Chatbot vs Human: How To Combine the Best of Both Worlds
04/03/2024 min de lectureCharles Kergaravat
Table of contents
Customer experience leaders are unsure when to use AI chatbots vs. human support. For many, it’s tough to find the balance between efficient scalability and an empathetic support experience.
Even before the widespread use of artificial intelligence, 59% of consumers felt companies had already lost touch with the human part of their customer experience. Yet, we know AI can help provide the 24/7 service customers crave.
So, how can support teams make these two functions work together?
In this article, we’ll compare the capabilities of AI chatbots and humans, diving into why both are critical ingredients in a winning customer support strategy. We’ll also give you some tools to upgrade the quality of your agents’ service.
By the end, you’ll be able to make more confident decisions about when to use AI and when to double down on the empathetic support only humans can provide.
What you’ll learn
The current state of play with AI chatbots
How much of your customer support should be AI
How other people are using AI chatbots
How to measure the effectiveness of AI support
How to upgrade the human element for a competitive advantage
Where are we now with AI chatbots?
In recent years, artificial intelligence has seen groundbreaking advancements, becoming more accessible than ever for most businesses to set up and deploy for various functions.
The most budget-friendly updates to the technology, particularly for the customer service sector, came in the form of chatbot advancements.
Today’s AI chatbots use technologies like machine learning (ML) and natural language processing (NLP) to converse with customers more humanly.
Both of these systems allow machines to interpret and learn language the way people do:
Machine learning helps chatbots improve and adapt as they speak to more customers and collect more data
Natural language processing perceives and weighs what each customer is saying (including context) to respond with best-fit answers
ChatGPT, for example, uses these processes to interpret user prompts and respond with natural-sounding messaging:
In this example, the system “understands” the positive meaning behind phrases like “definitely recommend.”
It also picks out other words to help understand the broader context. For example, “intuitive platform” refers to the product’s usability and “chat support” to its customer support team.
By highlighting these terms, the system can filter feedback for human team members accordingly.
Thanks to these use cases, it’s no wonder every company seems to be experimenting with chatbots.
“If you are not thinking about how to apply AI in your business, it’s a huge missed opportunity. Companies that use it will enhance their entire customer service journey by freeing up human agents to proactively deliver a value-adding experience for customers with more complex issues.”
Many websites feature live chatbots that provide quick responses and reduce the workload for human agents.
All this technology sounds impressive. But what’s the impact of AI on the customer experience?
Have we delivered on the promise of AI revolutionizing the customer experience?
AI and the use of chatbots are undoubtedly making waves in the digital customer experience sector. There’s been plenty of speculation about AI “replacing” humans, but chatbots have a long way to go before that becomes a reality.
More importantly, customers don’t want to deal solely with chatbots. Multiple scientific studies report “lower consumer trust in and preference for chatbot service” compared to human agents.
That’s not to say AI hasn’t improved anything with all its advances. AI tools have plenty of benefits, including:
Improved personalization. AI is helping businesses personalize their messaging and improve the sales experience. Zendesk’s 2024 CX Trends report suggests 70% of CX leaders believe bots are becoming “skilled architects of highly personalized customer journeys.”
Improved insights. AI can analyze vast amounts of CX data, interpret customer behavior and preferences accurately, and predict upcoming trends. For example, Starbucks’ proprietary AI platform Deep Brew analyzes data to tailor the customer experience with personalized marketing and customized menu recommendations.
But it’s not perfect.
For starters, AI is not unbiased. Training data, algorithms, and results often over-represent certain groups while underrepresenting others.
This bias has real world effects. In one case, scientists discovered a popular healthcare algorithm that led to substandard care for Black patients. Similar issues can occur with other marginalized groups like women or the LGBTQ community, especially in professional settings when they’re applying for leadership positions.
So, while AI is a cheaper way to scale, customers still want and need individually tailored experiences for more complex or nuanced problems.
Why human interaction is still key for customer support
AI is not a one-size-fits-all solution. Instead of one or the other, chatbots and human reps should help each other.
As the technology grows more sophisticated, chatbots will be able to respond instantly to a broader range of use cases.
However, a human touch is irreplaceable at crucial points in the support spectrum.
Only humans can understand the complexity of emotion
Chatbots won’t cut it when a customer is angry, dealing with a tricky problem, or navigating an irregular experience.
AI can say the words “I’m sorry,” but it can’t say them with the understanding and empathy a human can.
This difference can be a make-or-break moment when 80% of American consumers get frustrated with chatbots, leading to negative experiences and customers who don’t come back.
AI should be a supportive tool
Instead of replacing humans with AI for these purposes, you should implement the technology to support them. AI can be a useful tool, working with your agents to make their jobs easier instead of replacing them.
