Data Visualisation as a Design Discipline: Lessons from ESB

The ESB Real-Time Visualization Platform won Best Data Visualisation at the IDI Awards 2023 and was benchmarked by Microsoft for its design approach. What made it distinctive wasn't the technology stack — it was the decision to treat data visualisation as a design discipline before treating it as an engineering problem.

The Problem Data Visualisation Usually Solves

Most data dashboards are built to answer the question: 'How do we display this data?' The better question is: 'What decisions does this person need to make, and what data do they need to make them confidently?'

At ESB, station managers and energy traders needed to monitor real-time output across multiple generating assets in Ireland and the UK, track performance against financial targets, and identify anomalies that required immediate action. The dashboard wasn't a reporting tool — it was an operational command interface. Every design decision flowed from that understanding.

Hierarchy Over Completeness

The temptation in any data-rich environment is to display everything. The right approach is to display the right things at the right level of prominence. On the ESB platform, we established a strict information hierarchy: critical operational alerts at the top, near-real-time performance metrics in the primary view, and detailed historical analysis accessible but not dominant.

This hierarchy was validated through observation sessions with actual plant managers and traders — watching how they scanned the screen, what they looked for first, and where their eyes naturally landed. The AI saliency model we used for product testing confirmed and refined these observations with objective attention data.

Colour as Functional Language

In data visualisation, colour is a functional tool, not a decorative one. On the ESB platform, every colour choice had a specific semantic meaning — green for normal operating ranges, amber for approaching threshold, red for alert conditions. We enforced this discipline ruthlessly: no colour appeared in the interface that didn't carry functional meaning.

The result was a visual language that experienced users could read in seconds, without needing to parse labels or legends, because the colour system did most of the communicative work.

The ETL and Data Architecture Context

Great visualisation design is only possible when the data architecture beneath it is sound. On ESB, I worked closely with the engineering team on the Data Warehouse and ETL pipeline design — because if the data is arriving inconsistently, with latency, or in incompatible formats, no amount of design skill can produce a reliable operational dashboard.

The investment in a robust data foundation was as much a product decision as a technical one.

‍