How Analytics Leaders are Enhancing Customer Experiences
Executives at the inaugural Business of Data Festival revealed how they’re embedding cultures of experimentation, establishing efficient analytics development processes and setting their sights on hyper personalization
Using advanced analytics and AI to enhance the customer experience (CX) was one of the hottest topics at this year’s Business of Data Festival.
Analytics-focused executives are looking to differentiate their companies from competitors via superior CX. In doing so, they hope to improve the lifetime value of relationships with their customers by meeting their needs more intelligently.
“You must be creating value with a focus on the customer,” said Jose A Murillo, Chief Analytics Officer at Mexican bank Banorte. “If it makes sense to the customer, then you will be able to extend and deepen the business relationship that you have with [them].”
Ren Zhang, Chief Data Scientist at Bank of Montreal, added that improving customer experiences with AI has been a big focus for her since she joined the bank, alongside more conventional data science applications such as fighting fraud and optimizing prices.
“We are also working with digital,” she said. “That’s another big area, where people are looking for personalized experiences. So, in those areas, we started using deep learning to predict what products [customers] are going to like and make a seamless customer experience.”
Like many executives at this year’s festival, Zhang spent the past year delivering projects for the units within her business that are ready to use AI and measuring the impact of these initiatives. Now, executives like herself are setting their sites on larger AI-led business transformations.
Why ITV Embraced Analytics Experimentation
Improving customer experiences with analytics and AI requires enterprises to embrace a culture of experimentation. Analytics teams need processes that encourage them to generate ideas about how CX might be improved based on data and then test those ideas in the real-world.
UK TV channel ITV is one example of a company that has embraced a culture of experimentation to improve its digital CX, as ITV Head of Viewer Analytics Matthew Bryan revealed on the third day of the festival.
“[Experimentation analytics] is a really key part of our data strategy and our product strategy,” he said. “It’s really about, ‘How do we drive a culture of continuous improvement through micro-optimizations?’”
“What we’ve put in place is a rigorous workstream with mass testing, and we’ve integrated that into the product and development teams”
Matthew Bryan, Head of Viewer Analytics, ITV
“Experimentation at ITV is something that, over the next few years, will sweep through the entire business,” Bryan continued. “But we started that journey focusing on our digital products.”
Through applying a hypothesis-driven experimental approach, Bryan’s team has significantly improved the user experience (UX) its online steaming hub provides to viewers in recent months.
Similarly, his team observed that that customers were getting distracted by other links on the page during the signup process for ITV’s ad-free service. Starting with this insight, his team was able to significantly boost customer conversation rates following four weeks of testing and experimentation.
“What we’ve put in place is a rigorous workstream with mass testing, and we’ve integrated that into the product and development teams,” he explained. “We then go through a realty vigorous process of A/B or multivariant testing.”
Enterprises are Productionizing AI More Efficiently
As enterprises seek to scale up their AI programs to drive company-wide business transformations, they’re finding they must improve the efficiency of their end-to-end model development processes.
As Dino Bernicchi, Head of Data Science at home shopping retailer HomeChoice International, said during his festival session, inefficient development and deployment processes are holding many AI teams back.
“Where a lot of companies are struggling at the moment is, their AI strategies are a little bit disjointed,” he said. “They know they need to do stuff with AI. They can see the value. But they’re sort of running isolated little projects. So, they’ll try something [and] it won’t really be productionized.”
To address this challenge, HomeChoice International drew inspiration from Uber’s machine learning platform, Michelangelo, to design a machine learning platform that ensures model features can be stored in a library for easy access and reuse.
The company also automated many of its AI model deployment pipelines, so insights can be generated with minimal human intervention.
“At HomeChoice, a lot of it is batch prediction stuff,” he explained. “So, what’s the probability someone’s going to take up a loan? What’s the probability they’re going to want a specific product?”
He continued: “So, for a lot of our deployment pipelines, we could automate and just say, ‘OK, the models will run with those specific features, it will produce results and that will always get pumped into certain systems, and then those execution teams could use that.’”
Taking steps like these to ensure AI teams can drive value more efficiency will accelerate the pace of analytics and AI-led innovation within an organization and ensure projects can deliver returns more quickly.
How Executives are Targeting Hyper Personalization
Having an embedded culture of experimentation and efficient processes for the end-to-end AI lifecycle may prove to be essential prerequisites for those wishing to achieve the ‘Holy Grail’ of many customer analytics strategies: Hyper personalization.
Eddie Short, Chief Data and Analytics Officer at telecoms giant O2, used his festival session to talk about the role hyper personalization plays in O2’s analytics strategy. He said the commodification of the sector in recent years has made it all the more important to provide superior CX.
“What people expect now is what we would call hyper personalization,” he said. “Which merely means they want to have the right experience, at the right time, in the right context, wherever they are.”
“We want to be able to ensure that, at any one time, we offer [customers] the right message, the right action, the right experience”
Eddie Short, Chief Data and Analytics Officer, O2
Short said O2 collects data on roughly 2,500 attributes from consenting customers to build up a detailed profile of each customer’s situation, wants and needs. This data is then used to divide O2’s customer base into ‘behavioral tribes’ which each receive personalized treatment.
“We’re in the business of trying to offer that truly personalized experience on an almost one-to-one basis,” he said. “The only way you can achieve that is with really good data and sophisticated machine learning analytics. A human being cannot make that in-context decision at the right time, in the right place, to deliver to another human being the service they want.”
Of course, effective hyper personalization initiatives require a company to gather vast quantities of data about their customers’ behaviors and preferences. O2’s access to detailed location data from its customers’ phones gives it a real advantage in this area.
Moving forward, companies that wish to achieve differentiation through superior CX will need to streamline their processes for AI innovation. But they will also have to scale up the volume, variety and velocity of customer data they collect to fuel advanced personalization algorithms.
To catch up on all the customer analytics insights shared by top data and analytics executives at the 2021 Business of Data Festival, browse all the festival sessions on-demand here.