Saj Patel

Financial Operations (FinOps) is a mature practice for most organizations and is designed to evaluate various aspects of the business through a finance lens. Financial operations best practices ultimately aim to manage and improve key bottom-line metrics through efficiency efforts, responsible governance, and assigned accountability. Managing Data Management Costs with FinOps As companies have focused on implementing modern data architectures to enable their business strategies in the past several years, they are witnessing the fiscal challenges of implementing and operating those solutions. FinOps data management is quickly finding its relevance in helping organizations better manage the costs of their data management solutions. In 2025 we will witness an even greater emphasis on FinOps data management and understand its impact on decision making within the data ecosystem. Financial operations best practices will even drive innovation in the development roadmap for platforms to enable it natively. While the modernization of data platforms for companies is impactful in achieving their respective data strategies, there are some financial surprises along the way, sometimes risking the viability and continuity of the programs. The costs exceed beyond their original anticipation, impacted by the choice of technologies, the execution strategy to stand up the new solutions, or the underlying operating model. Tough conversations at the leadership level have become commonplace. What Are the Variables Impacting Costs? Integrating the chosen technologies, often a hybrid of best-fit-for-purpose platforms and tools that must align with an existing landscape, was more complex than imagined. We are not quite there with plug-and-play solutions or a single platform that can address everything. Another variable is the implementation strategy and plan itself. The original plans are often too optimistic, and the actual activation of initiatives delays the business' ROI. Migrating a legacy solution to the new platform, setting up a new operating model for a data management function, or implementing a new capability come with their own challenges to navigate. Stir in the challenge of business and technical team adoption, and you will have a recipe for cost overrun. This leads to a search for answers. Where can I create efficiencies to lower costs, reduce risk, and ultimately accelerate time to value? This is where looking for solutions through a lens of financial operations best practices provides a fresh perspective that will alter the thought process for decision-making. Here are some of the ways that FinOps drives towards implementing a fiscally responsible model. Improving efficiency — doing more with less, eliminating redundancies, reducing non-value-added or expensive activities, and introducing automation. Strengthening governance —processes to improve quality, establishing accountability goals, monitoring and reporting progress of key metrics If you are interested in applying financial operations best practices principles to your data management solution, here are a few topics to evaluate for inclusion in your strategy. Simplify the Data Stack Implementing a best-of-breed strategy for data technologies certainly can bring unique value to a solution but not without complexity. Consider limiting the number of vendors and platforms or look for a platform that combines the different functions into one cohesive and integrated solution. More and more vendors are looking to deliver as many data functions as possible into their platform or product family: data ingestion, transformation, data quality, data architecture, metadata management, data science, and sometimes BI/analytics. The tradeoff you must consider is getting locked into the platform for a long commitment. This type of decision alone can reduce the complexity of integration and naturally enable automation. Furthermore, it will provide more complete observability to manage key metrics through a single pane rather than requiring you to collate and curate from many platforms. Thoughtful Decision on Real-time Data Ingestion Every business leader believes that real-time analytics will provide them with the timely and actionable information they need to make business decisions. This capability, however, requires a true re-engineering of the data integration architecture, extending back into the technologies used at the sources of data and infrastructure components to enable integration. While event-driven architectures are becoming the norm, they can be expensive to sustain if not used intelligently. In many cases, such as traditional performance reporting, the business does not need real-time data. In those cases, there is no need to push a pattern that introduces complexity and new costs when it is not essential. If you have the opportunity, create an architecture that supports the various patterns for data ingestion, allowing you the flexibility to choose the most appropriate pattern based on the need for information. I believe in right-sizing a solution rather than over-engineering it. Consider a data fabric solution if there is a true real-time analytics value proposition. Data fabric solutions that leverage FinOps data management virtualization techniques eliminate or reduce the need for traditional movement of data and can be cost-effective and a great time-to-market solution. To emphasize my earlier point, an architecture that allows you to incorporate data virtualization into a thoughtful and flexible architecture will give you the levers to use what is appropriate for a fiscally responsible solution. Automation A common theme across FinOps data management is introducing automation wherever possible. Done well, there is a clear correlation between automation and value delivered. Automation can be applied in many forms in the data ecosystem. DevOps, once limited in context to application development domains, is now an inherent element of the design of a well-orchestrated data solution. Any part of the build and deployment phases that can be automated through DevOps will reduce an application's development lifecycle and aid with quality as it progresses from concept to production. Automated testing is of greater interest to most companies. Manual testing involving people is expensive, time-consuming, and requires strong discipline. If you can divert basic use cases and scenarios to automated approaches that you can also incorporate into your DevOps approach, you are better able to focus your manual testing efforts on the more complex scenarios. It will provide better coverage overall and deliver more confidence in the quality of the code delivered. Another area for automation is FinOps data management pipeline orchestration. The cohesive, broadly capable platforms I described to simplify the data stack inherently have solid orchestration methods. Leveraging these built-in features as much as possible to streamline workflows must be a goal. Even when employing a hybrid set of technologies, the major cloud providers provide various orchestration tools and methods to create an automated workflow of the different technologies. Any function that can be scheduled or triggered should be in consideration for automation. It requires work but delivers good payback. Automation should also be considered for other aspects of data management. Roles involved with data stewardship and quality management need their workload limited to actionable items. Automation implemented within metadata management tools to gather metadata, appropriately tag information, categorize data, or address anomalies can help these functions focus on what is important. This approach further enables other functions like governance, compliance, and security. Automation within data quality platforms to create scorecards, address anomalies or quality issues, and feed workflows for remediation will directly accelerate business value. Code Development Assist Technology Accelerating code development while ensuring quality and maintainability has always been an ambition for data engineering teams. Technology advancements and a heightened focus on this topic have advanced capabilities within vendor platforms in a couple of ways. The majority of data engineering platforms simplify the code development process with more no-code capabilities or abstraction techniques to hide the complexity. The platforms may also be designed to encourage the reuse or leverage of pre-built and community-developed components to eliminate time-consuming activities. While custom code development is essential for unique business logic, it should be completely unnecessary for basic transformation rules that comprise most business rules. The second advancement relies on AI and foundation models to generate and assess software code and artifacts. GitHub Copilot, Gemini, Augment Code, various Amazon platforms, and many language-specific solutions are showing proven ROI and will continue to get better. Automated documentation generation using AI is another big-time saver and often provides a richer, consistent quality that can improve manageability. Reducing the time to create, test, or document code and improving the quality of the end applications produced has a very measurable impact on efficiency and cost. The FinOps perspective definitely merits consideration within the data management domain. As you begin to implement the principles for financial operations best practices, always remember to measure and quantify progress and results. Enjoy the journey.