CMO roundtable: How machines will make or break your customer engagement success
- 19 February, 2018 06:51
There aren’t too many Aussie marketers right now who aren’t trying to connect the experience dots and turn their customer acquisition craftsmanship into retention and advocacy. To get there, however, there’s a growing complexity of digital channels to navigate, increasing swathes and types of data to stitch together, and ever-rising consumer expectations to contend with.
Just how brands are faring in this quest to meet modern customer engagement expectations, and the machine learning and artificial intelligence (AI) will play in helping them reach such expectations, was the subject of a recent CMO roundtable and panel discussion, sponsored by digital marketing and experience management platform vendor, Sitecore.
In an initial panel presentation, Sitecore strategist, Anthony Hook, outlined the current and future state of AI and machine learning application and what marketing teams must have in place in order to harness their potential. This led to a whole-of-table discussion, focused on what’s helping and hindering marketers and CX leaders from customer engagement success
Excerpts of the roundtable and AI panel are featured below. You can find a slideshow from the Melbourne event here.
Building a personalised approach
Personalisation is a key element to customer experience success, and roundtable attendees reflected mixed levels of maturity when it comes to achieving it. Bendigo and Adelaide Bank head of customer voice, Ian Jackman, for example, rated the bank highly in relation to face-to-face interactions, and relatively low - but rapidly evolving – in relation to the personalisation of digitised and outbound communications.
“For us, personalisation implies the ability to present highly customised content to customers that is relevant, valued and ‘in the moment’,” he said.
One way Bendigo Bank is looking to better understand what customers experience when they engage with the bank, and what opportunity exists to improve them, is a new customer metrics framework. This tracks key success measures across the categories of ‘attract’, ‘please’ and ‘grow’.
“Many of the metrics are sourced directly from customer surveys and feedback,” Jackman explained. ”We have also recently implemented the IBM Watson Customer Engagement platform to leverage the customer insights and analytics, and enable us to respond to customer behaviour with relevant content across channels.”
Over at retail property owner, Vicinity Centres, two key areas of focus took precedence in 2017, general manager of data science and insights, Genevieve Elliott, said. The first has been a 12-month project identifying and understanding key drivers of shopper satisfaction nation-wide.
“We have identified seven different shopper segments and are now using these to inform how we develop our shopping centre product,” she said. “The second area of focus has been on actual customer behaviour in our shopping centres, collected via our Wi-Fi network.”
There’s no doubt consumers are in the driving seat, something Vicinity has recognised in its corporate strategy, Elliott said. This puts consumer understanding at the heart of ongoing product development. However, historically, we have had an intermediated relationship with our consumers.
“Vicinity has had to develop a program to gather information about our shoppers in order to inform our new customer-led approach. This has meant significant investment in audience segmentation around shopping drivers, and the beginnings of an NPS [Net Promoter Score] program,” she said. “But the greatest change has come through investment in data gathering technology.”
At Village Cinemas, investments into CRM and customer loyalty programs have been the data collection step driving personalisation improvement.
“To build on the loyalty data set, we on-boarded a customer data platform [CDP], and ingested all our key data sources into one pool, enriching our view of customer. That’s informing all customer communications and tactics,” GM of sales and marketing, Mohit Bhargava, said.
“The challenge is to ensure we maintain the balance between becoming too mechanical and placing an equal emphasis on tone and delivery, both digitally and physically. True customer experience is human and emotionally led in our business.”
The PAS Group head of digital and loyalty, Anna Samkova, said the role of personalised messaging in near real-time has been highly impactful building engagement around brands such as Review.
“We moved from top-down conversion to personalised engagement,” she said, noting the use of the Oracle Marketing Cloud platform, Responsys, to assist. “With the help of LiveClicker software, we then took personalisation to the next level and now include dynamic content in our email communication.”
Tapping data for insight
As a way of elevating data’s role in customer recognition and engagement, Vicinity Centres has been working to create a data philosophy and environment that will allow teams to better leverage existing and new data sets, Elliott said. Vicinity has also built a cloud-based data environment that includes a data lake, team of data engineers and data scientists. “This environment, coupled with robust governance, means we are in a position to surface data and insights that haven’t been available to us historically.”
