A pair of feet standing on asphalt with two white arrows painted on
Feature

Powering Customer Experience Through Conversational AI, Analytics and Good Data

9 minute read
Scott Clark avatar
Delivering exceptional customer experiences that give consumers control comes down to the right tools and good data.

Customers today demand a personalized, seamless experience throughout their preferred channels — and they want to control the narrative.

By using conversational artificial intelligence (AI) along with good data and analytics, brands can provide an exceptional customer experience based on the customer’s previous interactions and data from current sessions.

This article will look at the ways brands can create such an experience, along with the challenges that often come about during the process.

How Does Conversational AI Enhance the Customer Experience?

Conversational AI that uses natural language processing (NLP), automatic speech recognition, advanced dialog management, deep learning and machine learning (ML) is advanced enough that it's likely to pass the Turing Test. This test determines if a computer program can perform well enough to fool a human into thinking they are talking to another human. As such, conversational AI provides a much more realistic experience than traditional chatbots.

Evan Macmillan, CEO of Gridspace, a contact center automation platform, told CMSWire that conversational AI was not very conversational until recently, and that previously, brands had to program hard rules for every possible customer intent and response.

“Today large, pre-trained language models cannot fully understand language, but they are extremely useful tools,” said Macmillan. “Instead of programming rules, brands can use these models to flexibly and accurately recognize meaning and converse in a more natural, less scripted way.”

AI-driven chatbots can be predictive and personalized, with more complex, fluid responses that are similar to human decision-making. These AI bots have access to a customer's previous interactions, typically through customer relationship management (CRM) software, can observe user-specific traits (location, age, mood, gender), learn conversational styles from past interactions and even take actions using tools such as robotic process automation (RPA).

Whether or not conversational AI is right for a brand depends on each brand’s specific use cases. “Conversational AI is just a tool, sometimes the right one and sometimes not, it all depends on the job to be done,” said Macmillan.

Related Article: 4 Ways Conversational AI Is Improving the Customer Experience

Do People Really Trust AI Tools?

Customer trust in AI has been improving greatly over the past few years. A Capgemini report indicated that 54% of customers have daily AI-based interactions with brands, and 49% of those customers found their interactions with AI to be trustworthy.

The trust in AI isn’t limited to customers either — employees trust AI too. An Oracle and Future Workplace [email protected] report revealed that 64% of employees would trust an AI chatbot rather than their manager, and 50% have used an AI chatbot rather than going to their manager for advice.

It also appears the majority of people enjoy having conversations with AI chatbots — 65% of employees surveyed said they are optimistic, excited and grateful about potentially having AI "co-workers" and almost 25% said they have a comfortable relationship with AI at their workplace.

When used appropriately, conversational AI can be an effective tool in a brand’s customer service toolbox. Customers want to be in control of their own narrative and prefer to solve their problems without having to speak to a live agent, provided that their issue is relatively minor.

Anthony Chavez, founder and CEO at Codelab303, a team of digital experience designers, engineers and producers, told CMSWire that conversational AI can be seen as a digital agent, a brand representative that is no less critical than a teammate working on-site in a brick-and-mortar location.

"Digital agents, albeit the meta equivalent of a teammate in a store, can also create hospitable, memorable and efficient interactions with guests and customers," he said.

How Can Data Be Bad, and What Makes It Good?

Data can be bad if it is unstructured, inaccurate, inconsistent, incomplete or contains duplications. Because data comes from a myriad of sources — some of which are siloed, while others are in different formats or databases and yet others are unformatted — there is no consistency, and it must be brought together in a structured, consistent way to be useful.

“Good data, within the context of conversational AI, means data that supports NLP, NLU (Natural Language Understanding) and ultimately intent identification — that is to say, the machine needs to understand what the user is asking, and then provide a human-like answer that the user is looking for,” explained Chavez.

To turn bad data into good data, it must be “cleansed.” Data cleansing is the process of fixing bad data in a data set, which involves identifying any errors and then updating, fixing or removing them, which improves the quality of said data.

Jory Hunga, business development manager at iPaydayLoans, an online payday loans provider, told CMSWire that most brands usually have a very large amount of customer data stored in their CRM systems, which is comprised of past interactions, transactions and chat and call session transcripts.

“However,” said Hunga, “a huge amount of this data often comes in an unstructured format like verbatim comments that can often prove to be difficult for human agents to sift through for insight.