For example, a recent study of over 5,000 chat agents found access to generative AI tools increased productivity by 14%.
So, how much of each solution (chatbots and humans) does your customer support function need?
Is AI a garnish, a side, or the whole dish?
People should be the foundation of your customer experience, while AI should supplement and help you scale.
CX leaders know AI can help them but don’t know whether to use a little (like a garnish) or a lot (like a side dish). Some companies make the mistake of letting AI take over the whole support department – which can lead to immense frustration, as you’ll see in later examples.
Zendesk’s 2024 research suggests customers are split on bot/human assistance. When people want “immediate service,” bots win by a fraction (51%).
The scope of this preference varies, but when booking airline tickets or requesting order statuses, digital interactions with AI win.
This research suggests your industry and budget will likely determine which tasks and interactions you hand over to AI.
For example, a company selling office equipment may want its chatbot to handle admin tasks, like lead generation and billing queries.
In contrast, a medication supplier may want human agents to handle inquiries that could impact a website visitor’s health or involve sensitive medical information.
Before making a decision, you need to know the strengths and weaknesses of each solution.
The upsides and downsides of chatbots
Customers are getting more used to automation and AI as widespread implementation continues.
For example, Qualtrics research suggests 48% of consumers are comfortable interacting with a company’s AI.
Here are four ways chatbots can help and hinder businesses.
Upside: Multilingual and available 24/7
AI chatbots are available for customers at any time, with most able to understand many languages — which can help you adapt when growing a global customer base.
This preference indicates not everyone wants instant service if it lacks human interaction, so you should offer customers access to both types of support.
Upside: Cost-effective
Chatbots can handle high volumes of queries faster than humans and require little maintenance after setup.
They’re cheaper than salaries and the expense of turnover, which is appealing if you’re looking to cut costs and increase profitability.
Appen VP of AI Strategy Mike Shwe says we can expect the tech to begin to “capture the expertise of IT specialists in high-leverage, time-consuming tasks such as data migration and software integration” in 2024.
While some CX leaders may worry they’ll lose the skills of human representatives, these trends suggest a shift from AI “generalists” to “specialists.”
Downside: Time-intensive to train
With more capability comes higher upfront and ongoing costs.
While chatbots provide almost instant responses to queries, the training to get them to that stage is far more in-depth.
You’ll need to compile a large dataset of conversations and corresponding responses. Then use machine learning algorithms to analyze patterns and teach the chatbot how to respond appropriately based on context.
For larger teams, Enterprise Bot estimates one chatbot could take around 4–12 months before it’s ready to deploy fully.
That’s because the knowledge base for an enterprise-sized company is much larger than the one for a small business.
First, you need to build a chatbot that can process this amount of data. After that, label all the information appropriately so the software can understand the context behind it. Then train and test it before deploying it to customers.
While human agents also take time to train and onboard, you can usually assign them to customer interactions within a couple of weeks.
For small businesses and startups, this difference in time and training could have a huge impact on your bottom line.
Downside: Lack of humanity
Chatbots lack the human touch some interactions demand.
For example, ECCCSA research suggests European consumers’ preferences shift toward human assistance when dealing with more emotional, urgent, and post-purchase interactions:
Human reps can also help you avoid negative publicity if the technology fails.
In one instance, a frustrated customer got European parcel group DPD’s chatbot to write a poem about how useless it was, use profanity, and label the company the “worst delivery firm in the world.”
According to French media, one of the posts got 800,000 views in 24 hours — ensuring the embarrassing blunder didn’t go under the radar.
While a disgruntled employee could technically do the same, the responsibility would lie with the individual.
The DPD case shows technology has limits that can impact brand image and customer retention.
Downside: Environmental impact
Training and running any AI chatbot requires a huge amount of power and electricity.
In fact, one University of Washington professor estimates ChatGPT requires the same daily power as around 33,000 homes. Training it even “directly evaporates” a vast amount of freshwater — around 700,000 liters.
As the use of this technology grows and organizations release more sophisticated models, these systems could end up using as much energy as entire nations.
In the midst of climate change, this huge carbon footprint isn’t an impact companies should brush off lightly.
The upsides and downsides of human agents
While people are becoming more comfortable with company chatbots and AI assistants, many still crave interactions with real humans.
According to Qualtrics research, 62% of customers prefer human customer service channels over digital in 2024.
Here are four ways agents can help and hinder businesses.
Upside: Personalized support
As more companies automate their interactions, you can stand out with customized support that people remember.
According to Adobe research, customers say “highly personalized experiences” are one of the top ways brands make themselves more memorable.
In general, personalization means creating a CX that meets each person’s situation and requirements.
