A clearer basis for allocating budget.
Compare acquisition cost and return across campaigns using shared definitions of CPA and ROAS. Normalized currencies and channel data help teams judge where spend creates value on a consistent basis.
Our work
Marketing intelligence for smarter investment.
Understand what drives return. Give budget decisions a stronger basis today and build the foundation for forecasting tomorrow.
IntelligenceThe project
For Barbra, every client budget raises the same questions: where is spend delivering value, what explains a change in return, and which campaigns deserve more investment? Disconnected reports made those answers difficult to compare. Wagner built the cloud and data foundation behind Barbra Intelligence so investment decisions can start with a consistent view of performance.
What we delivered
A marketing intelligence platform that brings Google Ads, Meta Ads, and TikTok Ads into one view of spend, conversions, acquisition cost, and return. Automated data pipelines and a governed BigQuery model power dashboards and AI-assisted analysis, helping Barbra investigate performance and compare campaigns. The same normalized history establishes the foundation for future forecasts of spend, conversions, and return.
Services
- Cloud architecture & engineering
- Direct API ingestion
- Data modeling & ELT
- Data quality & observability
- Multi-client product engineering
- Conversational AI integration
Outcomes & value created
- Advertising channels
- 3
- Google Ads, Meta Ads, and TikTok Ads connected for a shared view of investment and performance.
- Campaigns synchronized
- 195
- Campaign catalog recorded at the implementation validation checkpoint.
- Automated refresh
- Daily
- Updated campaign data for ongoing investment reviews, with historical backfills.
- ELT layers
- 4
- Raw, staging, core, and marts turn source records into consistent marketing metrics.
Answers that support the next decision.
Explore which campaigns perform best and what explains a change in return through dashboards and natural-language questions. Teams can investigate a new question without waiting for a new fixed report.
A foundation for forecasting investment.
A normalized campaign history gives Barbra a reusable starting point for future forecasts of spend, conversions, and return. Budget recommendations and predictive models are the next phase of this foundation.
Objectives & challenges
Budget decisions depend on comparing performance fairly. Barbra needed a consistent way to evaluate investment across channels and clients, explain changes in results, and prepare historical data for future forecasting.
- Compare spend and return across platforms with different currencies, formats, and metric definitions.
- Make CPA and ROAS consistent so budget discussions use the same numbers.
- Give each client an authorized view of performance while supporting portfolio-wide analysis for administrators.
- Build a traceable historical record for future forecasts and budget recommendations.
The data behind better investment.
Better investment decisions depend on trustworthy numbers. Direct APIs feed BigQuery, where Dataform validates and unifies campaign metrics. Dashboards and AI consume the same authorized marts, connecting every analysis to a consistent history that can support future forecasting.
Barbra Intelligence architectureLogical data path
Official advertising APIs
- Google Ads
- Meta Ads
- TikTok Ads
Scheduled ingestion
Python adapters · Bounded retries · Date-range backfills
BigQuery · Analytical warehouse
Dataform + SQL · Transformations and quality assertionsAutomation & operations
- Cloud SchedulerDaily refresh
- Secret ManagerManaged credentials
- Cloud LoggingRuns and errors
- Cloud MonitoringVisibility and alerts
Cloud Scheduler starts ingestion jobs. The adapters extract data through official APIs and preserve the source in BigQuery. Dataform transforms and validates the layers before publishing consumption marts.
Authentication and the catalog establish the organization and project context. The dashboard and conversational experience consume authorized marts with the same metric definitions. AI does not make live queries to the advertising platforms.
Cloud, data & AI stack
The engineering behind comparable metrics, trusted analysis, and a foundation for investment planning.
Ingestion & orchestration
Keep investment reviews supplied with updated campaign data through direct APIs and automated ingestion.
- Cloud Run
On-demand Python ingestion jobs.
- Python
Independent Google, Meta, and TikTok API adapters.
- Cloud Scheduler
Daily refresh scheduling.
- Secret Manager
Managed access to integration credentials.
Warehouse & data engineering
Make campaign comparisons reliable. BigQuery stores the history; Dataform standardizes marketing metrics through versioned SQL transformations and quality assertions.
- BigQuery
Raw records, canonical models, and analytical marts.
- GitHub
Version control for Dataform SQL and model definitions.
- Cloud Logging
Run state, loaded rows, and extraction errors.
- Cloud Monitoring
Operational visibility and alerting.
Identity & product
Put performance analysis in the hands of each authorized client, with a shared dashboard and conversational experience.
- Firebase Authentication
User authentication.
- Firestore
Profiles, roles, organizations, and campaign catalog.
- Next.js
Dashboard and conversational application.
- TypeScript
Typed application interfaces.
- Vercel
Web application delivery.
Our approach
Start with the investment questions.
Identify the decisions the data must support: comparing channel efficiency, understanding changes in ROAS, and evaluating campaigns for additional budget. Define metrics, currencies, and reporting granularity around those questions.
Make the numbers comparable.
Preserve platform records in BigQuery, then normalize them with Dataform. Validate required identifiers, uniqueness, and metric consistency so investment reviews draw from one traceable model rather than competing reports.
Connect insight to the decision.
Bring dashboards and conversational analysis onto the same authorized marts. Teams can move from a performance indicator to a focused question while retaining the same client context and metric definitions.
Key activities & takeaways
Keep decision data dependable.
Automate daily refresh and track each ingestion run. Logs, alerts, retries, and backfills help the team investigate missing data before relying on it in a campaign review.
Match cloud spend to the stage of growth.
On-demand processing, partitioned queries, and reused transformations keep the demonstration platform focused on the workload it serves. Quotas, budgets, and alerts support controlled expansion as more sources are added.
Prepare for forward-looking planning.
The next phase is to use the normalized history for forecasts of investment, conversions, and return, alongside anomaly detection and budget recommendations. The delivered data foundation makes that evolution possible.
Put your marketing data to work.
Build the cloud and data foundation to understand return, guide investment, and prepare for forecasting.
Talk to an expert