In Part 2 here, we will discuss the design and implementation of AI customer experience (AI CX), as well as our prediction of the future of AI in customer experience support services. This Part 2 is a continuation from Part 1 which covered assessment for adopting human-centered CX AI, and related issues of data, foundational capabilities, use cases, generic or domain-specific AI models, etc.
As AI is advancing by leaps and bounds across all industries, seamlessness and personalization are becoming the two key competitive “moats”. (“Moat” in AI is like a moat around a castle protecting a business from competitors with a unique AI competitive advantage. ) Customer-facing companies such as JP Morgan Chase, Home Depot, Starbucks, and Nike have announced their core AI strategy as “personalized and seamless omni channel experiences”. “We are now at the point where competitive advantage will derive from the ability to capture, analyze, and utilize personalized customer data at scale and from the use of AI to understand, shape, customize, and optimize the customer journey”, per Harvard Business Review.
Towards that end, this article will guide contact centers and CX managers how to approach design and implementation of AI CX.
Elevate ESAT and CSAT in tandem. We all know that happier and more engaged employees create happier and more satisfied customers. The purpose of AI, as a tool, is to free up humans to do the higher level work of delighting customers. By improving both the quality of work and its environment, happy and motivated frontline agents elevate customer satisfaction, acquisition, and retention.
In addition to using AI CX to cut operational costs, visionary and human centric companies need to consider designing their AI CX tools to elevate employee satisfaction (ESAT) and customer satisfaction (CSAT), in lockstep with each other.
Internally, design AI CX tools to improve workforce management, help onboard CX teams, reduce operational costs, increase productivity, and enhance workforce skills.
Externally, AI CX can be adopted “off the shelf” or custom-designed to analyze customer data, predict and prevent customer churn, acquire more customers, and create more consistent brand experiences.
Identify data assets and sources. After an initial assessment of what exact problems to solve or what customer-driven opportunities to achieve, and what specific CX outcomes to accomplish, you need to “identify what data assets and sources these initiatives and opportunities rely upon”, per an AWS whitepaper. If the data source lacks cybersecurity and privacy protections, or comes from a questionable or an illegal source, you must solve these foundational issues before proceeding any further.
During data extraction, having humans in the loop is important to prevent breach of data privacy and security.
Identify cross-organizational dependencies and disruption to the organization. “AI adoption is a cross-functional effort, much more so than is the case for other technologies.” Supra. Even though initially your AI adoption is limited to prototypes or pilot projects, once it is adopted, even a limited-scope project can disrupt more than one part of a technology infrastructure and digital processes. It may have a domino effect, impacting many other interrelated processes and the entire infrastructure. It is critical to first align every part of your organization along the goals and vision to ensure future AI readiness by the entire organization. Adopt locally but envision holistically. The success of your initial AI CX project can secure stakeholder buy-in for necessary upscaling and updating technology infrastructure, and even change management if needed down the road.
AI CX Experiential Design. AI algorithms must follow a real-world customer journey and customer experience that delights end users. However, at the time of this writing, much of the AI priority is weighted towards cost-cutting more than customer satisfaction, driven by venture capital investors’ desires for speedy and exponential return on their investment. We believe that focusing on short term and temporary savings is a myopic approach to AI CX experiential design, as illustrated in this article on AIXD. It will hinder future AI CX expansion beyond merely simplifying tasks and scheduling, deflecting tickets, and providing analytics.
At the onset, take a long term view for developing customer-centric CX AI capabilities, which will lay the groundwork for future scaling of AI data, applications, models, channels, and knowledge base, to substantially improve CX and the bottom line in both the short and long term.
Setting metrics around the problems to be solved. AI CX design needs multiple iterations. During the design stage, it is wise to build your own benchmarks and analytics around the problems to be addressed by AI and the underlying data, to facilitate trouble-shooting and evaluations with metrics such as latency, accuracy, tone, goals, etc.
Frameworks for evaluating AI models against the metrics and benchmarks. To ascertain that the deployed AI models have the desired results such as reducing latency or delays, and monitoring new customers, etc., you need to differentiate the metrics and frameworks from one use case to another. Frameworks for evaluation also need to be changed when selecting better fitting models, switching from open AI to bespoke models, from LLM to SLM, etc.
From prototyping to testing pilot projects to solve specific CX problems, to incrementally scaling AI CX, models and data are refined with every iteration. Mastering each iteration depends on a healthy feedback loop, competency of AI talents, strategic planning, visionary leadership from the top, coordinated execution by teams, the quality of data, right-fitting AI models, and other factors.
Launch pilot initiatives. “In this phase, you focus on delivering pilot initiatives from early proofs of concept to production and demonstrate incremental business value. Pilots should be highly impactful on the organization and the business, as well as meaningfully benefit from AI being applied to it. Regardless of whether they are successful or not, they can help influence your future direction. Learning from them helps you adjust your approach before scaling to full production”, per the AWS paper.
Feedback from frontline workers and customers. The frontline workers interact with customers daily and know better than anyone in the organization about customers’ sentiment, pain points, and friction points along the customer journey. Their feedback about AI CX is highly valuable. Organizations need to have a system and process to collect feedback from employees and customers at all times and in all stages of the CX journey.
Interaction-centric, cloud-native platforms can enable feedback in real time from employees and customers. Each interaction becomes an opportunity to create higher level experiences for employees, customers, and organizations, and to deploy and scale AI CX with speed, security, and CSAT.
Refine AI models to achieve higher CX goals. Use customer service metrics data to monitor and measure every step of the customer experience. By integrating AI with human centered workplace management and customer services, CX AI can become the catalyst for positive changes.
Scale: “This phase focuses on scaling pilots in production to achieve broad, sustained value. Scaling here can mean not only the technical capabilities of solutions or initiatives, but also the reach of them through the business and towards your customers. This activity translates your activities into customer value”, said the same AWS paper, which also cautioned that:
“While you iterate through these cycles, recognize the limits of what you can achieve in a single cycle. It is important to be ambitious and aim high, but trying to do everything in the same cycle can lead to discouragement in the organization. This is why pairing a larger picture with many pragmatic and actionable steps and measurable KPIs on these smaller steps is crucial. Every step then brings the organization closer to its goal. Do not try to do everything at once. Rather, evolve the foundational capabilities and improve your AI readiness as you progress through your AI transformation journey.”
As AI gets faster and smarter, it is anticipated to be more integrated with IoT (internet of things), AR (augmented reality), VR (virtual reality). Eventually AI will evolve towards fully autonomous, as co-pilot first, then even replace humans.
As product and support lines get blurred, the competitive differentiation is user experience or customer experience.
Starting from IVR, chatbot, voice agent, to LLM, gen AI tools, AI is predicted to evolve into AI agents who might just do all the human work as “Contact Center AI Agents” for insurance companies, banks, and many customer-facing services, even in the healthcare industry. Humans will just monitor and check the AI agent’s work. AI will be flying a plane, with a human next to it acting as a co-pilot. But that will take a LONG time. There is still a lot to be done before multi-model AI agents become fully functional.
©Jerry Briggs all rights reserved.
We at ApexCX will be always here for you, today and tomorrow, to increase your operational efficiency, optimize consumer experience, and enhance your brand value in the AI CX journey fo a better future.
Contact us for more information. Thank you.
