Читать книгу Self-Service Data Analytics and Governance for Managers - Nathan E. Myers - Страница 12

The breadth and scale of data are growing exponentially, and this growth of data is impacting the shape of organizations. Across industries, many companies have entire departments and functions devoted to processing vast numbers of data points into information, for delivery to internal and external stakeholders. Along with the growth of data, data analytics technology and tooling are advancing at a breakneck rate to process it, to identify and understand relationships and trends, and even to make predictions on future outcomes, before displaying them neatly in low-latency dashboard views for ultimate consumption by managers, executives, clients, counterparties, and regulators.

Data analytics is coming to the fore as an exciting strategic and tactical enabler of higher-order analysis and value creation through insight generation and automation of manual processes. Data analytics includes a number of analytics tools, technologies, and buzzwords readers will have heard thrown about more and more over the last 5 to 10 years: robotic process automation (RPA), machine learning (ML), artificial intelligence (AI), text mining technologies like natural language processing (NLP), optical character recognition (OCR), and intelligent character recognition (ICR), along with neural networks, logistic and linear regression analysis, and many more. At the most basic level, these are disciplines enabling descriptive techniques to understand past events and their drivers and to gain insight through the extraction of data trends. These technologies can allow us to forge more structured and intelligent processing steps, and at their sexiest, they can enable predictions, trigger recommendations or prescribed actions, and prompt informed decision-making.

More data analytics and automation tools are available to perform routinized tasks in the data-to-information processing chain than ever before. Digital transformation features in the concerns of most Fortune 500 CEOs, and while companies report that the chief goals of digital transformation are to understand their customers better or to improve products or services, quite often there are more practical motivations for digital transformation, particularly in finance, accounting, and operations functions. The goals are to build capacity and create efficiencies through automating routinized processes, improve process stability and control by structuring the work performed outside core systems, and to optimize human capital resources by reducing the proportion of low value-added processing tasks in their workday.

Many large firms employ dozens, hundreds, or even thousands of employees, who spend their days enriching, processing, transforming, and perhaps to a limited degree, even analyzing data in Microsoft Excel. They may work in a variety of functional silos in the organization, whether as product controllers, entity controllers, or accountants within the CFO organization, whether they work in an operations function, or whether they work in a business management or business intelligence function rolling up to a COO – or in any other part of the organization. Spreadsheet processing continues to dominate in accounting, finance, and operations functions, but the thick and lengthy manual processing tail performed outside of systems highlights the shortfall of core technology platforms in meeting users' needs. Advancements in data analytics and automation tooling delivers viable alternatives, with the potential to supplant and dethrone Microsoft Excel as the default business processing tool, and perhaps finally relegate it to where it belongs – one of several quick and dirty tactical tools available for selection if and as required, but not the default go-to, where the majority of processing teams live, day by day.

When we think of AI, data science, and related disciplines, for many of us, the most leading-edge advanced analytics capabilities come to mind – the ability to detect anomalies in financial data, the ability to detect and flag high-risk patient test results for further review by medical practitioners, and any of a host of proven use cases for descriptive and predictive analytics. While these represent striking and momentous opportunities to employ data analytics to great effect, we submit that these higher-order applications are the exception rather than the rule. The bread and butter use cases that will offer predictable and quantifiable benefits for rapid adoption of data analytics are the opportunities that exist in spreadsheet-driven environments to structure processes in self-service data analytics tooling. Data analytics is not always a theoretical predictive rocket science employing code-based technology and advanced algorithms to solve complex problems. It is the body of solutions that organizational leaders must cultivate and have at hand to forge ahead in their digital journey, at whatever pace is consistent with digital transformation goals and objectives. The development of practical data analytics tools has led to billions of dollars of investment across many service industries, aimed at building core competencies, increasing competitive advantage and organizational efficiency, doing more with fewer employees, or reducing employee costs and footprint.

It is this last goal that the authors predict will prompt a surge in adoption of data analytics tooling in the next five years, across medium to large-scale enterprises. In many organizations, the cost of employees is the most significant expense on the income statement. Managers are motivated to structure their spreadsheet-based processes in a more mature and robust way. By reducing the manual processing performed in Excel, managers can stabilize and lock down spreadsheet-driven processes into more repeatable, structured, and time-efficient processing steps. By minimizing both process variance and time spent performing routinized processing steps, spreadsheet-based jobs of the past will evolve to remove the most manual and least value-added steps in the processing chain. While this book cannot but introduce and acknowledge many more advanced data analytics capabilities and technologies to provide the backdrop, the focus of this book is largely around one subset of the emerging tool suite, self-service data analytics.

