Connecting millions of Australians to the national broadband network: Forecasting service demand for the NBN
In 2015, the Australian Government announced the rollout of a national broadband network (NBN). With millions of homes and businesses awaiting connection, the challenges were clear:
· How could NBN Co accurately forecast connection orders in different regions across Australia?
· How could they optimise workforce and resource allocations while maximising customer satisfaction?
· And how could these findings help inform NBN products and services in order to meet market demand?
The problem
Rolling out the NBN was a major undertaking. For NBN Co, the builder and service provider of Australia’s national broadband network, the challenge lay in effective workforce planning to support the delivery of top-tier customer service.
NBN Co originally relied on records of previous rollout activity to forecast demand and allocate workforce resources. This wasn’t always accurate, and as the project picked up speed, the company needed access to accurate data to determine data-driven strategies to manage the rollout process more effectively and improve customer satisfaction with their services.
The opportunity
Researchers at the UTS Data Science Institute, in partnership with NBN Co and Data61, Australia’s largest information communications technology research organisation, designed a series of machine learning models that could predict future orders and network utility.
These models enabled more accurate allocation of NBN Co’s technical workforce and more precisely timed rollout schedules to reduce customer waiting times. Publicly available Australian Bureau of Statistics (ABS) Census data was also used for insights into the demographic profile of different regions, further enabling more accurate order forecasts.
By using clustering and data mining algorithms, researchers were better able to identify unhappy customers, resulting in better understanding of customer behaviour. The result was a lift in efficiency and an overall increase in customer satisfaction.
The impact
As of May 2020, 96 per cent of homes and businesses were connected within agreed timeframes, up from 87 per cent in January of the previous year. Customer satisfaction with the connection experience rose by about 12 per cent, with 2.3 million new customers connecting to a plan since the new, model-based planning approach was introduced.
For the NBN Co. analysts tasked with supporting the rollout, there were also huge benefits. Compared to NBN Co’s previous planning approach, the new predictive models reduced their workload by 62.5 per cent and improved prediction accuracy by 30 per cent.
Find out more about how we work with companies to improve their data analytics capabilities, or contact Plus UTS.
