Following bushfires near Perth in nearly 2021, the Commonwealth Bank of Australia revealed it was analysing data to help to customers affected by natural disasters.
CBA explained it was using algorithms and data from emergency and weather services to offer same-day “tailored support solutions” to affected customers – including those affected by the Perth fires. For example, customers might receive an offer to defer a loan, or an offer of an emergency overdraft. This was about timely, personalised support, the bank stated.
This is one example of the way personalised interaction with large numbers of people is changing. Cloud, artificial intelligence and automation have made it more feasible to store and analyse large data sets about historic, current and predicted customer behaviour and take action quickly.
Some see these technologies enabling personalisation on a scale that is not always feasible with more manual approaches such as focus groups, surveys and marketing campaigns.
The implications for the way banks, retailers, universities, government agencies and other organisations engage with their customers are becoming clearer.
These technologies are making it possible to respond faster to customers. For example, some Australian banks are using apps to give customers ‘nudges’ encouraging them to save money. Utilities are analysing energy usage data for sudden spikes or drops in householders’ energy usage so they can offer appropriate support and tailor consumer tariff plans to be more suitable.
Combining population-scale data analysis with more personal research, such as face-to-face interviews or behavioural analysis, could also enable better appreciation and anticipation of customers’ preferences, according to Mason Davies, Partner at Digital Delta, KPMG Australia’s digital transformation practice.
“Just because we both like a certain hobby doesn't mean we like the same product. However, it is possible to infer certain brand affinity based on sentiment analysis from our social media interactions. Combine this with data on my spending, travel or other interests and you improve your customer hypotheses significantly. It's really about understanding those micro differences in intention and behaviour.
“You and I might live in the same street, have the same interests, spend roughly the same amount of money each month and we might even have the same type of mortgage. However, I might like doing my grocery shopping online, and you might like going in store because you'd like hunting out certain products.
“Or, you might like researching clothes online and then finishing that transaction in store, while I might like paying a bill online, but when I've got a query, preferring to speak to somebody. There's various nuances. Whilst these nuances were previously translated into set customer journeys, now the options are vast and yet more predictable over time,” Davies says.
Organisations are also using data to track whether services or initiatives are resonating. For instance, universities are using data dashboards to try and get a better understanding of what is and isn’t working in classrooms.
Davies calls this process of continually refining and validating understanding of the needs of a large number of people and responding dynamically, “empathy at scale”.
A more detailed customer picture
In the education sector, data-driven efforts to personalise education have been going on for years. Davies has been working in this sector and sees great potential for large-scale personalisation.
“What would happen if we were able to gather the data associated with all of the students – not just in a classroom or a school, but a state or a country – and really understand the types of indicators that show where students are struggling with a particular concept, and the types of teaching approaches that help those students overcome that particular challenge?”
“Every single student in a classroom could have a personalised learning experience, which is based on inputs from the last 20,000 children who had exactly the same problem with the same question. We might find that one student responds very well if taught a concept in a certain way, whereas another student is a different type of person, so there's a different data set that applies to them,” he says.
Davies has been helping educators explore the possibilities. This involves using digital platforms, combining data from a variety of sources and correlating it to support appropriate interventions. The data could be about student engagement and learning styles, and could also include information about student performance from learning platforms, including Google Classroom, Microsoft Teams, Sentral and Canvas, as well data from subject-specific platforms Mathletics, Mathspace and many others.
This data could be combined with information about teaching styles and curriculum material. Social media analysis could also indicate whether students are engaged with the student community, or even who might be having health problems that could be affecting their academic performance. Certain indicators could trigger automated guidance or support, or human intervention.
In other sectors, patterns in the use of services, such as phone, data or water services, might be taken into account. This could help organisations support customers in financial distress, recommend products and services, or cross-promote financial management services, Davies says. He imagines the savings made by switching to a new data plan being paid into a super account, for example.
The approach Davies touts combines this data with ethnographic study, persona development, contextual analysis and other tools and activities that add to an understanding of customers. “Whilst the data source is essential, context is critical,” he notes.
To make use of the data on a large scale, Davies sees organisations needing well-trained machine learning models and automation systems, refined using real world experience.
He sees experimentation as important: “You try experiments and you then need to listen to whether they were effective. Who responded positively, who was neutral who had a negative response? How do how do each of those responses affect your understanding of empathy? It's a continual process of understanding, trial and then refinement.”
Earning trust
Perhaps the most important requirement for any organisation undertaking these activities is to act responsibly.
The ethical and privacy risks are obvious – just look at Cambridge Analytica’s infamous use of data about Facebook users for the purposes of political advertising. Davies admits this was “a very good example of personalisation at scale” used “for the wrong purpose”.
Bad data or poor training of machine learning systems can also cause problems for or annoy customers.
Poor communication can also cause distrust. As anyone who’s received a personal product recommendation online for no apparent reason knows, personalisation can be creepy.
It can also come across as promotion, rather than authentic interaction, argued Aaron Spinley, vice president ANZ of customer journey analytics and orchestration company Thunderhead, in a Mumbrella article. Personalisation had an “irreversible PR problem”, he wrote.
Some organisations are tackling these issues as they seek to improve interactions with customers.
Davies point to one client which has a data and ethics committee. “It's looking at what can be captured, how to capture it and what is legally correct. It’s involving prominent advocates on safety, AI and ethics from academia and the business world to advise on this topic. That’s going to inform an ethical design authority for the whole business,” Davies says.
Organisations involved in these initiatives will also need to earn and maintain customers’ trust.
Davies argues trust could be earned if customers’ think they’re benefitting from a value exchange. “I think there's a lot more that needs to be done on what that value exchange looks like – you give us x and we'll give you y – and make it a lot clearer,” he says.
He also recognises a need for customer choice: “We need to make it very simple for the customer. If they want recommendations, then that's great. They should say, ‘Yes, I'd like those’. If they don't want to provide information, they can turn that off.”
These challenges, and the opportunities for organisations to personalise digital interactions, will continue to evolve with technology and data generation.
“An interesting journey lies ahead for businesses, government and the people that use their services,” Davies says.