Let’s say a user wants to subscribe to your product but believes it’s too expensive.
Your rep may offer a customized quote for your service by removing several features. The user appreciates you’ve tailored the product to their needs and saved them money, so they sign up.
Humans are naturally creative problem-solvers. Unlike AI, we can solve challenges with unique solutions and build long-term relationships in the process.
Upside: Emotional intelligence and empathy
Speedy resolutions are important. But customers place understanding and empathy for human emotions just as high on the priority list.
Customer service platform Dixa’s research indicates 96% of customers believe empathy is “very important” during support interactions.
For example, your reps may share with a caller that they’ve made a similar mistake in the past before suggesting a solution.
Sure, chatbots can use tools like sentiment analysis to interpret human behavior:
But it’s not always accurate. Many phrases have double meanings that change the context. Plus customers may use slang or idioms that chatbots don’t register and struggle to process.
On the other hand, most humans can register how others are feeling from tone and expressions — even if what they’re saying contradicts that.
For example, someone saying “Can I get an update?” might be calm or frustrated. Talented customer service reps will pick up on tone to provide a better customer experience.
Downside: Higher costs
While the ongoing cost of AI chatbots can vary widely, the maintenance can be a lot lower than human agents — especially for smaller businesses.
However, most businesses still need human agents to handle queries. Chatbots are great for fact-finding questions (like “what is my customer number”), but inevitably, customers will come across more complex issues and need to speak with a human.
The balance you’ll strike between human and chatbot comes down to this: Do you see your contact center as a cost center or profit center?
If the goal is to optimize customer support spend, think about how complex your offering is and consider how many basic queries could be answered with automation. If your agents deal with common questions and simple tasks every day, a chatbot may help you save money and speed up response time.
Downside: High employee turnover
The turnover rate for customer support agents hovers between 30% and 45% globally, with the average tenure for entry-level agents only around a year.
Businesses must then use valuable resources to hire and train new employees.
According to Dixa, one explanation for such high turnover is 59% of agents lack the knowledge to provide better service.
To encourage your reps to stay and help them succeed in their roles, support them with adequate tools and training. That’s one area where AI can help behind the scenes rather than out front in a customer-facing role.
For example, AI can support your agents by analyzing call sentiment and providing insights so they give the right answer next time. That way, you’ll combine human communication with AI-powered research and insights.
Supporting agents and giving them the tools to thrive (like Apizee) can help decrease turnover by ensuring agents feel supported and fulfilled at work.
When to use humans and when to use AI: 4 use cases
You can use chatbots for time-consuming, repetitive tasks and leave your human agents to creative problem-solving and relationship-building. This harmony can reduce response times and improve customer satisfaction.
Reps are even beginning to see the positives of AI rather than worrying about it replacing them.
While over half the support professionals in a recent Hiver survey believe a human-led support strategy works better, they also see the advantages of AI.
Here are four industries using both types of support to enhance the customer experience:
Ecommerce and retail
When selling products online, customer queries come in fast.
Most of them are pretty easy to handle. For example:
“Where’s my order?”
“How long are shipping times?”
“When will you process my refund?”
“How can I return my order?”
Chatbots can handle these straightforward order tracking and refund processing queries. Many brands have already implemented them for this purpose.
For example, fashion brand H&M offers a chatbot for support on its website:
If the chatbot can’t help with your query, you can speak to a human during business hours.
And if you use specific keywords and phrases the chatbot doesn’t understand, you’ll be routed to a human agent automatically. Or you can type “Speak to a human” to accelerate the process:
Customers with complex problems or complaints likely want to speak to live agents who can use critical thinking (something AI currently lacks) to solve problems.
Human reps can take over in these cases to ensure they resolve all queries and keep customers happy.
Finance
The finance sector deals with high-risk personal assets and digital platforms can make some customers nervous.
Nordea is a Norwegian bank that combines AI and human support teams to alleviate these worries.
Until recently, you could use its chatbot “Nova” to answer general queries from the bank’s “open pages.”
Now the technology can answer and solve over 90% of queries across 2,300 banking topics. Plus help with self-service functions like blocking a bank card or finding a bank account number.
However, human reps still have their place in the user experience.
If you need more tailored help or have a tricky issue, you can arrange a face-to-face video chat with a financial advisor.
The advisor can also share their screen (a feature Apizee also offers) to help troubleshoot problems with your account. Or show you relevant graphs and presentations if you need monetary guidance.
Healthcare
Personal health is a particularly delicate topic for customer service representatives — especially when self-diagnosis with “Dr. Google” tends to make patients fear the worst.
Thanks to this cyberchondria (the fear of illness made worse by online searching), people tend to want answers from trained medical providers.