At Bendigo Bank, a single customer view brings together core customer data and product relationships from a range of source systems, and includes 98 per cent of the group’s customers, Jackman said. This underlying data platform feeds into our engagement systems.
“We are continually extending this customer understanding through external datasets, interaction data, and analytical modelling,” he continued. “Most of the event-driven customer interactions we have automated so far relate to service or experience outcomes, alongside some sales-based interactions.”
Further steps being taken include bringing phone, digital and branch interactions data into the mix.
“This also involves connecting the new automated customer engagement platform into these channels to enable personalisation of these experiences,” Jackman said. “And we are transforming our underlying data architecture, which supports our digital and engagement platforms, to improve the speed, consistency and scalability.”
Jackman made the distinction between a single view of customer and a single view of customer experience.
“The former is an aggregation of what we know about the customer across our various products and brands. The latter extends this into an understanding of how and why the customer is interacting with us and the experiences that we are creating from the customer’s perspective,” he said.
“Achieving the latter is much more complex yet much more valuable when seeking to deliver personalised and relevant experiences. Customer voice is an important input into this deeper level of understanding, as is having technology to aggregate these experiences and an integrated set of capabilities across customer analytics, engagement and marketing teams.”
Staying relevant to customers has also become an overarching priority at The PAS Group, especially as customers expect continuous personal experiences that connect mobile Web interactions with brick-and-mortar ones, Samkova said. “We’ve integrated our IT platforms and created a central repository for our customer data which helps us to easily access and understand our data.”
However, Samkova said the group spends a lot of time focused on the creative and customer need, rather than just asking questions of data. “You have to have humans asking the right questions. The customer is at the centre of what the business does,” she said.
“Sales and profit are the outcome of this centricity. We use analytics to create unique user experiences by predicting the best offer, offer or content for each individual customer based on their profile, transactional and Web data. The next stage for us is to look at how we can get data about attitudes, behaviours and lifestyles.”
As technology further pervades the engagement process, Jackman warned digitisation and automation can lead to a lack of empathy and customer connection. Another challenge industry sectors such as banking have to still overcome is a traditional product-focused versus customer-centred approach in relation to organisational structures, systems and processes.
Today, it’s about working on understanding customer needs and the ‘context’ of the experience, based on all of the interactions they’re having, Jackman said.
A challenge Village is working on overcoming is singularity in data analysis and decisions, Bhargava said.
“Only a few years ago, various business divisions were using different data sets and sources for their analysis, often diluting credibility of information stemming just from CRM data or marketing dashboards,” he commented. “It’s critical the data source being used by marketing departments are not seen as marketing analysis but rather as accountable and acceptable to the whole business.”
As this journey has progressed, marketing has embraced metrics that matter to the business and product owners, Bhargava said.
“That said, we have a company wide NPS measure is place which ensures there is a clear focus on the customer across the board,” he said. “This shift in our language and reporting has helped garner stronger relationships and alignment across the leadership group.”
Elliott is another who believes measuring customer satisfaction in a robust and reliable way is key to driving any valid focus on customer experience. Having just kicked off its own NPS program, Vicinity’s vision is to develop this into an experience index correlated to traditional financial success metrics.
“In addition to this, our investment in Wi-Fi technology is providing metrics we haven’t previously seen, such as dwell time and frequency,” she added. “Once we understand the relationship between these metrics and commercial success measures, we will be able to deploy strategies and tactics that influence these.”
Up next: Getting a grip on AI and machine learning
Getting a grip on AI and machine learning
The exciting news is artificial intelligence and machine learning present new ways of understanding and actioning a customer engagement strategy driven by data and focused on personalisation.
To kick off our roundtable, Sitecore strategist, Anthony Hook, explained the current and future state of AI and ML application and what marketing teams must have in place in order to harness their potential.
AI versus ML: A definition
Hype is definitely where we are at with AI/ML right now in terms of what’s available to marketers, Hook said, and understanding the distinction between both is important. He described ML as taking data and applying algorithms to derive outcomes.
