The Enabling Data Model and Manifesto

Abstract

Advanced analytics use cases and genAI are attracting new resources to pushing value out of data. Still, many stumbling blocks remain before successful value delivery. We propose a new holistic view of data and a data-first manifesto to address them. These two concepts can be used to a) explain the complexity of data to the business and b) guide a first-time data leader in designing a successful portfolio of enabling data activities.

Keywords: data strategy; data governance; operating model; organizational design

We now operate in a deluge of data. So much so that an argument can be made that we have data. The hyperscalers provide ready and relatively cost-efficient services to do both data engineering and advanced analytics at scale from the start . Universities and bootcamps regularly upskill talented and educated future employees to implement many new use cases . So why do we still see so little value generated from data? Why are AI models in production so rare? Why do we have so many stories of interesting pilot projects, yet the few success stories of large-scale products are limited to technology giants?

We argue that the main reasons are a need for holistic understanding of what is needed to achieve value (data education) and a set of principles to guide the bridge between strategy and execution. Here, to address those two issues respectively, we propose the and the . This work is intended as reference material for a business audience since data leaders are commonly aligned on the topics.

Enabling Data Model

In analytics, we recognize the spectrum of activities between data producers and consumers. We argue that we need to shift resources and attention - closer to the producers. This dichotomy can be visualized as an iceberg, where the obvious, attractive use cases (such as advanced analytics and generative AI) are above the water. At the same time, the bulk of the work is below - invisible. This is presented on the x-axis in the model below:

The Enabling Data Model

Before any work on data is done, a clear understanding of business goals (for example, by creating a Value Stream Map), the subsequent strategic initiatives and baseline metrics (with a KPI value tree connecting both) are essential. With the addition of a maturity assessment, we can then set up the data initiatives in four pillars - People, Technology, Governance, and Data. Many of the initiatives can have an overlap. For example, a data catalog spans data, governance, and technology. The tip of the iceberg use cases require different configurations for successful execution and some might have a shorter path than others. For example, a dashboard often has fewer requirements and dependencies than a customer churn model. This can be seen as a visual representation of technical feasibility. Finally, the use cases feed into the aligned baseline metrics set at the start in a feedback loop.

The EDM allows for a) stable requirements, dependency tracking, and not reinventing the wheel, b) Portfolio management of “shiny” use cases and “invisible” data initiatives, c) dynamic tracking of business performance and impact, d) opening the possibility of a working backwards approach and e) visualizing a balanced approach ensuring value delivery of data.

Manifesto

Here are the manifest points, in no particular order:

  • If your use case is not measurable, you don’t have a use case
  • Stop measuring everything: start measuring what matters
  • Beware of pilotitis (but do experiments)
  • You have no right to ask for a budget unless you know how you contribute to the value chain
  • Start with the end: work backwards
  • Your new framework is probably a cargo cult
  • Preach data, but not to the choir
  • If everyone owns the data, no one does
  • Tech is never the problem: you are
  • If data is not an asset, it is a liability
  • Scalability, performance, and cost: choose two
  • Buy, don’t build (unless you can afford it)
  • Data modeling is more important than anything (do it as early as possible)
  • Balance offense and defense in your projects

Outlook

By using the model and manifesto, we aim to help data leaders communicate better with business managers so that we finally get value from data.

References

Angelov, B. (2024, July 16). 2 Essential Strategies for CDOs to Balance Visible and Invisible Data Work Under Pressure. . https://www.cdomagazine.tech/opinion-analysis/2-essential-strategies-for-cdos-to-balance-visible-and-invisible-data-work-under-pressure

Angelov, B. (2021). Elements of Data Strategy: A Framework for Data and AI-Driven Transformation

Ates, A., & Suppayah, K. (2024). Disciplined Innovation: A Case Study of the Amazon Working Backwards Approach to Internal Corporate Venturing. Research-Technology Management, 67(3), 23-33.

DalleMule, L., & Davenport, T. H. (2017). What’s your data strategy. Harvard business review, 95(3), 112-121.

Floerecke, S., Ertl, C., & Herzfeldt, A. (2023). Major drivers for the rising dominance of the hyperscalers in the infrastructure as a service market segment. International Journal of Cloud Computing, 12(1), 23-39.

Rawlings-Goss, R., & Rawlings-Goss, R. (2019). Building Data Careers. Data Science Careers, Training, and Hiring: A Comprehensive Guide to the Data Ecosystem: How to Build a Successful Data Science Career, Program, or Unit, 5-30.