Changing the way Australian businesses access capital
Faster time-to-market for new solutions
Credit approvals for customers
Shift is tackling the traditionally burdensome business finance experience to make it simpler, more convenient, and more empowering. This meant changing the way businesses accessed capital so that the process was focused on their needs — not on products. To get there, Shift needed a consolidated view of bank transactional data from disparate sources that they could quickly gather to better understand their customers’ situations. With the help of the Azure Databricks Data Intelligence Platform, Shift built back-end data pools overlaid with proprietary in-house classification models with machine learning capabilities, resulting in a level of automation that reduced processing times and that consistently delivers efficiencies, flexibility and cost savings to customers.
Hurdles to accessing and processing data at scale
For a fast-growing, technology-centric business like Shift — which is built on speed — processing the ever-increasing volumes of customer records efficiently was becoming a challenge. Furthermore, the data had to be processed and presented to users across the company in an interactive, easy-to-understand format.
“Transferring and processing the different data sources was slow, making it difficult to get a consolidated picture of what the customer insights actually were,” said Shift data engineer Thomas D’Arcy.
Structuring and running ETL pipelines was another challenge the data team had to overcome. They usually had to stay past business hours to run a pipeline and were wary of experimenting on new data models due to the huge effort required.
“We deeply consider every single element of the digital user experience, and this allows us to challenge the traditional ways of business lending,” said James Spence, Shift’s product strategy manager. “Today, data intelligence, automation and machine learning let us make decisions accurately and quickly to enhance the customer experience and make the complex simple.”
Shift turned to the Databricks Data Intelligence Platform to centralize their disparate data sources within a unified, scalable infrastructure to improve access and the efficiency of their data team.
Improved data reliability to generate a dynamic customer view
Databricks has provided Shift with the means to unlock and structure all their data to uncover insights that enable them to have more meaningful conversations with customers and personalize the end-to-end experience. With Delta Lake — an open format storage layer that delivers reliability, security and performance on a data lake — Shift is able to easily migrate all their machine learning data into Delta Lake files. This empowers data engineers like Tom to manage the full data lifecycle more reliably, allowing data scientists to concentrate on experimenting with models to enhance the customer experience.
“Through using Databricks we’re able to process a significantly larger volume of data faster, which in turn allows us to focus on stitching together and understanding both the systems and the customer better,” said James.
Data analysts can also leverage the information stored in Delta Lake to generate a 360-degree dynamic view of the customer via Tableau dashboards, providing personalized assessments and optimizing recommendations to increase customer interactions.
“Our data teams are not afraid to test things anymore. Quick and efficient testing means we can be more innovative, agile and confident in using the power of data analytics to help solve our problems,” said James. “Such tools are essential for a compact team like ours, to fast-track product offerings that allow us to enhance the customer experience.”
Cross-team collaboration is also facilitated between the teams in Australia and Bangalore with easy access to all their data and compute clusters.
Reducing time-to-market for new data models
With Databricks, Shift has aggressively reduced their time-to-market and time-to-test, and they’ve implemented real-time decisioning for certain segments of their customers — an unprecedented feat in this line of business.
Quicker iterations and retraining of models within two days have allowed the organization to better reallocate resources and aggressively cut down time-to-market. With a 90% faster data processing capability, the team has grown predictive capabilities dramatically.
“We’ve managed to uplift the capabilities of our data teams and we’re much more efficient internally. This speed leads to better products, more features and more creative solutions for our customers,” said James.
To further expand their service offering and networks, Shift is now looking to implement unified credit and risk scores enabled by MLflow. This is set to improve the entire machine learning lifecycle, boost the distilling of mainstream information and enhance experiment tracking.
As Shift continues to grow as a go-to source for business finance and embedded solutions at the core of everyday business transactions, they rely on Databricks to enable massive-scale data engineering, collaborative data science, full-lifecycle machine learning and business analytics.
GetCapital leverages customer data to generate Databricks-powered dashboards via Tableau to provide a 360-degree view of their customers. As a result, they are able to deliver personalized recommendations to increase customer engagement.