How google/pmax_best_practices_dashboard works
Project Type & Purpose
• This project is a full-stack reporting/dashboard solution built to monitor and report on performance metrics for pMax campaigns. Its main deliverable is a Looker Studio dashboard that aggregates data from Google Ads (via GAQL queries) and processes information using Google Cloud Platform (GCP) services.
• The solution is designed to provide best-practice insights for campaign and asset performance using SQL queries (both BigQuery and GAQL), shell scripts for deployment/upgrades, and configuration files.
Analyzing the File Structure
• The top-level files and directories (such as the SQL queries folders – “bq_queries”, “google_ads_queries”, plus additional queries in the “plugins/retail” directory) indicate that the system extracts and processes data from Google Ads and uses BigQuery for analysis.
• Multiple shell scripts (build-retail.sh, no_gmc_update.sh, setup-wfs.sh, upgrade_pmaximizer.sh, and non_retail_to_retail_upgrade.sh) show the automation and deployment aspects of the solution.
• The presence of package.json and package-lock.json implies some Node.js-based tooling or dependency management that might be involved in orchestrating parts of the project.
• Configuration details (with files like answers.json and various SQL files) suggest that the project is structured to support different deployment scenarios (e.g., standard and retail versions) along with regular updates.
Reading the README for Architectural Guidance
• What happens during installation (automated deployment)
• Automatic deployment of components like Storage, Scheduler, Workflows, Run, and BigQuery.
• The README explains the daily pipeline, GCP authentication, and troubleshooting steps which further confirm the cloud-based design.
a. Main Components to Include:
• Frontend/Presentation Layer: Represented by the Looker Studio dashboard that displays the reporting results.
• Data Extraction Layer: Google Ads queries (GAQL queries) that retrieve data from Google Ads.
• Data Processing/ETL Layer: SQL queries in “bq_queries” and additional scripts; this is where data is processed, transformed, and stored.
• Google Cloud Infrastructure:
• Google Cloud Storage (for file storage)
• Cloud Scheduler & Workflows (for scheduling, triggering ETL pipelines)
• Cloud Run (to run containerized processing tasks)
• BigQuery (for data warehousing and analytics)
• Auxiliary Components: The shell scripts and plugins that facilitate deployments and updates, which could be annotated as supporting utilities.
b. Relationships & Interactions:
• That shell scripts (like upgrade and deployment scripts) control or trigger the automation of processes within the GCP environment.
• Separation between different deployment variants (e.g., standard vs. retail) by including the “plugins/retail” module and how these might integrate with the central pipeline.
c. Architectural Patterns/Design Principles:
• Microservices/cloud functions design: each GCP component (Storage, Scheduler, Workflows, Run, BigQuery) is modular and integrated to build an automated, scalable pipeline.
• Automation and the “push button” nature of the deployment, as the project uses several shell scripts to manage updates and deployments.
• Separation of concerns: queries and workloads for data extraction (google_ads_queries, bq_queries) are separated from the orchestration and deployment mechanisms.
d. Relevant Technologies & Tools:
• Looker Studio for visualization.
• Google Ads API with GAQL queries.
• Google Cloud services (Storage, Scheduler, Workflows, Run, BigQuery).
• SQL for data processing.
• Shell scripting for automation.
• NodeJS (implied by package.json) for dependency management or possible backend utilities.
Final Advice
• Double-check that each deployment script or module (as seen in the file tree) is represented where relevant within the cloud orchestration flow (especially scripts like upgrade_pmaximizer.sh and non_retail_to_retail_upgrade.sh which adjust the system’s behavior).