Building a robust and efficient path to production for cross functional teams to deliver applications
User Journey Mapping and Business Improvement
How might we build a paths to production framework that allows for cross functional teams to deliver applications in the quickest, cheapest and most cohesive manner.
The Challenge...
Within the CDO(Chief Data Office) a critical problem that nbn is facing is the risk of having Shadow IT teams working on business critical models. In order to mitigate this, we needed to map out an end to end process to allow data science models to be productionised with the appropriate risk and audit measures in the System Engineering and Operations(SEO) chapter.
Our approach
-
A define phase to...
-
to set the scene and understand what the current state journeys are and their pain points and opportunity areas
-
understand the purpose of creating one singular path to production
-
-
A research phase to...
-
understand what the target state for this project is
-
form an underlying goal statement for the workshops to come
-
-
A workshop synthesis phase to...
-
Map out a new path to production with colleagues across CDO and SEO (Chief Data Office and Systems Engineering & Operations)
-
Co-design what the target state process might look like
-
Capture any risk and audit requirements to form as guardrails for the process
-
My Role
My role in this engagement was to conduct design research, conducting interviews, facilitating workshops and managing stakeholders . The other members of my team included another UX Designer.
Define
Stakeholder Engagement:
This piece required involvement from five different chapters within nbn, with each of our stakeholder being the General Manager of the chapter. It was crucial to set up a cadence that we could update them as well as define methods of doing so. For this particular piece we provided weekly email updates, monthly stakeholder meetings as well as a Confluence page to document all our work.
Unpacking The Current Experience:
-
Run a Design Thinking workshops to collect experiences, theme and vote on them and to prioritise them
-
Interviews with each General Manager from Engineering, Architecture, Operations and CDO to break down further the background, problem statement and identify "what good looks like"
-
Finding use cases and running a series of workshops to map out current state journeys that will aid in developing the future state journey
Shaping the Future Development Experience:
-
Identify risk and audit requirements that will help carve out key milestones in the journey
-
Weave in solutions to any pain points the current journey is facing
Research
Focus:
Capture pain points in the methods in which a data science model is productionised and to find out key milestones that will need to be incorporated in the future state journey.
Purpose:
To build a paths to production framework that allows for the capability and requirements to drive which incubation or strategic environment is needed, for the quickest and most efficient delivery model.
The process of defining the purpose was not a straightforward one, it required several iterations and analysis into what the the goal was and what we wanted to achieve.
This led us to creating an anchor overarching statement as well as three other statements that were underpinned by the broader one.
OVERARCHING STATEMENT
How might we build a paths to production framework that allows for the capability
and requirements to drive which incubation or strategic environment is needed, for the
quickest and most efficient delivery model.
PEOPLE FOCUSED
How might we build a collaborative culture underpinned by trust that motivates our people to innovate or fail fast to unlock business value quicker.
DATA QUALITY
How might we release business capability faster that we are confident has come from accurate and reconciled data. So that the organisation can confidently make decisions from traceable insights.
FUTURE PROOF
How might we build a framework that has a clear path for outcomes, production and decommission use cases.
Workshop
Significance:
This process defined the course of this research and also identified three other paths for a data science model, these included a path to; an outcome, production and decommission.
Outcome:
The outcome of this research piece was a decision tree, three current state journey maps and feedback from our working group.
As there were no existing journey maps that we could reference off of, it required 2 workshops per journey to map out the end to end process.
Adding a new data set to the model
Responding to business demand
Responding to a consumers request
Adding a new data set to the model
Through the conversations to build the current state journey, pain points and opportunity areas were captured.
Current state decision trees and a desired state decision tree was provided by the working group to find the gaps between the two and to come up with a journey that would fill those gaps and create a desired state.
Pain points and Opportunity areas
Difference between current and desired state
Target State Journey
Focus:
Utilise captured pain points and opportunity areas that were identified in the research phase to map out a future state journey.
This process was by no means linear, it required several iterations due to changes in business demands, organisation restructure and time constraints. Since our working group involved so many business units, it was important to find ways to best use the time we had with them.
We mapped out each segment of the journey with the involved business unit before bringing together the wider working group to discuss connect points and to walkthrough the overall flow of the journey to best utilise the limited time we had with them.
The final target state journey was a combination of the two journeys that we mapped out, these journeys were then used to aid the business transition their shadow IT teams into ones that could be properly regulated and audited, ones that were only required for experimental purposes and ones that could be decommissioned.