Self-service data analytics is an important growing subset of the suite of data analytics tools that is emerging as a focal point of digital transformations across large companies. It is distinct from the other sets of tools in the analytics toolkit in important ways. First, self-service tools are typically built in off-the-shelf vendor products with which individual operators, not technologists, can interact and configure directly, due to their ease of use. Process owners, with no prior technology background and that may have never seen a piece of code, are well equipped to lay out a customized, automated process, armed only with their knowledge of the raw data inputs and the processing steps they routinely perform in spreadsheets. Intelligent source data parsing and drag-and-drop operations replace SQL and Visual Basic commands, enabling the most inexperienced, inexpert, if not maladroit and bungling of us to quickly roll up our sleeves, forge and test processing steps, and implement a processing workflow, all in an afternoon (“small” automation). Benefits cases based on control and efficiency are self-evident to those performing processing, and the decisions to invest time in structuring spreadsheet processes into automated workflows are directly in the hands of operators and their managers.

The benefits of self-service data analytics tools include increased process stability, reduced dependency on the often over-subscribed technology function, improved time-to-market – and the instant relief as capacity is recaptured through process automation. Removing the technology function from the critical path is an important end that has led to a raft of self-service and user-configurable tools spanning processing and reporting. Data democratization, or the widespread availability and accessibility of critical datasets throughout organizations, is a significant driver behind the growth of self-service data analytics, as operators sitting directly on top of business processes are well-placed to unlock data value if provided with the tools and capabilities to harness it. Perhaps chiefly, work that was previously unstructured, risky, and manual-intensive can now be structured in a tool, emulating system-driven processing. Laborious and time-consuming spreadsheet processing can now be easily replaced with tactical application-driven processing, leading to time savings and efficiency.

We all know that systems (“big” automation) are integral to nearly every job across nearly every industry, much as mechanization was a game-changer for manufacturing during the Industrial Revolution. In our digital and information age, where vast data is captured, stored, transformed, and processed for use to inform business decisions, core technology systems offer both coded processing and considerable mature technology governance, tied up with a bow and a ribbon. Systems are a central hub for structured development – offering data storage and processing operations, long lists of features and functionality, they offer user experience (UX) styles, and perhaps a rich reporting suite – but importantly, they quietly serve unnoticed as a centralized funnel around which to build internal controls and governance to ensure the accuracy and stability of processing.

The fact that systems are widely subscribed to in an organization ties many operators to enveloping control frameworks built around them. Thanks to mature system governance, many systems are designed with embedded internal controls such as those that ensure appropriate user entitlements, automatic system reconciliations and check totals to ensure data integrity, and perhaps even workflows providing for required supervisory approvals. Critically, many companies already enforce robust change governance around technology deliveries, including the mandated production of key project development artifacts such as documented requirements, evidence of test scripts and testing results, evidence of sign-off, and post-implementation reviews following a release. However, with the development decentralization that results from putting low-code, no-code tooling in the hands of users throughout the organization, governance frameworks built around systems will ignore an increasing segment of solution development and processing. This is one factor leading to a rift we are referring to in this book as the self-service data analytics governance gap.

This book is unique, because it is written not for data scientists or those with PhDs in statistics, mathematics, or neural sciences. It is not written for machine learning experts or AI specialists. This book is written for managers at large organizations, where there are pockets of operators – or likely entire functions – performing manual processes in spreadsheets. This edition will be valuable for managers that are increasingly under pressure to cut costs, produce more, and overall to do more with fewer hands. We will introduce process discovery methods to identify manual pain points and opportunities to capture efficiencies across functions, we will demonstrate how to size and cost these efforts, how to quantify the benefits cases, and how best to compare projects to assess priority. We will showcase prominent tools in action, as we take our readers through case studies featuring common use cases that they are likely to encounter in their own processing plants. The process consistency and stability benefits, which result as manual work done outside of systems is structured in tools, will be crystallized for the reader. Further, the reader will walk away convinced of the need for an overarching governance framework to protect the value created by the analytics portfolio. Our readers will surely have heard about AI and some of the disciplines within it, they may have heard about data analytics and understand that it is transformative, but they will benefit from a clear introduction to the language of data analytics, and they will benefit from simple demonstrations of emergent data analytics technologies. Finally, they will be armed with ideas on how to influence the approach, the course, and the speed of their organization's digital progression.

