Optimizing inflation assessment for NBP

The client

National Bank of Poland (NBP), Poland’s central bank, serves as the backbone of the nation’s economic resilience. With key responsibilities such as shaping monetary policy, managing currency issuance, and ensuring the stability of financial systems, NBP plays a vital role in Poland’s financial landscape.

Among its strategic priorities is accurate inflation monitoring, a critical process that guides decisions influencing the economic well-being of millions and the nation’s overall prosperity.  

One of the key departments driving this effort is the Economic Analysis and Research Department, which is responsible for conducting inflation forecasts. 

The challenge

Inflation assessment is a cornerstone of NBP’s mission, providing essential insights that influence national economic policies and underpin financial stability. 

The Department of Economic Analysis and Research (DABE) at NBP is responsible for creating inflation projections, forecasting economic trends under the assumption of fixed NBP interest rates. 

To maintain its operation, NBP required a reliable, daily stream of price data to support inflation index calculation. However, the bank’s existing data collection system presented significant challenges. Price data was coming from a wide range of providers in varying formats and with inconsistent availability, making the process of obtaining information inefficient and complicating analysis while hindering timely decision-making.

One of the main challenges was also managing significant fluctuations in demand during peak hours while maintaining system stability throughout the business day. An additional complexity came from industry-specific patterns, such as restaurants opening later compared to other retail sectors.

The solution

Recognizing the strategic importance of reliable inflation monitoring, BitPeak transformed NBP’s data collection and analysis processes. We designed and deployed a fully automated pipeline for web scraping and data processing in Microsoft Azure. The system integrates open data sources to track product prices aligned with key items in the inflation reference basket and delivers consistent, accessible, and well-structured outputs.

Modern analytic platform

Our repeatable and scalable system in Microsoft Azure automatically collects data from around 40 e-commerce websites across various sectors, processes it, and stores it in a structured format for analysis. To enhance scalability and efficiency, we seamlessly migrated NBP’s operations from the SAS environment to Azure, providing a modern, cloud-based platform designed for high-performance analytics.

The solution was created as an evolution of existing processes, with a strong focus on reliability, standardization, and automation. Our support extended beyond implementation, offering comprehensive maintenance to ensure the system’s uninterrupted performance and long-term reliability.

As a result, BitPeak’s solution significantly strengthened NBP’s ability to produce accurate inflation and GDP forecasts, contributing to more informed economic policymaking.

Monitoring and data quality

A robust monitoring process ensures continuous insight into both scraper performance and the quality of collected data. This includes automated reporting, data inspection routines, and stability tracking.

Telemetry from Azure Application Insights, combined with task metrics and states from Azure Batch, enables real-time visibility into execution duration, error rates, and system health. Rather than relying solely on synthetic tests, quality control is grounded in production data, verifying completeness, schema conformity, and logical consistency.

This approach ensures that analysts can trust the data being delivered for economic modeling and inflation forecasting.

Security and compliance

Security and data governance were also key considerations throughout the system design. All secrets and credentials are securely stored in Azure Key Vault and accessed at runtime by containerized tasks, ensuring minimal exposure and adherence to least-privilege principles.

Containerization also enhances security by isolating execution environments and simplifying image updates, including the deployment of patches and configuration changes.

Collected data is stored in Azure Data Lake Storage Gen2 using the Parquet format. This not only supports efficient analytics but also enables fine-grained access control at the storage level and ensures consistent, standards-based data processing across the pipeline.

To reach our goal, we used the following Microsoft Azure tech stack:

• Azure Cloud Hosting: Provided scalable infrastructure to support stable data scraping operations
• Azure Key Vault: Used for secrets management 
• Azure Data Lake Storage: Secured and managed large volumes of structured data in a standardized format for analysis
• Azure Batch and Jobs: Automated data collection for tasks, ensuring updates occurred daily, Monday to Friday, without manual intervention 
• Azure Container Registry: Platform to Store Containerized version of code-based solution for daily scraping
• Azure Monitor: Tracked data quality and scraper performance, offering real-time alerts to preempt and resolve any issues