Lior Barak, a speaker, data strategist, and author of Data is Like a Plate of Hummus, spoke on the Data Engineering Podcast about using a data vision board to drive strategic planning. Barak argued that data teams are often a costly afterthought in organisations’ data-centric strategies and proposed a collaborative 3-year roadmap to ensure alignment, evolution, and delivery of impactful capabilities.
Barak explained that data engineering teams are generally focused on enabling other teams’ initiatives and become disconnected from the value delivered by their data. Barak shared his experience of data engineering teams often being reactive, resulting from their being measured on KPIs targeted at immediate priorities. Given the cost and latency of data engineering solutions, this would frequently result in falling short of demonstrating a longer-term return on investment. Speaking of typical planning for the immediate, one-year horizon, Barak said:
We talk about the processing engine not being fast enough and that things are crashing for the analysts … We’re looking very much at what our problems are today (and) don’t talk about the future. We don’t talk about how data is actually shaping the course of the organisation … Marketing needs to have data for a campaign to steer it and steer its budget … (we should ask) what is the actual impact? … Or maybe we should talk about automation or giving them a revenue forecast that’s gonna help them actually to understand what is the value of a user in 20 to 60 days.
Gaëlle Seret, group product owner at Decathlon Digital, spoke at the Paris Data Ladies meetup in October about Organizing Data Teams for AI Success. Seret talked of her “data as a product approach,” and emphasised that her internal teams are all treated as customers, including “data analysts and AI.” Seret’s approach involved frequent collaboration with “sources,” who she defined as business specialists, data producers, data consumers and “data quality defenders.” Seret said that the most important key to a platform’s success is collaboration, and being “servant leaders” who understand and drive the success of customers’ strategies.
Barak suggested starting the elicitation process by having teams and stakeholders define the present using a shared document or Mural board to outline their needs, gaps, and concerns. This information is then analysed to identify connections, trends, and stakeholder relationships, feeding into a long-term strategy.
Barak proposed the use of his Data Ecosystem Vision Board to align strategy with the business and other stakeholders. He recommended building a 3-year strategy for Data Engineering, with intentional collaboration across stakeholder groups, from teams to business stakeholders. Barak recently blogged about the use of Vision Boards, describing an approach to moving away from strategy workshops which “never look beyond current problems.” He proposes building a strategy using workshops and collaboration to define:
- The present layer of data use, data pains, ethical concerns, cost and revenue potential.
- The future layer is based on big-picture goals, the resources and capabilities required, and a gradual path that iterates through clear milestones.
- Success metrics derived from ROI and revenue expectations, cost targets, and the impact and utilisation of data.
Barak presented the vision board as a lean canvas which considers these three states as seen below.
Seret also stressed the need for alignment on a “core vision” co-written with stakeholders, defining goals and acknowledging the differing definitions of metrics and domain language across teams. Using “weight” as an example with multiple meanings, she highlighted the importance of defining concepts in business language to prevent confusion. To build a shared taxonomy, she selected a low-hanging product area and delivery, and then assembled experts from across the business to define it. Seret explained that the taxonomy’s evolution was an incremental process, iterating through the selection of specific projects and specialist groups to refine it. She said:
As a data product manager, you are going to think ‘Who can I onboard in my expert Community and what is the first easiest product that can bring value?’ … When customers start consuming the data you will bring more experts and maybe you will change the taxonomy that you built initially. This taxonomy is going to evolve with the data strategy of your sources … Your taxonomy is going to live with your company and it’s going to evolve with the more experts that you bring on.
Barak and Seret advocated for strategic shifts toward proactive, long-term, business-focused strategies for data engineering teams. Barak highlighted the pitfalls of reactive approaches, where cost, impact, and urgency misalign. Seret redefined data product managers’ roles, emphasising deep domain knowledge, usability, discovery, operational stability, legacy decommissioning, and observability. She also stressed the importance of building the right long-term capability, saying, “a robust core data-product foundation fuels analytics and AI capabilities.”
Describing the need to define and measure overall business impact, Barak wrote:
The success metrics layer can change everything. It gives us a way to evaluate initiatives based on their impact, ensuring we focus on what truly matters. By introducing clear indicators of success, we can embrace imperfection, find joy in the journey, and above all, discover the middle way to move forward.
