0–1 Product Design · Chart Framework · Interaction Design

Designing a data exploration system for IoT data

Transformed a meter-focused chart into a flexible experience for exploring device data and comparing multiple signals.

Problem

Utility teams needed to explore raw device data and correlate different signals, but the existing chart was built for meter-read validation.

What I designed

Designed a configurable charting system for time-series analysis across multiple data sources.

What changed

Shipped in NextGen MDM, with customers describing the capability as essential to their operations.

My Role

Sole designer, leading end-to-end design in partnership with 1 PM and 2 Engineers.

  • Defined the experience, design direction, and scope

  • Partnered with engineering on feasibility and trade-offs

  • Owned interaction design, UI specifications, and implementation support

Problem

With residential solar and EV charging creating new analytical needs, utilities needed better ways to explore consumption patterns and understand changing demand.

But the existing chart was designed for meter-read validation, not exploratory analysis. Raw channel data was buried in drill-down, and users couldn’t easily compare it with other signals.

Key issues

  • Chart is built for meter-read validation, not data exploration

  • Raw data is buried inside drill-down

  • No flexible way to correlate with other signals

Legacy Chart Workflow

1

Select channels.

2

In the Long-Range chart, select an area to zoom in based on Read Status and Device Events.

3

Zoomed chart is populated with actual channel readings selected from step 1. (E.g., KWH 15 Interval Read, KWH Total Cumulative Register Read, KWH Usage Read)

Solution

Utilities have different analytical needs depending on their data setup, commodities, and business priorities. A fixed charting experience would require ongoing customer-specific development to accommodate these differences.

I designed a configurable charting experience that lets users choose what to visualize, compare multiple signals in context, and reuse chart configurations across records.

Visualize raw data

Users can browse available data channel information and select which ones to visualize on the chart.

Compare signals in context

Users can compare multiple data series on a shared timeline to explore patterns and relationships across signals.

Customize and reuse charts

User can customize chart settings, save configurations, and reuse them to analyze other records.

Iteration & Key Improvements

A. Improving chart readability

Multiple data series on a single chart would make it difficult to read. Alternative is to create separate charts for each series, but that will take too much space and makes comparison harder.

I introduced synchronized chart with 3 slots, allowing users to group related series while maintaining clarity and reducing cognitive load within the limited chart space.

Overlaid two data series in a single slot for direct side-by-side comparison (diff).

Overlaid two data series in a single slot for direct side-by-side comparison (diff).

Each dataset gets plotted in its own separate slot.

Each dataset gets plotted in its own separate slot.

Using all three slots to plot 4 different datasets.

Using all three slots to plot 4 different datasets.

B. Supporting different types of analytics

To effectively analyze trends or resolve operational issues, users cannot look at consumption data in a vacuum. They must cross-reference and correlate with various other data sources.

  • Historical data (Time-shifted)

  • Profile data

  • Event data

  • Weather

  • Compare data across multiple service points

Each data type behaves differently, so I designed distinct interactions for each type based on what made most sense.

Adding time-shifted data

Adding time-shifted data

Adding profile data

Adding profile data

Compare across multiple service points

Compare across multiple service points

C. Making event data meaningful

My initial design focused on whether events had occurred and how many were detected, but this didn’t help users understand the types of issues occurring. Based on user testing, I redesigned the visualization to group events by category and added filters, allowing users to identify relevant issue types.

Process

I partnered closely with PM through multiple rounds of ideation to define key hypotheses and user scenarios.

Brainstorming sketches of how to select data and save a chart as a favorite.

Brainstorming sketches of how to select data and save a chart as a favorite.

Before jumping into the UI, I collaborated with engineering to align on the underlying data models. I created diagrams to map our high level decisions about what components we need, and how data is passed across the UI.

Next, in order to define the capabilities of the data visualization experience, I established a set of design principles and system-level guidelines.

Data Visualizer (Core Experience)

The data visualizer is designed for time-series analysis, where the X-axis represents dateTime and the Y-axis represents measurement values.

  • Users can plot multiple data series within a single chart to enable comparison across data streams

  • System limits are based on data volume and performance, rather than the number of selected series

  • Each data series is individually configurable (e.g., color, chart type), supporting flexible analysis workflows

  • Date/time formatting is controlled at the global level, ensuring consistency across views while allowing user preference

Chart Behavior & Interaction Model

To ensure consistency and usability across different chart use cases, I defined default behaviors and interaction patterns:

  • Chart titles are handled outside the visualization, allowing greater layout flexibility and clearer hierarchy

  • Legends can be hidden and restructured as a selectable list, improving usability in multi-series selection

  • Zoom is enabled by default, with a zoom reset button to support exploratory analysis

  • A navigator is included to provide temporal context and enable efficient range selection

  • Tooltips are enabled by default to support data inspection, with flexibility to disable them for simplified views (e.g., compact trend charts)

With these decisions in place, I moved into wireframes with the goal of gathering feedback early on.

I evaluated the charting library’s capabilities and worked with engineering to expose the controls that are most relevant to users.

We used configuration and data mapping approach using a 3rd party chart library.

We used configuration and data mapping approach using a 3rd party chart library.

“

This is instrumental to our business. We wouldn’t be able to operate without it.

–– Utility Customer, Denmark

Outcomes

Due to the on-premise nature of most customers, direct usage metrics were limited. Instead, impact was assessed through stakeholder validation, customer evaluations, and sales pipeline.

  • Data Visualizer released in NextGen MDM

  • Strong customer validation

  • A foundation that can support additional use cases

Potential Applications

Energy demand forecasting

  • Predict peak demands and grid overload

  • Analyze consumption trend

Renewable energy integration

  • Forecast solar generation and smooth out supply fluctuations

Usage analytics

  • Plan for more revenue options with different pricing

Water network efficiency

  • Run mass balance analytics to detect water leaks in hidden pipes

Predictive asset maintenance

  • Spot aging and low-performing assets for timely repair