Customer experience has emerged as the current darling of tech innovation. Born out of mounting ability to cheaply store, better integrate and intelligently analyze massive new information resources, customer experience appears as a distinct item in analyst forecasting, as the preferred marketing approach – the “future of marketing,” according to Forbes, and, in one consumer study, as more important than product or price in differentiating a brand. Defined in an HBR article as the combination of digital experiences and interactions that may occur on a website or a smartphone, retail or customer service, and the speed at which a problem is resolved in a call center, customer experience (CX) encompasses an individual’s entire engagement with the provider throughout the relationship. Successful CX strategies depend on the provider’s ability to read and segment customers in order to fine tune ongoing journey messaging, and on quick response to service requests or negative social media – at root, on the ability to merge and parse massive amounts of data about the individual, and on access to computing resources with the power and flexibility needed to support push/pull communications as well as the demands of sometimes erratic ecommerce transactions.
Arriving first to the online retail space, CX is now deployed across a range of industries, and is especially vibrant today in the banking sector. TD Bank, for example, has established a Design Centre of Excellence devoted to the creation of digital services that are customer-centric in nature. Imran Khan, TD’s VP of Digital Customer Experience described his team’s and the CoE’s mandate as “data driven design” in which “personalization is key, and data is key to building it” – ‘it’ being defined as the development of a holistic customer profile through the use of advanced analytics. AI is key to how TD now serves digital customers, as it can provide the insights needed to ensure that “customer experience is solid,” he explained. But Khan also argued the importance of “getting the right insights quickly” through a platform that can handle the bank’s data volume, while delivering the speed of response that customers now expect. The customer landscape is evolving, changing more quickly than ever before, he added, and the bank cannot wait to build solutions – it needs to react within weeks. According to Khan, cloud plays a critical role in servicing the needs of the increasingly important digital customer segment: TD currently has 12 million digital customers worldwide, and processes 100 million records online per day. To access the infrastructure needed to address the speed and volume requirements of CX innovation in applications, such as the bank’s location-based TD for Me, TD turned to Microsoft’s Azure, customizing the cloud platform to deliver what Khan called “speed to market but also best to market.”
The vendor community has been quick to recognize the synergies between cloud and AI, and how this plays out in the realm of CX. Running AI, machine learning and deep learning applications on the Big Data sets that form the foundation of insights requires heavy use of computing resources. For many organizations, building and powering the necessary infrastructure locally is cost prohibitive, and utilization of the massive cloud infrastructure available online a better option, even for traditional enterprises like TD that ultimately operate based on a hybrid IT model. In addition, the implementation of complex AI, such as neural networks in deep learning, relies on specialized GPU or other chipsets and on computationally intense architectures that can support rapid processing – for training in machine learning, for example, or for the multiple, often parallel, calculations that drive algorithms statistical modelling. To access these kinds of platforms, users would need to make different infrastructure investments, or deploy in public cloud environments. Going forward, the success of advanced AI deployments in areas such as CX will depend on the organization’s ability to connect user-facing services and interactions with back-end enterprise processes and supply chain to deliver an optimal experience, a significant task made easier by spinning up infrastructure when necessary.
To address AI-specific needs, many of the hyperscale service providers have built solutions aimed at capturing the burgeoning interest in advanced AI – and the new, and rapidly growing use case for consumption of cloud infrastructure. Among others, Microsoft has launched an AI platform that offers a range of AI tools, data storage, and services (including cognitive, machine learning and bot services) on its Azure AI infrastructure, designed to democratize AI – to enable users who are unable/unwilling to make the necessary infrastructure investment to benefit from AI capabilities. And to encourage broader use of these advanced platforms, vendors are investing in the creation of various innovation hubs and accelerator programs. Microsoft, for example, has approximately 8,000 researchers in AI globally and acquired Waterloo-based deep learning research firm Maluuba, and as new president of Microsoft Canada Kevin Peesker explained, provides support in kind to startup businesses across the country. The BizSpark program, for example, offers free software, service, support, and Microsoft Azure cloud services to thousands of young Canadian startups (less than 5 years old, with less than $1 million in annual revenues), and has partnered with incubators including MaRS, Communitech, DMZ at Ryerson, Founder Fuel, and L-SPARK, offering access to the Microsoft “brain trust,” training on how to leverage the Azure platform and intelligent cloud, free Azure credits to provide early stage businesses with the cloud tools they need to scale on hyperscale infrastructure.