In the healthcare sector, chatbots are mostly used for admin tasks like appointment scheduling and prescription ordering:
Patients understandably want to speak to trained, real people when discussing medication or symptoms.
However, some apps like Visiba Care are using cutting-edge AI technology to combine AI with doctors, nurses, and other healthcare professionals.
A chatbot starts by asking multiple questions to take initial details and symptoms from patients. It then offers three diagnoses based on probability and history of answers.
It acts as an all-in-one human and chatbot solution. The technology can:
Access your company’s knowledge base and write responses for agents in seconds
Search the web for additional information and sources
Guide new agents with “Smart Scripts” as they learn
Transcribe calls in real time
Provide “next best action” recommendations based on past calls
Agent Assist helps automate contact centers through the speed and efficiency of chatbot software combined with a human’s critical thinking and scope of emotional responses.
Talkdesk also integrates with other support tools like Apizee to empower your team to resolve queries even faster and more effectively.
Measuring AI support vs. human support with metrics
Measuring return on investment (ROI) is crucial for all customer service solutions. But remember your chatbots and human agents are likely handling different types of queries.
You won’t track the success of FAQ answers in the same way as more complex interactions. So make sure your ROI metrics for AI vs. human support account for this.
AI metrics
You’ll most likely implement chatbots to enhance operational efficiency.
In this case, you can use these three metrics to measure success:
First contact resolution (FCR). FCR measures how often your chatbots can resolve an issue during the first interaction. It’s worth noting that chatbots often respond to easy questions like “What are your hours,” so they typically have a higher FCR rate than human contacts who are dealing with more complex queries.
Average handle time (AHT). Handing over basic interactions to chatbots should reduce your AHT, but don’t sacrifice quality to resolve queries quickly.
Escalation rate. You can track the efficiency of your chatbots by how often they have to escalate interactions to human agents. The better they get, the less this should happen.
Keeping track of these figures will ensure you’re providing the best and most efficient service.
Human metrics
While AI focuses on more basic tasks, human agents can focus on more complex or nuanced customer needs.
Tracking the quality of service is a little more complicated, but these four metrics can help dictate success:
Customer effort score (CES). Low-effort customer interactions can be a good indicator of your team’s problem-solving abilities.
Customer satisfaction (CSAT). An excellent CSAT score can suggest high-quality human-agent interactions.
Net promoter score (NPS). The score will be higher if your digital touchpoints address customer needs quickly and your human agents are empathetic. Notably, respondents who started digitally but ended by speaking with a human gave higher loyalty scores than customers who used exclusively digital means.
Issue resolution rate. While you can’t solve 100% of issues, your resolution rate should be high and always end with concrete next steps (e.g., “We’ll be in touch when we release that new feature”).
Customers still appreciate the human touch, so keeping an eye on these numbers is key.
Upgrade your human customer service with Apizee’s visual engagement tools
You may rely on your human agents to develop innovative solutions when new issues arise. But are you setting them up for success?
One scientific study suggests it’s not only the complexity of the query that dictates whether people would prefer to speak to a chatbot or a human. The “problem-solving capability” of the agent plays a huge part.
Solutions also require in-depth, accurate background information. Sometimes a phone call or live chat isn’t enough for your reps to collect the facts they need.
Doing so can help if your business struggles with these problems:
Customers need a seamless way to connect with you for efficient issue resolution
Service agents don’t always understand complex customer problems remotely
You want to foster stronger customer engagement through personalized and empathetic support experiences
You need tools that enable lasting customer relationships and loyalty
Customers don’t need to download software (like Zoom) or have an account to access these services.
Apizee’s secure click-to-video functionality means human agents can manage and resolve remote cases quickly with extra visual help.
Align your chatbot and human support for a more fulfilling customer experience
Many of the support tickets your team deals with are low-value and repetitive. In these cases, AI chatbots are the perfect solution to get customers faster answers and reduce the workload for human agents.
But for the rest of those queries? The unique situations, angry customers, and delicate matters? Only humans can navigate situational and emotional complexity to instill customer trust and confidence.
In an ideal world, chatbots will give your customers instant answers to 80% of their problems. For the other 20%, your human agents can be on hand to provide higher-quality, one-to-one support through visual engagement tools.
The chatbot should augment the agent’s experience by relaying information so the customer doesn’t have to repeat themselves. A bot can also provide summaries and customer interaction notes while guiding team members through the next steps.
The human agent can then focus on solving complex problems and serving customers quickly with personalized attention that improves the overall experience.
Get in touch today
Learn how Apizee can facilitate more efficient, engaging customer conversations when chatbots alone aren’t enough.
Learn how to implement co-browsing — a more secure form of screen sharing — to protect your customers’ data while walking them through digital processes.
