better, faster, smarter
how might we help dispatchers schedule the most optimal deliveries by simplifying complex decisions?
route scheduling for fuel deliveries at this oil and gas company is currently carried out by dispatchers using manual methods that rely heavily on their intuition and experience. the client approached us with an interest in applying decision optimization models to their existing data to provide dispatchers with driven insights when scheduling deliveries in order to capture new revenue opportunities.
throughout this mvp build, there was close collaboration between designers, data scientists, and developers to research user needs, build a back-end optimization model, and deliver a front-end interactive dashboard using real data.
6 weeks / confidential oil and gas company / design research, workshop facilitation, dashboard design
01_my role
during the research phase, i designed and led delivery of a series of workshops that allowed us to quickly understand the client's problem space, data constraints, user needs, and scope success measures for the mvp build. throughout this process, i identified common themes and ensured that discussions and materials produced could be understood by both the technical (data scientists and developers) and non-technical (dispatchers, business analysts and operations managers)participants from the client side.
during the mvp build, i led a team of designers to consider user needs in tandem with available outputs from the data model to design and wireframe a dashboard interface. the design vision was to both enable dispatchers to monitor their customers' fuel supply, and see recommendations for how best to schedule deliveries at a glance. this was validated through user testing, with one dispatcher noting that she is extremely excited to implement the tool into her workflow.
02_design research
through a series of four facilitated workshops, i led the client and internal team to create an understanding of recurring themes and reach consensus on how to best tackle opportunities. given that there were stakeholders representing management, operations, development, and data science, a large part of this process was focused on breaking down siloes and cross-pollinating ideas.
in parallel to the workshops i also conducted a series of 1-on-1 interviews with dispatchers (our main user) from ontario and the atlantic provinces around their workflow and tools. i discovered that while there were similarities such as the manual and repetitive nature of their tasks, inconsistencies in technology maturity, and how supply data is collected from their customers meant their needs were slightly different. for this reason, we focused on addressing common painpoints for the mvp build.
by the end of the research process, we had a nuanced understanding of the client's existing situation, identified areas of opportunity, and prioritized these opportunities so we had a scoped focus and clear success measures to guide the accelerated mvp build.
03_mvp build
i led a team of two other designers to translate insights from our user research into a dashboard design that would address the dispatcher's painpoints and needs. a key finding was that dispatchers' tasks can broadly be divided into two categories:
1. monitoring fuel readings and consumption trends to understand which gas stations will need re-fueling and when
2. scheduling fuel deliveries by assigning a truck to a route, and then assigning the relevant gas station fuel order per truck compartment
in this way, we designed a dashboard with two pages, the first with a layout and interactive visualizations optimized for monitoring tasks, and a second that suggests optimized routes and schedules for dispatchers. for the second page in particular, i worked closely with the data science team to understand what kind of data their prediction and optimization models would produce, and translated that into a layout that would give dispatchers all the information they'd need at a glance.
04_testing and outcome
once design was completed, i carried out user testing with dispatchers using clickable wireframes. feedback was overwhelmingly positive, and comments were used to improve the design for usability. for example, we originally had a general truck loading visualization oriented horizontally, but because this might confuse new dispatchers who might associate a specific fuel type to a particular compartment in the truck, we rotated the visualization to be stacked vertically.
the mvp build was completed at the end of the 3 week sprint, and based on our objectives and feedback, was considered a success. it was mentioned that the dashboard would enable the business to shorten the time it takes to train new dispatchers from 6 months to a matter of weeks.
continued work towards production will be completed by the client team working directly under the cio - as part of the handoff, i put together an experience roadmap to help guide the team in planning for subsequent user-driven development.