OVO
Designing a new internal tool to help customers and call center agents diagnose problems with Gas and Electric meters.
Energy customers can experience a variety of issues with their gas or electric meters, from broken buttons to faulty billing or even total loss of service. In these situations, the first point of contact is usually the energy provider's customer care team. I was brought on to design an internal tool for OVO that would streamline the process for their customer care agents, allowing them to diagnose issues more effectively and book emergency or maintenance jobs for engineers.
The Challenge
At OVO, customer care agents were using up to eight different tools to troubleshoot and diagnose meter issues. This fragmentation led to inefficiencies, errors, and frustration for both agents and customers. Inexperienced agents can struggled to find the correct information, or ask the right questions which often leads to delays in booking jobs or engineers being ill-prepared for the situation. The overall experience was cumbersome for agents, leading to longer call times and sometimes incorrect job bookings.
Research and discovery
To better understand the challenges agents faced, I conducted user interviews with customer care agents and set up frequent call shadowing sessions to observe their workflow in real time. I also analysed data on common meter issues and job booking volumes to identify key friction points.
Working closely with the product team, we uncovered the most critical pain points:
To align the team and stakeholders around a clear path forward, I developed high-level personas and journey maps, which helped frame discussions and refine our approach. The recommended solution was to design a single-platform booking tool with an integrated problem diagnosis feature, allowing agents to triage customer issues more effectively and ensure the right job was booked the first time.
Working closely with the product team, we uncovered the most critical pain points:
- Agents were spending too much time switching between multiple tools.
- Booking the correct job was difficult, leading to inefficiencies and repeat visits.
- Engineers often lacked the necessary details to complete jobs efficiently.
Design Process
This was a tight time-bound project. The team focused on delivering incremental value, aiming to build and deploy an MTP (Minimum Testable Product) within the first three months. This allowed us to test the tool with a small group of agents early on, gathering critical feedback to ensure it accurately guided agents to the correct job bookings.
We broke the project down into two main phases. Phase 01 was to 'Find the right job'. This began with:
Wireframing & Prototyping:
Super low-fidelity wireframes helped us to map out the flow and key features, allowing some experimentation with different layouts and interactions. We built a basic prototype and ran some user testing with agents to get early feedback and build confidence that we were on the right track. With some confidence that the structure and layout of the tool was correct, we moved on to focus on how the 'booking guide' concept would work.
Defining information architecture
Creating a 'Booking guide' was a major challenges. We needed to ensure the agents would ask the right questions, guiding a conversation to the correct job booking, or problem resolution for the customer. There are around 60 different job types, each with different troubleshooting steps and details. I spent several weeks in close-collaboration and co-design sessions with metering subject matter experts, agents, and engineers to design a flow that was intuitive, streamlined and accurate.
Collaboration with content designers
From the outset, I established a design principle that the language used within the tool needed to be simple and customer-friendly, with a strategic view of re-using components within customer facing journeys in the future. Metering is a complex field, and translating technical jargon into user-friendly language required multiple rounds of collaboration with a content designer. Together, we ensured that the language was both clear and compliant with necessary regulations.
Design and prototyping
Once we had a clear flow, we moved quickly to create design concepts and prototypes for testing with agents. I organised regular user testing sessions using role-playing techniques to replicate real customer scenarios, enabling us to test the tool’s effectiveness in real-world-like conditions.
Pilot groups
To assess the tool’s performance in a live environment, we set up a pilot group of 20 agents. Over a two-week period, they used a minimum testable product (MTP) version of the tool during customer calls. We gathered feedback using a combination of Mixpanel data, shadowing sessions, surveys, and user interviews to continuously assess the tool’s impact and identify areas for improvement.
This concluded Phase 01.