“We may even know the answers we want, we’re just validating them. Or we’re looking for answers or questions and their subsequent answers,” he said.
Where AI can be seen in marketing is in things like intelligent chatbots. “The technology behind it is machine learning, but it feels more AI because the human is having a conversation with a machine,” Hook said.
“The key word there is artificial as opposed to intelligence. That will grow of course as we achieve singularity, which is the point where a machine can have an intelligent conversation, and that technology is accessible to the people we have around the table today. But machine learning is where it’s at for us right now.”
Playing a role in martech, adtech and customer tech
Current applications of ML comes in two flavours. One is in productisation, or where vendors such as Sitecore provide features within their platforms. These include predictive sendtime optimisation for email marketing, and automatically and semantically tagging digital content and images.
The second way is process driven, or what Hook positioned as ‘bring your own algorithm’.
“That’s where we have this sea of data we’re collecting, often in different silos, and it’s about applying data scientists, machine learning and code to those processes, trying to drive that,” he explained. “Productisation is accessible to everyone around the table; it’s usually a case of the more you pay, the better the technology. Process driven – that is way more complex.”
Ultimately, productised implementation of ML helps marketers save time. Hook noted recent research that found the average time each marketer spends pulling data together to do analysis is 3.6 hours per week.
“The reality is the answers to the questions are all up in their head somewhere, and most machine learning in the process driven space is supervised. That’s how I think ML will really help marketers: Free their time up and let them focus on doing other stuff,” he said.
Gleaning customer insight
Where the rubber hits the road in Hook’s opinion in utilising these technologies right now is customer segmentation and audience discovery and in creating “living personas”.
“We’re targeting all our active campaigns to a bunch of personas we believe are true and correct, but ML could well show us there are a huge number of microsegments that could be reprioritised or readjusted,” he said.
To illustrate the point, Hook noted a number of companies using Invision’s prototyping software to tell the story and creatively plan around these living personas. These are then regularly updated based on data and fed back into audience managers and segmentation tools.
“If we can get into levels of hyper segmentation at the upper end of our buying cycles, that’s where becoming more personalised and relevant becomes achievable,” Hook said.
Of course, using machine learning to free up time creates another time-based problem: Getting data into one place where it can be useful. With productised ML solutions, Hook said data is usually plugged in.
When you get into process-driven ML, however, most companies face an age-old barrier: Data silos. “Bringing all that customer experience data into one place does make a difference,” Hook said.
“That might be in a data warehouse, where you have a team happy transcoding and translating data formats into one common format, or it might be a proprietary system where you have connectors in those platforms to pull things together. Once you have the data together, you have the entire customer journey in a big silo you can run data analytics over.
“The success of decisions you make in email automation are 100 per cent impacted by the point-of-sale, retail, offline and ecommerce journey. You need to centralise data. That is both a time and technology thing, but it has to be solved.”
The second must if you’re going to start investing in ML is trust, Hook said. He noted many instance of ML being rolled out have been via incubation hubs within organisations, and are usually spearheaded by representatives from marketing, IT and data analytics.
One example is Valtech, which worked with Rotmans School of Management in Canada to analyse all data using ML in a proof of concept pilot without telling any c-levels they were doing it. Getting early results paved the way for scaling the solution.
“Getting c-level to trust the process you’re going through is right, and not to demand answers in the first month, is key,” Hook said.
One local Sitecore customer investing in ML is RAC WA. Under a proof of concept, the motor insurance provider took customer data from marketing and digital platforms as well as other business datasets and pooled this into Azure ML in order to start delivering personalised exit journeys. As soon as a digital customer reaches a thank you or completion page, they’re presented with another product that is relevant to them. In its first month, exit rates on pages dropped more than 4 per cent.
For Hook, the next part of the journey where the model gets clever will see it continually learning to improve accuracy.
“If RAC WA sees a 1 per cent drop per month over the next 12 months on that exit journey page, that’s a massive success,” Hook said. “And it’s justified internally from a c-level perspective on how much work needs to go into delivering that.”