“Through the use of conversational AI bots, brands are able to make use of advanced machine learning to quickly analyze their databases and find links between pieces of information much faster and more accurately than any human agent ever could.”

Related Article: How to Prepare Data for Ingestion and Integration

What Role Does Analytics Play in CX?

Customer analytics begins with aggregating and unifying data from all possible sources, including websites, mobile apps, email, chat, social media, customer service tickets and in-store visits. Once brands unify and structure that data, they can use it to create a holistic 360-degree view of each customer.

Using analytics to determine whether a user finds the answer they’re seeking is the key to crafting the best experiences, said Chavez. “The best sets of data for this purpose are a blend of quantitative metrics and qualitative metrics; after all, every user will have a unique fundamental opinion about having a conversation with a machine.”

Companies can use customer analytics to create personalized customer experiences and assist with customer service inquiries. Real-time data can help funnel live inquiries to the most appropriate agents at the beginning of the interaction with the customer. By interpreting and analyzing this data, brands can make the “next best decision” in the customer journey.

Learning Opportunities

Ben Hookway, CEO of UK-based Relative Insight, a text analysis software provider, told CMSWire that effective and efficient analytics are critical to ensure that brands are delivering experiences that consumers want.

"Conversational AI is a great leap forward but absolutely key to success is a willingness from businesses to leverage the untapped gold mine that is unstructured text data as a whole — the likes of customer reviews, call transcripts and survey open ends,” said Hookway, who added that it’s customer data that holds the answers to new engagement strategies.

With the ongoing and ever-increasing deprecation of third-party cookies, the effective use of first-party data is more important than ever.

“To engage effectively with consumers in an increasingly competitive and noisy landscape,” Hookway added, “businesses need to look at the major repositories of data which they already have — and which they are generating all the time.”

Because third-party cookies will eventually be phased out, marketers are looking for innovative ways to understand their audiences, he explained. “There’s no doubt that technology-driven analysis surrounding first-party data — often in the form of conversations with your customers — will be fundamental to future success.”

What Are the Challenges That Brands Face With AI?

Brands today want to stand out for the exceptional customer experiences they provide, so they must be able to monitor and improve experiences on a minute-to-minute basis. This real-time personalization creates an emotionally positive connection and shows the responsiveness of the brand.

That said, there are many challenges to creating such an experience, especially for companies using disparate technologies.

Tom Summerfield, retail director at Peak, a decision intelligence company, told CMSWire that he’s seen a number of businesses that have been approaching these challenges by designing and procuring “point solution tooling” to enable scaled personalized messaging, i.e., content management systems, customer data platforms, apps, website “search and merch” toolings, etc.

“All of these tools contribute to the modern customer stack,” said Summerfield, “but they aren’t integrated and they don’t talk to each other. They’re optimising channels in silos and, without a connected approach, risk enhancing one at the expense of another.”

A much better solution for brands interested in creating a seamless experience is a connected approach to personalization. “That’s achieved by adding an agile SaaS (software as a service) layer between incumbent backend systems of record and stacks of point solutions.”

“Increasingly, this is an AI/ML platform that can identify trends and add a layer of intelligence into those processes,” explained Summerfield, who believes that this represents the future of CRM: a central, agile intelligence layer fueling automated segmentation and product recommendations via an application programming interface (API).

On the other hand, Hookway told CMSWire that brands can layer data and compare it, allowing them to get more value out of it and gain rounder and more accurate views of audiences.

Doing this successfully, however, is not easy. “First up,” he said, “it calls for a rethinking of the labels of what is thought of as ‘marketing data’ and what is ‘customer experience data.’ It also requires a mindset change; a process of true collaboration — with leadership ready to grasp and encourage the opportunities that effective data mining and analysis presents. We can’t rely on any one tool to give us all the answers.”

Given that there are multiple open-source and licensable tools for NLP, NLU and intent prediction that exist today, the technology has become the easy part, according to Chavez. “Designing the experience, however, in a way that is differentiated and competitive, will always be a more nuanced and persistent challenge for businesses because, in essence, you are designing a brain.”

Related Article: What's Next for Artificial Intelligence in Customer Experience?

Final Thoughts

Once marketers have cleansed, structured and optimized customer data, AI, ML and NLP can analyze and use it to enhance the customer experience.

Conversational AI enables brands to provide customers with digital agents that know their shopping, purchasing and service details, facilitating a personalized conversation while enabling live agents to handle more complex inquiries — all while allowing customers to control their own narrative.