A Guide to Self Service: Understand, Plan & Implement
Written by Practitioners for Practitioners
One important hallmark of a data-driven organization is when business users meet their own data and analytics needs without IT assistance.
What does this look like in practice? Business users ask questions of data and get instantaneous answers; power users easily find, access, prepare, analyze, and visualize data for business consumption; data analysts and scientists create departmental reports, dashboards, and AI models from libraries of trusted data; and executives track performance using KPIs in custom scorecards and dashboards.
For the past 30 years, self-service has been the holy grail for data leaders—a cherished, but elusive goal. What seems simple and straightforward becomes an odyssey filled with pitfalls, dead ends, and frustration. For many data leaders, self-service backfires, spawning data chaos; only a few finish the journey and achieve self-service nirvana.
This Guide to Self-Service reflects the knowledge and frameworks we’ve developed over more than three decades of delivering data strategies for commercial and non-profit organizations around the world. Our Guide outlines the 18 assets to help you understand, plan,?and implement?self-service that empowers business users without creating data chaos.
Understand Self Service
Self-service is hard because it requires data leaders to balance the conflicting forces of empowerment and control. This is the primary focus of our e-book “How to Succeed with Self-Service Analytics: Balancing Empowerment and Control.” This guide steps readers through the pitfalls and paradoxes of self-service, provides a framework for classifying business users and understanding their requirements (see image below). Finally, it advises readers how to structure governance processes, architect data environments, federate teams, and select tools to facilitate self-service. I highly recommend it.
Silver service means the data team gives users the information and functionality they need to do their jobs i.e., deliver data on a “silver platter”. Self-service means users have the knowledge, means, and imperative to create reports and models from scratch to meet business needs.
The article “Self-Service is the Outcome, Not the Driver of a Data-Driven Organization” argues that self-service is not possible until data leaders design and deliver an effective data architecture, governance program, operating model, and product delivery method. Our blog series on self-service discusses pitfalls of self-service (“Self-Service Analytics: What Could Possibly Go Wrong?”); how to tailor self-service to business users (“One Size Does Not Fit All”); and how to create governed self-service workflows (“Curate, Create, Consume”).
Plan for Self Service
A crowning achievement for data architects and engineers is to devise a data environment than enables data-savvy domain users to create their own data marts, data pipelines, and/or data preparation workflows within the enterprise data architecture. The goal is to provide a baseline set of building blocks that business users can extend to support local applications without creating silos with conflicting semantics and metrics. The podcast with Jeff Magnuson of Stitch Fix recorded in 2018 on “How to Create a Self-Service Data Platform for Data Scientists” explores this important topic.
Operating Model.?More recently, we’ve discussed how it’s important for the data team to establish data engineering standards and educate business users how to implement them. The goal, according to one of our consulting clients, is to “deputize” business users who’ve received the training and tools to build their own data structures in conformance with enterprise standards. This is vital for a data mesh environment, but it’s critical in every environment where you want to empower data-hungry business units to meet their own data needs without creating data disharmony.
Our e-book titled “Operating Models for Data & Analytics: How to Align Resources Across the Enterprise” shows how enterprise data teams need to support domain developers through education, training, and coaching. It also shows how support and coaching trickle down from the enterprise team through various levels of data users and a data & analytics help desk. (See below.)
Implement Self Service
Data Architecture.?The report, “A Reference Architecture for Self-Service Analytics”, dives into how to create a data architecture that supports self-service. We’ve updated the framework in that report to simplify how our standard user roles map to architectural structures. (See below). A good companion slide deck is “Data Zones for All Consumers” (or watch the webinar) which explores how to design data processing zones within a data platform to support self-service use cases as well as other use cases (e.g., real-time, API access).
领英推荐
DAaaS. More recently, the e-book “Data Architecture as a Service: Empowering the Business to Build Compliant Pipelines” discusses how next-generation data transformation tools, such as Coalesce, enable data architects to embed standards into the development process, making it easier for domain users to build compliant workflows. ?
Power Users.?We believe the fastest way to succeed with data & analytics is to focus on power users. The report “Making the Most of Your Data Analysts: Strategies to Empower, Align, and Retain Them” shows the pivotal role of power users in a data strategy, and how they can make or break your data program. (See below). It also offers a checklist so you can see the effectiveness of your power users and a career pathway for data professionals in which power users play a central role.
Tools. We also believe the market for power user tools is evolving rapidly. The article “Self-Service Triumverate: The New Data Analyst Workbench” shows how new platforms are emerging to support power user workflows to discover and validate data (i.e., a data catalog), massage and prepare data (i.e., data pipelines) and visualize and analyze data (i.e., data visualization).
Newer tools, such as Promethium, add a data fabric with query federation and a GenAI interface, enabling non-SQL users to generate ad hoc queries across multiple systems without coding. See Promethium’s blog, “How to Simplify SQL with Text-to-SQL Technology”.
Data Governance.?Finally, self-service only works if the output of business users is governed. My podcast interview with Angie Davis?shows how one organization implemented a comprehensive governance model to ensure that new enterprise reports or changes to existing ones goes through an agile process to ensure conformity with enterprise standards. (See below).
Summary
Self-service is a journey, not a one-time project or a tool. To succeed with self-service, data leaders first need a deep understanding of their business users. Then, they need to implement the right data architecture, right data governance, right operating model, and finally, the right tools. Once that work is done, self-service can take root and blossom.
Need Help with Self Service? Talk to an Expert.
Browse our Insights
About Eckerson Group
Eckerson Group helps organizations get more value from data. Our consultants have 25+ years of experience in all facets of data & analytics. Learn how we can help your organization create actionable data strategies and highly tailored solutions.