We broke the project down into two main phases. Phase 01 was to 'Find the right job'. This began with:
Wireframing & Prototyping:
Super low-fidelity wireframes helped us to map out the flow and key features, allowing some experimentation with different layouts and interactions. We built a basic prototype and ran some user testing with agents to get early feedback and build confidence that we were on the right track. With some confidence that the structure and layout of the tool was correct, we moved on to focus on how the 'booking guide' concept would work.
Defining information architecture
Creating a 'Booking guide' was a major challenges. We needed to ensure the agents would ask the right questions, guiding a conversation to the correct job booking, or problem resolution for the customer. There are around 60 different job types, each with different troubleshooting steps and details. I spent several weeks in close-collaboration and co-design sessions with metering subject matter experts, agents, and engineers to design a flow that was intuitive, streamlined and accurate.
Collaboration with content designers
From the outset, I established a design principle that the language used within the tool needed to be simple and customer-friendly, with a strategic view of re-using components within customer facing journeys in the future. Metering is a complex field, and translating technical jargon into user-friendly language required multiple rounds of collaboration with a content designer. Together, we ensured that the language was both clear and compliant with necessary regulations.
Design and prototyping
Once we had a clear flow, we moved quickly to create design concepts and prototypes for testing with agents. I organised regular user testing sessions using role-playing techniques to replicate real customer scenarios, enabling us to test the tool’s effectiveness in real-world-like conditions.
Pilot groups
To assess the tool’s performance in a live environment, we set up a pilot group of 20 agents. Over a two-week period, they used a minimum testable product (MTP) version of the tool during customer calls. We gathered feedback using a combination of Mixpanel data, shadowing sessions, surveys, and user interviews to continuously assess the tool’s impact and identify areas for improvement.
This concluded Phase 01.
The solution
The final design consolidated the troubleshooting process into a single, user-friendly interface. Key features included:
Unified Dashboard
A central hub displaying all essential customer information, including account details, meter data, and job booking history—giving agents the context they need at a glance.
Step-by-Step Diagnosis Guide
A guided troubleshooting flow that helped agents systematically diagnose different types of meter issues, ensuring consistency and accuracy.
Seamless Booking System
Job details were automatically generated based on the troubleshooting steps, reducing manual errors and ensuring engineers had the right information before arriving on-site.
AI-Ready Self-Serve
The information architecture was designed with future automation in mind, ensuring a seamless transition to AI-driven self-service. By structuring the troubleshooting flow in a clear and intuitive way, the tool set the foundation for automating common diagnoses—enabling customers to resolve issues independently without needing to contact an agent.
Unified Dashboard
A central hub displaying all essential customer information, including account details, meter data, and job booking history—giving agents the context they need at a glance.
Step-by-Step Diagnosis Guide
A guided troubleshooting flow that helped agents systematically diagnose different types of meter issues, ensuring consistency and accuracy.
Seamless Booking System
Job details were automatically generated based on the troubleshooting steps, reducing manual errors and ensuring engineers had the right information before arriving on-site.
AI-Ready Self-Serve
The information architecture was designed with future automation in mind, ensuring a seamless transition to AI-driven self-service. By structuring the troubleshooting flow in a clear and intuitive way, the tool set the foundation for automating common diagnoses—enabling customers to resolve issues independently without needing to contact an agent.
Results and impact
The new tool is still in development and is set to launch in late 2025. It is projected to generate up to ÂŁ6 million in savings through improved job bookings, a reduction in aborted jobs, and a reduction in customer complaints.
Beyond cost savings, I am also focused on measuring the impact on agent experience. This includes tracking quantitative metrics - such as time saved per booking - and qualitative insights, like agent job satisfaction and confidence in diagnosing issues. These measures will help ensure the tool not only improves efficiency but also enhances the day-to-day experience of those using it.
Beyond cost savings, I am also focused on measuring the impact on agent experience. This includes tracking quantitative metrics - such as time saved per booking - and qualitative insights, like agent job satisfaction and confidence in diagnosing issues. These measures will help ensure the tool not only improves efficiency but also enhances the day-to-day experience of those using it.
