As we embarked on this journey to build a prototype, we spent a lot of time thinking about the foundation and what we needed to establish to make sure that we had a solution that was robust and scalable. That started with data architecture and then modeling based upon the requirements that we had for reporting and analytics before we even looked at data ingestion or management and governance. But the data ingestion was the key area that Alteryx was involved in, in terms of bringing in multiple sources of data across the landscape of NBN's sales and marketing activity. And wrapping all of this, the solution was data management and governance which we partnered very closely with Chief Data Office of NBN to make sure that we were following best practices.
There are 3 phases of this journey: Prototype, Operationalize, Cloud Big Data
The journey itself is interesting. Each of these horizons average of around six to seven months. Each of the horizons faced three key challenging areas of business, cultural, and technical. And in varying degrees, those challenges allowed us to be able to still deliver and scale out the solution to get us to the point in the last horizon.
Phase 1: Prototype
The challenges that we faced when we were in this prototype phase was getting senior management support to fund the prototype initiative. But being able to show them quickly what data can do and how sexy data can be was quite easy. We needed to bring IT and CDO on the journey also. We were in a fortunate position to have Alteryx already in the organization. But unfortunately, we were using a traditional data warehouse tool. We wanted to ensure that we were providing this comprehensive view of the data. The prototype was essentially to deliver to the marketing analytics team.
With every change in technology, you face cultural challenges as well. Many of you would experience this with your stakeholders, they're used to re-reporting. We need to shift that kind of view and present them data in a more visual manner, so they don't necessarily do the re-reporting. On the technical side, in this phase of our project we spent a lot of time choosing the right tool and data platform and presenting it as technology agnostic.
We commissioned a marketing mix and attribution model to be built out. The foundation, or the construct of this prototype, was to centralize a lot of that data from the model which consisted of a little over a hundred models, linear regression and random forest models as well. And that particular model provided insights back to our marketing function, not just where to spend and where the most efficient channels are to invest in, but when to actually go out and spend based on our rollout of the network over time. They also provided insights around hyperlocal marketing that helped us understand the segmentation in particular areas. With that,we could actually do local area marketing to influence the population to jump on to the NBN network.
And to help us transition into the next horizon, we started thinking about change management. What we did was we branded this solution. We named it Nexus and the definition of Nexus is essentially the connection of multiple points.
Phase 2: Operational
As we moved into the more operational state where we started increasing a lot of our stakeholders, we needed to think about, "How do we create this solution with more robustness and processes?" Not just firming the process or the governance but also the workflows. We worked very closely with the CDO to centralize this platform and create governance around it. We ensured that we identified key data owners and stewards, and the responsibilities and accountabilities around that.
Centralizing the reporting platform was somewhat of a challenge but because we had Alteryx and Tableau, it was easy to serve that up to a wider audience across the organization. The next key challenge was shifting the mentality of re-reporting within Excel, to empowering the user to use self-service capability through Tableau. So the team and I developed self-service dashboards. We also created Alteryx apps for our commercial finance team to extract data in a controlled manner from Nexus. We're essentially creating this window into Nexus for the highly trained analysts to no longer wrangle the data, but to analyze that information that's coming through. And making it easy for them to be able to access that information.
On the technical side, we were modelling to scale to an enterprise level going from weekly refreshes of Excel files to daily refreshers. We worked to get enterprise architects on-board with their vote of confidence to point other teams to Nexus as a “source of truth”. Other teams started using Nexus called for increased importance in Governance and reliability.
We put a process in place for scheduling, reconciliation and error reporting and communication.
Phase 3: Cloud Big Data
We produced the performance reporting as well as operational reporting. Our CEO recognized the value of our solution and we were able to get additional funding to move on to the next phase. We also implemented an AWS stack because IT saw that our solution was mature enough to be able to be migrated from our traditional B.I. warehouse into an AWS. But that then it caused a number of challenges. And those challenged is the increased usage of this Nexus platform. The question was around “How do I actually enable the data scientists as well as utilizing the AWS stack?” We wanted to ensure that we managed the prioritization of all those enhancements of this growing user group as well.
On the technical side, some of the challenges that we see are "How do we continue to utilize Alteryx as a platform, make use of the various new connectors, as well as some of the new exciting features that have been announced at the conference this week, especially assisted modeling?" We see Alteryx as a key platform for bridging the traditional data science teams within NBN and the business analyst community. We're actually implementing a Digital Management Platform (DMP) to collect and blend data with the usage that we currently have. We developed a usage-based segmentation model. Then, we can actually understand the customer experience, predict what the customer experience could potentially be in certain areas of Australia, and be a lot more proactive as to how to actually communicate.
Describe the benefits you have achieved