Professional services firms are seeing strong demand for predictive analytics projects, but say Australian businesses still face challenges implementing or operationalising the insights gained from the statistical models.

While predictive analytics is not a new capability partners from KPMG and Accenture say businesses still struggle when it comes to optimising business processes based on data.

“Predictive analytics is not a new capability but the level of demand from clients is constant,” says Raphael James, Partner, KPMG Digital Delta, a newly formed business unit to assist organisations with their transformation journeys.

“The reality is that if you were to sit with the leadership of any organisation and you were to ask them to put on the table the list of questions that they want to answer, I think a significant portion of those would require predictive analytics to answer.”

Predictive analytics helps businesses answers questions like, which of my customers is about to churn? Or from a supply chain perspective, how much product do I need to hold? Or, what will the demand for product be in the upcoming period of time?

The big challenge: Insight to action

According to James, the challenge is not the ability to build a predictive model but how business processes or humans “consume the insight when making decisions.”

For example if a customer is rated highly likely to churn in the next week, what does the organisation do to intervene in the appropriate time frame?

“The ability for the organisation to consume that knowledge and that insight within the time frame, and then act upon it is where the biggest challenges are,” James said.

“Organisations in Australia are still growing in their maturity of leveraging advanced technologies in customer service and marketing functions and integrating the end-to-end flow of insight into taking action.”

James said large organisations across Australia are making significant investments in technology which will give them greater data analytics capabilities but need to make sure they are adapting process to get the return on their investment.

“The question that we pose to our clients is whether they are truly able to derive the maximum value from that investment. What that comes down to is: how sophisticated and how automated is that flow from generation of a predictive insight through to taking action that results in achieving the objective of the organisation?”

For example, KPMG is currently working with an Australian retailer to develop a “lights out” marketing platform, which is able to recognise consumer triggers and automatically determine the next best action. It’s a process that’s too big for humans to handle.

“If you’re scaling across your entire customer base, you can’t manage this process end-to-end by humans. That’s why you need decision engines and decisioning platforms that are constantly observing, learning and interacting with your customers and managing the experience.”

The last mile

Amit Bansal, Accenture’s Managing Director, Applied Intelligence Leader, told Which-50 Australian organisations are well established when deploying descriptive analytics (visual reporting or looking at what happened) and predictive analytics, but translating it into action is often neglected.

“Everyone builds a model but people forget the last mile,” Bansal told Which-50.

“So what? You can build hundreds of models but if you don’t take action it’s pointless doing it.”

Bansal says businesses need to understand how the output of the model translates to real actions. “How do you embed those actions into your business process or the behaviour of the people or the experience of your end customer has?”

So for example, if the model is telling you that a large population of your client base prefers to be contacted through the app, not phone calls or emails, etc. then businesses need to change the experience  delivery for those customers, Bansal said.

Technology advances will go some way to completing the “last mile” of predictive analytics by automatically alter the customer experience.

Bansal noted machine learning and predictive analytics is beginning to be used to drive hyper personalisation of products for individual consumers. For example, the home screen of a banking app could be automatically customised based on user behaviour by featuring commonly used menus and hiding ones which are accessed less frequently.

In that scenario the machine is able to understand online behaviour, analyse it and provide a different experience.

Previous post

Three Simple Steps for Implementing Programmatic Brand Safety

Next post

Unsupervised Artificial Intelligence is not ready for direct customer engagement, says Pegasystems AI expert

Join the digital transformation discussion and sign up for the Which-50 Irregular Insights newsletter.