Major international corporations face a critical challenge: how to effectively manage analytics when teams are scattered worldwide and data needs grow exponentially.
Traditional approaches to building BI systems often lead to work duplication, inconsistent analysis and, consequently, significant waste of time and resources.
“When teams across different countries solve identical tasks, creating similar dashboards and setting up the same processes – that’s the first sign it’s time to change your analytics approach”, notes Kliment Merzlyakov, an expert in analytics systems and machine learning.
With over ten years of experience in building complex analytical solutions, Kliment’s latest project – developing an end-to-end analytics system for an international FMCG corporation – now serves dozens of teams worldwide, from the US and Canada to Switzerland and India.
“In large international companies, I consistently observe the same pattern: multiple disconnected teams working with identical data sources, but in different ways. Solving this requires more than just process automation – it needs a unified, flexible system that works for everyone”, Merzlyakov explains.
How exactly does one build such an efficient system? The answer lies in using a modern data stack – an approach that fundamentally changes the rules of corporate analytics.
“A modern data stack isn’t just a collection of trendy tools. It’s a fundamentally new approach to building analytical systems”, explains Merzlyakov. “We’re moving away from cumbersome monolithic solutions towards a flexible set of interconnected components, each solving its task with maximum efficiency”.
At its core, the modern stack builds on four key elements. First is cloud data warehousing. “I frequently use Snowflake or BigQuery – they excel at handling large data volumes and offer flexible resource scaling”, notes the expert.
The second vital component is ETL (Extract, Transform, Load) tools. “We’ve seen a genuine revolution here”, Merzlyakov emphasises. “We no longer waste time on preliminary data processing and building custom data movement pipelines”.†
The third element is data transformation tools. “dbt was a real game-changer for me”, shares the specialist. “It allows us to apply software development best practices to data processing.
Versioning, testing, documentation – it’s all now available to analysts”. Additionally, with such transformation tools, the concept shifted to ELT, where data is loaded as is and transformed within the warehouse. “It’s far more efficient”, Kliment claims.
The final component is BI platforms for visualisation. “The choice of the tool here is less critical”, Merzlyakov suggests. “What matters is properly structured data architecture in the previous stages. Then any modern BI tool, be it Tableau, Looker, or Superset will do the job”.
“The key advantage of this approach is the ability to quickly scale the solution across the entire organisation”, the expert concludes. “Once you’ve set up the processes, you have a reliable infrastructure that dozens of teams worldwide can use”.
While these components form the foundation, the real question remains: what concrete benefits does it bring to business?
“In large corporations, every technical initiative must have a clear business justification”, Merzlyakov emphasises. “And here, the modern data stack wins on all fronts”.
The first and foremost advantage is scalability. “When building an analytical system for an international corporation, you need to think ahead”, the expert explains. “The modern stack allows you to start small and expand smoothly without radical architecture changes. Adding new teams or data sources becomes routine rather than a separate project”.
Cost-effectiveness is another crucial factor. “Cloud infrastructure and pay-as-you-go models mean you only pay for resources you actually use”, notes Merzlyakov. “No more significant upfront investments in infrastructure that will sit idle most of the time”.
Modern development practices bring additional benefits. “Git for versioning, automated testing, CI/CD – all these make analytics more reliable and transparent. We can track any change, quickly roll back when necessary, and work effectively as a team”, the specialist continues.
Theory might sound convincing, but the real value of any approach reveals itself in practice. A telling example comes from one of his recent projects, where the client faced a classic situation: ten teams across different countries were setting up identical analytical processes for their local divisions.
Before implementing a centralised solution, each team independently collected data from various sources (e.g., Google Analytics), created their own ETL processes, and developed their own dashboards.
“It was a proper nightmare in terms of efficiency”, recalls the expert. “The same mistakes were repeated across different countries, metrics were calculated differently, and approving changes took weeks”.
The solution came with implementing a centralised data stack. “We created a unified pipeline for every data source, set up access management through Row Level Security, and standardised key metric calculations”, explains Merzlyakov. “Now, when a new team needs analytics, they simply connect to the existing system and gain access to all tools and developments”.
The results exceeded expectations. “The time to launch analytics for a new team decreased from several months to several weeks”, notes the specialist.
“Moreover, data quality improved significantly thanks to unified processing and validation standards. Most importantly – teams finally speak the same language, using identical metric definitions”.
“The essence is that you do the work once, but you do it properly”, Merzlyakov concludes. “Instead of reinventing the wheel ten times over, you create a reliable solution that everyone can use. That’s real optimisation”.
Successful technical implementation is only half the battle. Equally important is establishing the right organisational structure to maximise these new capabilities.
“The key to success is proper role and responsibility distribution”, notes Merzlyakov.
From his experience, the optimal structure includes two key groups: a core team and local analysts. “The core team is the infrastructure and standards custodians”, Merzlyakov explains. “They’re responsible for architecture, security, and system performance. These are the people ensuring the foundation remains solid as we scale”.
However, centralisation shouldn’t become a bureaucracy. “Local analysts retain significant autonomy”, the specialist emphasises. “They can create their own data transformations, develop department-specific reports, and experiment with new metrics. The crucial point is they do this within the common infrastructure, following established principles”.
“This approach resolves the classic dilemma between standardisation and flexibility”, Merzlyakov continues. “The core team provides a reliable foundation and common standards, whilst local teams can quickly respond to their divisions’ specific needs”.
Moving from organisational structure to technical implementation, attention to detail becomes crucial. “In analytics, the devil is in the details”, notes Merzlyakov. “Project success often depends on correctly chosen technical solutions and their proper configuration”.
Merzlyakov pays particular attention to data security. “Row Level Security isn’t just a feature, it’s fundamental for working with data in an international company”, he emphasises. “We configure RLS at the warehouse level, ensuring each team sees only their data, regardless of the BI tool used”.
“In ETL processes, reliability is paramount”, Merzlyakov continues. “I always implement detailed logging and monitoring. Every failure must be immediately detected and fixed before it affects business users”.
“Last but not least – documentation”, the expert concludes. “Modern tools allow for automatic code-based documentation generation. This isn’t just convenience – it’s essential for distributed team operations. Every metric and data transformation must be clearly described and easily accessible to all system users”.
“We’re on the cusp of a new era in corporate analytics”, asserts Merzlyakov. “The modern data stack represents a new philosophy of working with data, enabling companies to be truly global”.
“In the coming years, we’ll see even deeper integration of analytics tools with decision-making processes”, the expert predicts. “Companies investing in proper analytical infrastructure today will gain significant advantages in the future”.
Merzlyakov’s approach to building analytical systems isn’t just a technical recipe. It’s a fresh perspective on how analytics should function in a modern international company.
This view, grounded in real experience, deep technical understanding and, crucially, business acumen, suggests that the future of corporate analytics isn’t just coming – it’s already here. The only remaining challenge is implementing it correctly.
