Increasing maintenance efficiency
Updated: Sep 17
Ocado’s automated warehouses are populated with hundreds of robots that do much of the heavy lifting. They are monitored and maintained by teams of support staff trained to diagnose and recover faults. Here I describe the process and IT changes that increased their productivity and reduced costs.
Background
When robotic picking arms were introduced to Ocado warehouses in 2025, their troubleshooting process was different compared to other types of robots. Being a relatively new product there were several inefficiencies and opportunities for improvement. First, the arm’s original diagnostics tool was not easy to learn and use, leading to human error and lower productivity. Second, the complex coordination between teams meant resolution often took longer than necessary.
Usability problems
Usability problems with the diagnostic tool were found after shadowing sessions with several support staff. The tool was full of confusing technical jargon that were designed for the hardware engineers rather than the support staff. Not only did it cause confusion, it also made communication between support staff difficult and inefficient. Furthermore, some terminologies had ambiguous meanings, making it difficult to decide the correct action to take. Lastly, the user interface did not prevent potentially dangerous actions from being performed, increasing the likelihood of failure due to human error.
Usability solutions
We made a number of changes to fix the usability problems. Many technical jargon words were replaced with more accessible terms. Ambiguous terminologies were separated into clear and distinct labels. Certain UI controls were made active depending on context, which prevented dangerous actions from taking place. Information was organised to fit more closely with the users' mental model, increasing ease of learning.
Coordination problems
The inefficient coordination problem was discovered by working with stakeholders and support team managers. By mapping out their current process and looking at analytics data, we were able to find the root causes. Often one team member would wait several minutes for another team member to complete an action or provide some information. In many cases, extra clarification was needed because the system did not provide a way to share the information in a clear way. The need to manually type instructions was another source of wasted time.
Coordination solutions
We redesigned the system so that clear information could be easily generated and sent to another team member. The full task instructions were clearly displayed on their screen. Standardised task types and instructions meant there was no longer a need to enter instructions manually. Finally, the workflow was redesigned to eliminate waiting time, allowing team members to stay productive while the task was not ready to resume. Their total time spent resolving an issue was reduced by approximately 70%.
