A deep dive into the CX question and answer life cycle

Introduction

Customer experience (CX) teams face increasing pressure to deliver timely, accurate, and personalized responses to a growing number of customer inquiries. This challenge is compounded by the wide range of customer needs and multiple touchpoints. CX agents must also master an overwhelming amount of information about their organization's policies, procedures, and processes.

Do we really need to think hard about how questions and answers work? A customer asks a question, my CX agent answers that question. Simple, right? When we look closely, we see that it is not simple at all.

Each time our agent answers a customer's question, and delights that customer in the process, it means our agent has navigated a very narrow path. Providing support that our customers love might seem easy, but when we take a closer look there is more to it. It only looks easy when we get it right.

Let's dive into what I call the CX question and answer life cycle. We will outline the four key steps involved in answering a customer query, understand the challenges and complexities of each step, and finally see if AI might be the perfect partner to help teams streamline, enhance, and personalize this process.

The CX question and answer life cycle: an overview

The life cycle can be broken down into four stages:

• Understanding the question
• Finding the knowledge
• Understanding the content
• Answering the question

These steps form the foundation of any customer service interaction, whether in a call center, via email, or in a chat session. They may seem straightforward, but each presents different challenges that can bog down our teams and result in longer response times, low-quality answers, and frustrated customers.

Step 1: Understand the question

This seems easy and obvious, but is it? If a customer already knows the answer, they will ask a perfectly clear question. But they are asking for help precisely because they do not know the answer, so their questions are often confused and off the mark.

It is our agent's job to interpret the intent behind the question. They must also understand the context in which it is being asked. This can be difficult even if the agent is experienced. If the agent is inexperienced, it becomes even harder.

What if the question is asked in a foreign language? How much extra effort will it be to translate the question into the language the agent can handle?

It is not just about the words. It is about understanding the underlying need and divining the intent.

Step 2: Find the knowledge

Every organization provides a knowledgebase for their agents to store up-to-date information about the organization's processes, procedures, and policies. These knowledgebases range from files on network drives, to SharePoint servers, to intranets, to dedicated knowledge management tools.

Finding the data you need in these systems is universally an awful experience. Over time the organization's information becomes fragmented and strewn across multiple repositories, and the information can be incomplete, conflicting, or obsolete. The built-in search mechanisms are typically poor.

The news is not all bad. Our research finds that the Pareto 80/20 rule holds true in CX too. Experienced agents can typically answer 80% of questions off the top of their head — basically an instant answer for the customer. That leaves 20% of questions that require a trip to the knowledgebase.

We have found that a trip to the knowledgebase takes an average of three to seven minutes to find the required information. At best, this means a long hold time and a frustrated customer. At worst, it means the dreaded "can I take your number and call you back?" — which is then often followed by the agent needing to interrupt their supervisor to get help.

Step 3: Understand the content

The agent is not at the finish line when they have found the right documentation.

Now our agent must read and understand the content of the documents they found. This read-and-understand step can be more time consuming than the search. There may be multiple documents to read and review, the information might be scattered throughout them, there might be conflicting information, or the documentation can simply be difficult to decipher.

We cannot answer the question until we have read and understood the information. That takes time.

Step 4: Answer the question

Finally. Our agent has found, read, and understood our process, procedure, or policy. Now we can craft the response to the customer. The agent must synthesize everything found into a single coherent answer — an answer that fits the intent of what that customer asked originally, and that aligns with our preferred corporate tone and organizational voice.

We did it. Now on to the next call.

Recap

Looking through the lens of the CX question and answer life cycle, we see we are asking a lot of our CX agents. We want our agent's answer to leave our customer happy, satisfied, and confident. To do that, the answer must be fast, accurate, and match the customer's original intent.

The life cycle operates in a continuous loop. Each new call, email, or chat starts the cycle over again. Each question leads to interpretation, new searches, and new knowledge to understand and synthesize. If we improve any one of these steps, we improve our whole process, multiplied by each agent and each customer interaction.

Where can AI fit in the CX question and answer life cycle?

I often warn leaders that AI is not the solution to every problem, or even most problems. For many problems you are trying to solve in your business, generative AI solutions do not provide significant improvements over the existing tools you are already using. But:

AI is perfect for the question and answer life cycle.

AI can clearly and measurably improve each step. It is almost an ideal optimizer, and it can help CX teams respond to customers faster, more accurately, and more empathetically. Here is how.

Step 1: AI for understanding questions

Generative AI is excellent at quickly interpreting customer intent, regardless of the language used or the phrasing of the question. With AI, context is understood in real time, and intent is extracted with accuracy.

Step 2: AI for finding knowledge

AI-driven search tools can sift through massive volumes of documentation at lightning speed. The magic of vector databases gives us dramatic improvements in searching for relevant information buried deep in our documents. The benefit is not just the increased speed — it is the search quality.

A vector database paired with a large language model can find the right information where traditional keyword searches simply fail due to the reliance on exact matches. That pairing gives us the ability to search by understanding context and meaning.

Step 3: AI for reading and understanding content

AI can read your content at computer speed. Even if our search finds hundreds of relevant documents, a language model can read and understand those passages in milliseconds. Just think of the difference in time between your agent reading three pages of text and the AI doing it. And language models have no difficulty understanding your content — remember that the second L stands for language.

This processing speed difference might be the single biggest advantage of using AI in CX.

Step 4: AI for answering questions

This is the easy and obvious part. Language models are also great at writing text. AI can synthesize the data from the relevant documents and draft a great answer in your company's tone of voice, across various languages.

This allows our CX agent to review and personalize the response, allowing for high-quality, consistent, and fast answers to customer questions.

A caveat

You will note that I do not automatically advocate that AI should replace your CX agents. In fact, in many use cases the best use of AI is to support and enhance the CX team's efforts. Instead of replacing your team, AI can give your CX team superpowers.

Conclusion: AI is the future of CX

We have described the CX question and answer life cycle to highlight the hidden complexity of our customer service interactions. We do not realize how much load we put on our CX teams' shoulders.

AI is a surprisingly good fit to improve and enhance each step in the process. It helps reduce customer wait times and improve answer quality and accuracy. The resulting improvements in customer satisfaction and engagement are obvious.

AI is not here to replace customer service agents. It is here to empower them. CX teams are a critical part of the customer relationship we have worked so hard to build. AI can simply provide faster, more accurate access to the information our teams need to do their jobs. This enables our CX teams to focus on delivering exceptional, empathetic service instead of slogging through yet another painful search.

Embracing AI is a key to future-proofing customer experience operations and meeting the ever-evolving demands of customers.

Tell me what you think.

Keep reading