By Metin Sarikaya, Head of Data and Business Intelligence

Introduction

Artificial intelligence (AI) has emerged as a transformative force in banking, reshaping how we approach data warehousing and business intelligence (BI). As the head of Data Warehouse, Business Intelligence, and Big Data initiatives at Akbank, I’ve witnessed firsthand the profound impact AI is having on our ability to derive insights from data and make informed decisions. This article explores how AI is revolutionizing data management and analytics in the banking sector, offering both opportunities and challenges for institutions looking to stay competitive in an increasingly data-driven world.

The Evolution of Data Warehousing and Business Intelligence

Before delving into the impact of AI, it’s crucial to understand the evolution of data warehousing and BI in banking. Traditionally, our data warehouses served as centralized repositories for structured data, primarily from transactional systems. Business intelligence tools allowed us to query this data, generate reports, and perform basic analytics.

However, as data volumes exploded and new data types emerged, traditional systems began to show their limitations. At Akbank, we found ourselves struggling to integrate unstructured data from sources like customer emails, social media, and call center logs into our existing BI framework. The need for more advanced analytics capabilities became increasingly apparent.

Enter Artificial Intelligence

The integration of AI into data warehousing and BI has been nothing short of revolutionary. AI technologies, particularly machine learning (ML) and natural language processing (NLP) have enhanced our ability to process, analyse, and derive insights from vast amounts of structured and unstructured data.

Data Integration and Preparation

One of the most significant challenges in data warehousing has always been the integration and preparation of data from diverse sources. AI has dramatically streamlined this process. At Akbank, we’ve implemented machine learning algorithms that automatically classify and tag incoming data, significantly reducing the manual effort required for data preparation. For example, our AI system can now automatically categorize customer transactions from various channels (mobile app, website, branches) into predefined categories such as “product campaigns,” “complaints,” or “fraud alerts”. This not only saves time but also improves the accuracy of our data categorization.

Advanced Analytics and Predictive Modeling

AI has taken our analytics capabilities to new heights. Machine learning models now allow us to move beyond descriptive analytics to predictive and prescriptive analytics. For instance, we’ve developed an AI-powered credit scoring model that analyzes a wide range of data points, including traditional financial data and alternative data sources like location-based payment activity and mobile phone app usage patterns. This model has improved our ability to assess credit risk, resulting in a significant percentage reduction in default rates for new loans while simultaneously increasing our approval rates for creditworthy customers who might have been overlooked by traditional scoring methods.

Large Language Models for Unstructured Data Analysis

LLM has been a game-changer in our ability to derive insights from unstructured data. We’ve implemented an LLM-based AI Chatbot to analyze employees’ needs. This system has a 10000-article knowledge base, generates correct answers 90 percent of the time and allows customer representatives to easily save time. This enhances quality and accuracy and helps customer support agents save three minutes in every interaction, freeing them up for more proactive service.

Automated Reporting and Data Visualization

AI has transformed how we generate and consume BI reports. We now use AI-powered tools that can automatically generate insights from our data and present them in easily digestible formats. These tools use natural language generation to create narrative explanations alongside visual data representations, making complex data more accessible to non-technical stakeholders. For example, our dashboards now include an AI assistant that can answer questions about the data in natural language. Business units can ask questions like “What was our loan growth in the SME sector last quarter compared to the same period last year?” and receive instant, accurate responses.

Real-time Decision Making

The combination of AI with real-time data processing has enabled us to make split-second decisions based on up-to-the-moment data. We’ve implemented a real-time ATM efficiency system that uses machine learning to analyze ATM logs and flag potential out-of-service ATMs instantly. Thus, related units can take action as soon as possible. This system has increased our customer satisfaction by up to 5% while also reducing the out-of-service times of those ATMs, improving customer experience.

Challenges and Considerations

While the benefits of AI in data warehousing and BI are substantial, the implementation journey is not without challenges:

Data Quality and Governance

AI models are only as good as the data they’re trained on. Ensuring high-quality, well-governed data across our entire data ecosystem has been crucial. We’ve had to invest significantly in data quality tools and processes to maintain the integrity of our AI-driven insights.

Skill Gap

Integrating AI into our data and BI processes required new skill sets. We’ve had to invest in training our existing team and recruit specialists in areas like machine learning and data science.

Explainability and Transparency

In the highly regulated banking sector, the “black box” nature of some AI models presents challenges. We’ve had to work on developing explainable AI models, particularly for critical applications like credit scoring, to ensure we can provide clear rationales for AI-driven decisions.

Ethical Considerations

The use of AI in analyzing personal data raises important ethical questions. We’ve established an AI ethics committee to ensure our use of AI aligns with our values and respects customer privacy.

Best Practices for AI Integration in Data Warehousing and BI

Based on our experience at Akbank, here are some best practices for banks looking to leverage AI in their data and BI strategies:

1. Start with a clear use case and ROI calculation
2. Invest in data quality and governance from the outset
3. Build a multidisciplinary team combining domain expertise with AI skills
4. Prioritize explainable AI, especially for customer-facing applications
5. Continuously monitor and validate AI models to ensure ongoing accuracy and relevance
6. Establish clear ethical guidelines for AI use

Future Trends

Looking ahead, we see several exciting trends in the intersection of AI, data warehousing, and BI:

1. Augmented Analytics: AI will increasingly guide users to the most relevant insights, automating much of the analysis process.
2. Edge Analytics: With the growth of IoT in banking (e.g., ATMs, POS devices), we expect to see more AI-driven analytics happening at the edge, closer to the data source.
3. AI-Driven Data Management: AI will play a larger role in automating data management tasks, from data integration to quality control and lifecycle management.
4. Conversational BI: Natural language interfaces will become the norm, allowing non-technical users to interact with data using conversational queries.

Conclusion

The integration of AI into data warehousing and business intelligence represents a significant leap forward in our ability to derive value from data. At Akbank, it has allowed us to enhance our decision-making processes, improve customer experiences, and gain a competitive edge in an increasingly data-driven banking landscape.

As we continue to navigate the complexities of modern banking, the symbiosis of AI with our data and BI systems will undoubtedly play a crucial role in shaping our strategies and operations. For banks looking to thrive in the digital age, embracing AI in their data warehousing and BI practices is no longer just an option—it’s a necessity.

Author Bio: Metin Sarıkaya leads the Data Warehouse, Business Intelligence, and Big Data initiatives at Akbank, one of Turkey’s major banks. With almost a decade of experience in banking technology, he has been at the forefront of integrating cutting-edge AI technologies into traditional data management practices.