According to Microsoft, the ability to harvest data insights will be a key competitive advantage going forward. Citing an internal study of the performance of its own born-in-the-cloud businesses, Toni Townes Whitley, corporate vice president, Worldwide Public Sector and Industry, Microsoft, argued that “digital feedback” is the attribute that is making a difference, helping companies that can access and act on digital assets to grow disproportionately as compared to others who cannot. And citing the experience of one online retail customer, Whitley noted that ASOS, which has transactional volumes that outpace Amazon, has seen revenue growth of 25 percent since moving its ecommerce business to the Microsoft platform, and since introduction of “digital feedback” to hone customer experience through better segmenting, targeting and curation of customer engagements. To help support best practice sharing – or the application of digital feedback lessons learned in sectors like retail to other industries, Microsoft has introduced a new program called “Customer Success Manager.” In banking, Whitley explained, “fintech used to be the disrupter,” but going forward industry disruption will occur through a combination of fintech and enterprise digital “cross innovation.”
This redoubled focus on building services around customer data begs some important questions on the collection, use and distribution of personal information. Whitley believes that demographic shifts will resolve many issues for users of CX and other advanced analytics applications: referring to a recent Accenture global report, she noted that 65 percent of customers are now willing to provide personal data in order to benefit from new service levels in the financial sector, a finding she feels speaks to a generational shift in thinking about privacy and security. In any case, she argued, Microsoft policy on privacy is to adhere to what is understands as the outer limits on privacy legislation, and has introduced measures to ensure the company is GDU compliant and to push EU data residency requirements through its supply chain. However, new AI technologies have introduced other problems outside the surveillance of personal information by foreign governments and others, which extend beyond data storage requirements, namely, the introduction of bias in mathematical constructs that drive algorithmic analysis. Today, proprietary algorithms are used in hiring, in criminal predictions and in financial decisions: if bias in a system is unchecked – or grows as the machine “learns” – there could be serious negative impacts for individuals, for minorities and for poorer communities, and ultimately, the progress of a technology that has huge useful potential stalled. Challenges with bias are not insurmountable, assuming the user is interested in investing the time needed to create bias-free analysis, and can be addressed through greater transparency and awareness of the problem.
For its part, Microsoft has representation in a new project called the AI Now initiative, co-founded by principal researcher at Microsoft Research Kate Crawford to educate users on AI biases, and is a supporter of the “Algorithmic justice League,” a collective dedicated to mobilizing all stakeholders, including vendors, developers, citizens, and regulators, in the fight against bias. Internally, Microsoft CEO Statya Nadella has established an “ethics committee” designed to tackle this kind of issue, which Whitley said is currently focused on the misuse of facial recognition technology in web, AR and other technology areas. Going forward, the company may have more to do to keep pace with legislation aimed at making algorithms accountable: the European Union has recently adopted a due process requirement for data-driven decisions based “solely on automated processing” that “significantly affects” individuals, scheduled to go into effect in May 2018. The new rules will give EU citizens the right to obtain an explanation and to challenge automated decisions.
Commenting on challenges in the use of personal information, TD’s Imran Khan noted that this issue “will become more complex.” While customer research tells the bank that individuals want the bank “to know them,” they want to understand how this happens to ensure that personal information is not shared, but rather used only to service the customer. Khan senses that “customers may become uncomfortable about the use of their data,” and so the bank intends to be very careful going forward, bolstering the formal work of its structured compliance organization with regular reviews of new service initiatives, working only from a “statistically significant viewpoint” on what is acceptable use of personal information.
In Kevin Peesker’s view, the TD experience is but one example of digital transformation that is accelerating across Canada. While he has observed the top CEOs from across the country searching for better understanding on how they need to prepare their businesses for transformation, Peesker has also had recent conversations at the federal government level with Shared Services Canada and the Canadian Research Council on upgrading infrastructure, the use of cloud, the creation of “digital hubs,” and investment in “CoEs for Digital Transformation.” “There’s a lot talk about AI, about investment in deep learning in language and computation,” he explained. At the same time, academic institutions are creating the IP needed to drive the digital transformation agenda, often with the support of vendors such as Microsoft which has helped with curriculum development at the universities of Waterloo, Toronto and UBC, and has made direct investment of $7.5 million in AI research at Université de Montréal and McGill University, as are pockets of innovation in areas across the country such as the Toronto-Waterloo corridor, the Incubation hub in Saskatoon, or the Innovation Centre launched recently by Prime Minister Trudeau to facilitate cross border collaboration on digital progress in Vancouver/Seattle. Marriage of this enthusiasm for digital shift with the transparency needed for ethical transformation will be welcome. Who’s watching?
