Client: PETRA Data Science
Date: 2020-2024
Type:UX, UI, SaaS Product
This feature was for an AI based simulator for how geology can impact certain mining processes.
Optimal operating decisions are learned by the software and show engineers how to achieve their best day every day.
In particular, MAXTA Drill&Blast is purpose built to allow engineers to create bulk blast design patterns using AI.
Drill and blast affects load & haul, crusher and mill performance as well as materials handling and energy efficiency.
Boost confidence in Ai
The Challenge
Sometimes AI insights are hard to sell and considered high risk. Once a drill & blast prediction is made, how could we allow a user to compare different AI/ML models to see the trade-off between different outputs following different blast designs?
Find a way for the user to see predictions in a way that allows them to compare different scenarios such as Cost vs Energy Consumption, Throughput vs Energy Consumption, Fuel/Diesel vs Digrate
Believing there is a solution
Hypothesis
We believed that giving the user a greater degree of comparison as well as purposeful data visualisation would give extra dimension to the predictions and also increase confidence in using our software, in turn reducing risk.
Facilitate a design sprint
Workshop
We conducted group workshops with around 12 users from 3 separate personas. Attendees were selected given their hands-on engineering and geological backgrounds.
We introduced the hypothesis and presented the user with different solutions to the problem in the UI.
I facilitated and group prototyping sessions (short design sprints). During the workshop we designed and exchanged simple interfaces built in Power BI with the help of our data scientists. We captured a series of ‘how might we statements’ and as a group were able to vote on the most tangible solutions. These were conducted face to face.
We obtained qualitative data surrounding the users beliefs, reactions, behaviours and motivations and chose which solution to implement for initial testing.
Creating & testing
Protoyping
I designed a prototype in Figma with all possible micro interactions, which allowed our developers to then create a proof of concept which we rolled out for live testing and feedback.
Qualitative Data
Analysis
We included feedback requests/pop ups in the UI and obtained useful feedback.
Through analytics and heat mapping, quantitative data was obtained so that it would validate our hypothesis.
We conducted before and after user journey mapping.
We connected to New Relic for Real User Monitoring and observed user interaction.
We also conducted post-deployment interviews for feedback and validation
Significant behavioural acceptance
The Results
More than 400 users onboarded successfully.
Trends showed that users spend 30% less time the tabular view of the drill & blast predictions, but opt to hang around the chart view, exploring various model trade-offs and obtaining further insight equating to more than 12 minutes spent on that screen only.
We’ve also been given feedback that drill & blast predictions alone have contributed to over $460million dollars of uplift in EOFY2023.
Further feedback has given us insight that users are more confident and are looking at ways to now validate the predictions in real time.
