Conversational AI has officially entered the chat…and voice…and video…and email. While it’s tempting for business leaders and IT decision makers to be enamored by the 24/7 reliability, instant expertise, and unflappable demeanor of these tools, your investments in this technology should be heavily influenced by the customer journey.
A misconception persists that the modern customer journey remains both straight and siloed – a marketing campaign drives customers to a website where they purchase a product or service, and contact is only made if they experience a problem. The reality today is this journey is neither, and AI is only accelerating this disruption, where a conversation can and should occur anytime across any channel – from website and mobile app chatbots to intelligent voice or text-based agents.
It’s little surprise that last year, more than one in five startups (22%) in Y Combinator’s fall batch were building with voice AI (a much-hyped component of conversational AI), a 70% surge compared with the winter 2023-2024 cohort. In fact, Gartner predicts that 85% of customer service leaders will explore or pilot conversational AI this year and by 2029, agentic AI will resolve 80% of customer service issues without human intervention.
As the temperature around conversational AI approaches fever pitch, it’s paramount that before companies start automating calls or texts they first have a firm grasp of their customer journey, the dominant channels where their customers are now, and the robust ecosystem of their customer data that acts as the glue for more contextual, personalized experiences – before, during, and after conversations.
Inbound Versus Outbound, High Volume Versus High Value
When I say ‘Triple D’ most of us probably think of Guy Fieri’s Rojo Red ‘68 Camaro rolling into Flavortown, but it also represents a process all companies must master to get a 360-degree view of their customer journey before conversational AI is even a twinkle in the eye of their CIOs or CTOs: document(a comprehensive record of it), debate(a healthy challenge to it), and data(an unequivocal verification of it).
Together, these three components should inform a blueprint for customer engagement, what conversations – inbound or outbound – and what channels – voice, chat, text, email – funnel most leads. Remember, what you may want the customer journey to look like is irrelevant. It’s a fool’s errand to try and change customers’ communication behavior – there’s a long history of the juice just not being worth the squeeze. Focus on how and where your customers communicate and map a tiered framework of conversational touchpoints from high volume/low complexity to low volume/high value where automation can liberate bandwidth for more complex and potentially perilous use cases that demand human tact and touch.
For companies largely driven by outbound engagement, automation brings heightened scrutiny on regulation and permission requirements, such as HIPAA compliance or notifying customers they’re speaking with an AI agent. Outbound use cases ripe for automation include appointment reminders/reschedules in healthcare and automotive, billing and payment reminders in utilities and banking, and delivery status notification in retail and grocery. More dynamic, sensitive outreach such as delinquent payment recovery in banking or real estate or churn win-back calls in telecom or fitness/wellness are on the AI horizon but still require human-led, high-touch strategies.
If engagement is largely through inbound, automation will fall flat or even frustrate unless it’s underpinned by months or years-worth of secure, well-governed call and chat data to properly train agents on the most common customer queries. Routine inbound use cases include password/login resets, store hours/directions, and loan balance/credit limit checks. Again, issues that require strategic or even legal intervention such as regulatory questions in healthcare or security breaches in finance will still require humans in the driver’s seat (for now).
Then, and only then, should companies move onto proof of concept. Like any AI application, conversational AI shouldn’t just be released into the wild on a hope and digital prayer – test, iterate, deploy, evaluate. Have employees eat their own dogfood; challenge them to repeatedly call and message the agent with a gamut of real-world scenarios – encourage obstinance and reward failure. When you’re ready to put it in front of customers, manage risk by initially feeding it to only small pieces of traffic, such as callers who press a certain number on legacy IVR (Interactive Voice Response) systems, before easing slack on the leash.
More Than Just ‘Show Me the Money’
The impact of conversational AI investments can’t just be measured in dollars and cents. Every company’s definition of ROI is unique and should reflect the outcomes and metrics already defined and documented at each stage of their customer journey – what parts of the journey are the most complicated where automation risks a worse experience? What are the simplest where automation brings consistency and scale? Start by looking at how those channels have been measured historically as a baseline.
An obvious indicator of success: do customers want to have less or more interaction with an AI agent? Be wary of metrics that capture speed to resolution – did an intelligent voice or text agent relieve support tickets in record time, or did customers just quickly pivot to a human agent because the AI was ineffective and led to more questions? Companies must have the capabilities to recognize, the courage to act, and the conviction to regroup when their implementation is a dud rather than a difference maker.
Undoubtedly, conversational AI is enjoying a renaissance moment, but it’s not simply an all-purpose brush to paint over root customer service issues – sometimes a problem just needs a website refresh or improved UX. Companies that understand this and approach their investment in intelligent communication technology from the lens of the customer journey are poised to usher in an era of more satisfied customers, more productive teams, and more tailored brand experiences.