A subset of readers may be further along the digital transformation journey at their respective organizations. They have been exposed to common use cases for analytics tooling, they may have assembled a project portfolio, or perhaps even adopted self-service tooling across their organizations. These readers will benefit as the basics are clarified and reinforced, and they will appreciate the pains we have taken to present complex subject matter colloquially, as though we are explaining it to a trusted friend. Perhaps readers have already witnessed risks introduced by the unchecked proliferation of self-service analytics, in an environment where fragmented legacy governance frameworks better suited to system development have failed to close the coverage gap. In months of research prior to writing this book, your authors failed to identify an existing comprehensive governance framework suited entirely for data analytics or self-service data analytics portfolios. We resolved, therefore, to examine the existing body of frameworks and guidance and extend key principals to create and present an actionable governance runbook to optimize control over data analytics portfolios in this book – the first of its kind. In an era of increased accountability to internal auditors, external auditors, and regulators, we want to provide readers with the means to protect portfolio value through risk identification and mitigation and prepare you to meet the inevitable scrutiny head-on.

In the following chapters, we will take a practical approach to familiarizing readers with data analytics technologies and practices: we will discuss the rise of big data, whether from the internet of things (IoT) or the increasing body of digitized text from optical character recognition (OCR) or intelligent character recognition (ICR), the large data arrays that can be interacted with through natural language processing (NLP), subscribed to with ready-made application programming interfaces (APIs), and visualized with customizable interactive dashboarding solutions. From there, we will move on to explain the ways processing has evolved to digest the increasing body of available data. We will highlight decentralized, crowd-sourced distributed ledger technology (DLT) and blockchain, as well as cloud computing as a shift in the consumption model for storage, data subscription, and software. We will define AI as the nebulous pursuit of technologies that may ultimately emerge to emulate human thinking, capabilities, and interactions, encompassing all of the disciplines above – and more.

Having provided a backdrop to ease readers into our key topics, we will discuss data analytics disciplines that are rapidly evolving to supplant routinized manual spreadsheet operations. We will walk through robotic process automation (RPA) use cases and describe how RPA is being widely adopted to capture efficiency for (repetitive and stable) data capture and manipulation. We will briefly touch on machine learning and predictive analytics as advanced areas of analytics. Likely, the most immediate opportunity to transform processing will come about from the widespread adoption of self-service data analytics, which will emerge as a theme throughout this book. Putting low-code/no-code capabilities directly into the hands of process owners to automate extract, transform, and load (ETL) steps and to perform formulaic calculations, extensions, and comparisons will lead to a growth pattern and a pace of change never before encountered in the legacy system development environment. We will further devote a chapter to demonstrating the use of Alteryx, an increasingly prevalent tool that is rapidly transforming data analysis, processing, and reporting, to bring you firmly aboard the journey.

You may wonder just where to find promising opportunities to deploy data analytics tooling in your own organizations. We will propose methods for surveying processing operations to uncover use cases with significant benefits warranting prioritization and investment, matching tools to opportunities, and deriving an achievable digital roadmap. We will discuss the levers that can be pulled to increase control and drive efficiency, and decisions that can be made to adapt the organization to expectations that routine processes can be accelerated and streamlined to build capacity, unlock value, and to enhance decision-making. Finally, we will get you thinking about how to prepare your organization for the impending seismic shift, by establishing key governance procedures now, to maintain control as your organization adopts self-service data analytics and business intelligence tooling.

One key contribution of this book to the body of data analytics literature and discourse is our broad-based digest of existing governance frameworks and the systematic revelation of a significant governance gap that must be addressed, as data analytics saturates our respective organizations. There are any number of relevant and overlapping frameworks that cover portions of IT governance and even portions of data analytics governance in the finance and accounting environment. However, no single framework exists that is fit for the universe of self-service data analytics builds. We will draw from mature system governance, model governance, data governance, process governance, SOX 404, COSO IC (internal control framework for the financial reporting process), COSO ERM, and COBIT 2019 (ISACA) frameworks, and even the AICPA’s Statements on Auditing Standards – and distill fundamental governance principles that can be extended more suitably to govern data analytics portfolios. Building on such principles, we will call out recommended action steps that will help managers to satisfy them. Such action steps will be presented in the form of two checklists that readers can take back to their own organizations and implement straightaway. This must be done early and determinedly, so it is in place and can play a formative role in safeguarding your organization, as it embarks on what we believe to be an inevitable digital journey.

No matter which segment of readers you fall into, our goal in writing this work was to present you with a plain-language digest of the mosaic of developing analytics disciplines that warrant your attention. All pretense aside, we offer you our viewpoints and perspectives, with all deference to the insight and views of others that are concurrently embracing these same forces of change. We acknowledge that we cannot get our arms around the growing body of data analytics technology in this work of several hundred pages, alone. It can be fairly said that in the pages to follow, we barely scratch the surface. However, our goal is not to provide a comprehensive treatment of an extremely broad subject area that advances in leaps and bounds day by day. Rather, we set out to distill for you the significant developments in the space, to provide detailed discussions of key topics relevant to universal management goals of control and efficiency, and to empower you to succeed as the imminent data analytics–led digital transformation overtakes us. Best wishes for the journey!