Energy Australia will use data analytics to understand customer happiness, predict their ability to pay bills and intervene if they show signs of churning to another provider.
The utility last year revealed it has taken the axe to a “spiderweb” of disparate information systems and consolidated to a single data platform built with Oracle technologies.
With Oracle in place, Energy Australia is now in the process of building a central "customer event store" to keep track of billions of customer interactions such as phone calls, emails, live chat and activities logged through Energy Australia's MyAccount self-service portal.
This data store will act as the foundation for a series of analytics projects that aim to improve the customer experience.
The customer event store consists of a "scalable structure and ingestion process to store every single event against an Energy Australia customer", the utility said in a presentation to Oracle's OpenWorld 2017 last week.
The store already houses 250 million events. It is expected to be ready for production use in three-to-six months, information architect Jahanzeb Azim said.
Once it gets to that point, Energy Australia plans to run a number of analytical and correlation models on top to assess how best to communicate with customers and keep them happy.
One of those models combines internal and external data assets to index customers based on their experience when dealing with Energy Australia.
The index draws on the customer event store to analyse everything the customer does, but also takes into account external factors such as interest rates, weather and the average income of the customer's home suburb to assess their happiness at any point in time.
The utility wants to use that information to better tailor communications and engagement with customers, and in particular so it can target unhappy customers before they churn to a competitor.
“If you look at the customer lifecycle it runs from acquisition, experience and churn," Azim said.
"We want to manage the experience so that churn never happen or that people thinking about churning wait as long as possible."
Data analysts could use the index to reward a group of customers for their loyalty. For example, the analytics platform could be used to identify customers with tenures of 80 months or more, who have never made a complaint or asked for a transfer out.
Analysts could also segment customers by looking for those who used to pay on time but now pay late with increasing delays in payment.
If their electricity consumption and bills have remained relatively constant, then it’s likely that some other factor is making it difficult for these loyal consumers to pay.
“So this is a loyal customer who is having personal problems. How can we interject and show that we care?” Azim said.
History showed that customers struggling to pay their bills would eventually request a discount, churn to a cheaper retailer or become a bad debt.
But Energy Australia hoped to calculate an optimal window in which it could reach out to the customer and come up with a way to keep everyone happy.
The analytics team has also looked at the times of day when customer outreach has the greatest chance of success.
By conducting a phone survey and comparing the results to meter data, the utility found customers most receptive to offers to their mobile phone after they woke up and before they went to work. They have since built a tool to predict customers' wake-up times.
Call centre flow
Customer intelligence could also be used to improve the flow of calls through Energy Australia's call centre.
Currently, when a customer calls in, an interactive voice response (IVR) system captures the basic reason for the call, assigns the customer to a queue and the first available agent picks up the call.
But the customer event store could hold intelligence on the reason a customer is calling, providing an opportunity to personalise the engagement upfront.
For example, if a customer regularly calls to check their bill amount, the phone system could automatically switch to a custom greeting like: “Welcome to Energy Australia. Your latest bill is $154.30,” before giving the standard menu options.
Another idea being canvassed is to use intelligence to prioritise customers in the phone queue.
Wait priority could be calculated based on lifetime value, the customer experience index score and the predicted call reason.
The IVR could them, for example, decide to move a high-value customer who joined the queue later to the next agent if they were unhappy.
Sholto Macpherson attended Oracle OpenWorld 2017 in San Francisco as a guest of Oracle.