data and build visualizations, dashboards, and looks. Looker defines key metrics once in a unified data model, so business users can self-serve answers without creating conflicting versions of revenue, cost, or performance. Looker supports advanced analytics, predictive analysis, and custom applications and can connect to transactional databases in other public clouds. Complementary paths extend the same flow through analysis and business intelligence tools (opens in a new tab), including Connected Sheets for billions of rows in Google Sheets without exports and Looker Studio for simple graphics and dashboards.
Vertex AI Adds Forecasts and Predictive Insight is the predictive layer that turns past patterns into forecasts for future choices. Vertex AI is a fully managed, unified AI development platform for building and using generative AI. Data flows from BigQuery tables into a model for training, and the resulting forecast flows back into dashboards and applications where decisions are made. BigQuery ML (BQML) builds and trains machine learning models directly with SQL queries inside BigQuery, which lowers the barrier for analysts who already query data. Vertex AI supports fuller generative AI and machine learning development and pipelines, Vertex AI Workbench provides JupyterLab notebooks for Python, Java, or R alongside SQL, and Dataflow can deploy and manage complete machine learning pipelines for batch and streaming data.
Exam tip: Use BigQuery ML when the model can be built and trained directly with SQL inside BigQuery; use Vertex AI when the work needs a unified platform to build and use generative AI beyond SQL.
How Insight Outputs Improve Operations, Strategy, and Measurement ties trends, patterns, and forecasts to three decision types. The same pipeline from BigQuery analysis to Looker visualization to Vertex AI prediction supports daily work, future plans, and proof of results.
The three tools divide the work by the kind of insight needed:
| Service | Primary unit | Choose when |
|---|---|---|
| BigQuery | SQL or Python query and ML model | The work is analyzing large datasets to surface trends, patterns, and forecasts |
| Looker | Governed dashboard and visualization | The work is sharing one trusted view of metrics across teams |
| Vertex AI | Generative AI and ML development | The work is building and using predictive or generative models beyond SQL analysis |
Together, BigQuery provides scale and speed, Looker provides governance and clarity, and Vertex AI provides foresight for the next action.
Assess How Data Creates New Business Value explains how organizations turn data into better decisions, improved operations, and new products or revenue. Value is created when data moves out of isolated systems, is analyzed in one place, and the results are fed into processes and applications. Data analytics (opens in a new tab) on Google Cloud spans acquisition, processing, analysis, and visualization, with services that store, query, visualize, and govern data across environments.
Turning Data into Business Insights is the core analytics motion that converts raw records into repeatable analysis. BigQuery is a fully managed, highly scalable data platform with built-in machine learning, and BigQuery analytics runs analytic queries on large datasets so results keep pace as data volume grows. Looker lets users access, analyze, and act on an up-to-date, trusted version of data, while Looker Studio helps teams tell data stories that support better business decisions. Together these services produce in-depth understanding that decision-makers can act on, rather than one-time reports.
Real-Time Analytics and Governed Data is the combination of fast analysis with trusted, well-managed data. Governed data has been managed and validated before analysis, so real-time business intelligence does not spread errors quickly. Knowledge Catalog grounds agents and users with AI-powered, high-quality context for enterprise data assets. Data governance and sharing cover the end-to-end data lifecycle, making it easier to manage, discover, govern, and share data and AI assets across an organization. An analytics lakehouse stores, processes, and activates data, which standardizes storage formats and gives teams the flexibility, scalability, and agility to make data generate value instead of inefficiency.
The analytics hybrid and multicloud pattern (opens in a new tab) is the ordered design that keeps interactive systems separate from analytical systems. Transactional workloads are interactive applications such as sales, financial processing, enterprise resource planning, or communication. Analytics workloads transform, analyze, refine, or visualize data to aid decision-making. Analytics systems get data from transactional systems by querying APIs or accessing databases, and in most enterprises the two are separate and loosely coupled. The pattern moves data in a fixed order:
Each movement and transformation step can use ETL or ELT depending on data quality and the target use case. Cloud Interconnect and Cross-Cloud Interconnect provide dedicated connectivity to on-premises and other cloud providers when high-volume transfer is required.
Exam tip: The analytics hybrid and multicloud pattern keeps transactional workloads in the private environment and moves raw data into Google Cloud for analytical processing, then may feed results back to the transactional systems.
Advantages of Running Analytics in the Cloud are lower upfront cost, elastic scaling, and full lifecycle coverage. Moving data from a private environment or other clouds into Google Cloud as inbound traffic might be free of charge. Analytics workloads often process substantial amounts of data and can be bursty, so they fit well in a public cloud where compute scales dynamically to process large datasets quickly without overprovisioning equipment. Google Cloud provides services across the entire data lifecycle from acquisition through processing and analysis to visualization, and Cloud Storage is well suited for building a data lake that feeds those stages.
Multicloud Analytics with BigQuery Omni is the capability to analyze data where it already lives in other clouds. BigQuery Omni runs compute power locally to storage on AWS or Azure, letting teams query data stored in Amazon Simple Storage Service (Amazon S3) or Azure Blob Storage without copying it first. This breaks down data silos by bringing analytics to distributed datasets, which supports new insights and value without moving every dataset into one location.
The path captures raw data from files, applications, databases or event streams, stores it reliably, prepares it for analysis, then analyzes and interprets it for business decisions. BigQuery is the fully managed, AI-ready data platform at the center of this path, with serverless architecture and zero infrastructure management.
Use BigQuery ML when the model can be built and trained directly with SQL queries inside BigQuery, which lowers the barrier for analysts who already query data. Use Vertex AI when the work needs a fully managed, unified AI development platform for building and using generative AI beyond SQL analysis.
Looker is Google Cloud's platform for business intelligence, data applications, and embedded analytics that turns analysis into one governed view for every team. It defines key metrics once in a unified data model so business users can self-serve answers without creating conflicting versions of revenue, cost, or performance.
Raw data is extracted from workloads running in the private computing environment and loaded into Google Cloud where it is used for analytical processing. Some of the results may then be fed back to the transactional systems, keeping interactive transactional workloads separate and loosely coupled from analytics workloads.
An online retail company wants to boost customer engagement and unlock new revenue streams on its digital storefront. The organization plans to capture live user interactions and browsing behavior as they happen, analyze user intent instantly, and automatically display personalized product recommendations while the shopper is active on the website.
Which approach enables the organization to create new business value through real-time data and predictive analytics on Google Cloud?
Last updated