Does My Company Need AI?

Written by Emily Cummins  on   Oct 4, 2018

It’s easy to get excited about AI and customer service. After-all, it promises to elevate the customer experience by making sure that every time a person reaches out to your company, they have a brilliant customer experience: personal, immediate and convenient support on their terms.  But it doesn’t make sense for every company to adopt AI customer support agents today.

Here are the things you should consider when you’re evaluating whether or not to adopt AI to power your customer service.

How to Evaluate Whether You Need an AI Agent

#1. Volume of Tickets

The higher the volume, the more benefits you’ll reap with an AI simply due to the number of customers AI can manage automatically, optimizing support resources. The benefit of high volume, however, goes deeper than that. AI that leverages deep learning gets better from experience, and the more interactions it has, the smarter it will become. If an AI is managing a fraction of tickets in the first month, this would be expected to rise significantly higher over time with high volume.  If you have lower volume, an AI can still impact your customer service, but its knowledge trajectory will not be as steep.

#2. Repeatable vs. Non Repeatable Issues

Do at least 40% of your tickets span repeatable issues? AI can learn how to offer high-quality responses to repeatable issues, enabling your human agents to manage the rest. If your company gets a lot of unique, first-time and complex scenarios, your support would be better managed by a combination of human and AI agents. Use AI to classify tickets and route it to the right agent automatically, so it helps your human agents perform even more effectively.


Help agents work faster and smarter with a Zoho chatbot.


#3. Business System Flexibility

Can the business systems your company uses for support tickets be accessed via APIs? A Support AI needs the ability to resolve, not just respond. To meet consumer expectations, it needs to handle both generic and account-specific inquiries. Depending on your business, resolving issues might be accessing an order management system to provide shipping status, tapping into a CRM to pull up loyalty information and help with reward redemption, or sync up to a booking engine to facilitate the rebooking of a flight. Being able to pull relevant data in real-time will provide the authority that consumers expect an AI to have.

#4. Training Data

AIs need to be trained, and the more data you have available, the more tickets it will be able to resolve and the higher the confidence it will have in providing a response. Training data can include historic emails, call center logs, social media exchanges, FAQs and other marketing collateral. If you don’t have access to this type of data, compiling a database of the issues and how the AI should respond will be your first step.

#5. KPIs  

What will success look like for you?  Is it how many tickets an AI can solve on its own versus needing to escalate to a human? Is it CSAT?

What metrics are you trying to improve? Many of our customers seek to decrease the amount of time it takes to resolve a ticket, the total cost it takes to resolve an issue and even monetization led by AI-driven personalized support.


Click here for the 15 best customer experience tools to elevate CX.


While many people have traditionally measured AI on its ability to behave like a human using the Turing Test, this is not what should be the focus. It should be on how well the AI can solve problems. How well it is at providing high-quality resolutions, how effectively it is supporting human agents with recommended replies and intelligent routing, the impact on the time it takes to close tickets, and how it is learning over time.

Asking how you are going to measure the success of your AI upfront allows you to set clear expectations.  It’s essential that all of the stakeholders within the organization understand the AI training process, and that how well an AI performs on day 1 is not indicative of its overall impact. It will improve over time, but it needs experience.

#6. Support Channels  

What are the most common channels your customers are reaching out on for support? AI Agents work well on email, social and Web/app chat. While AI Agents can deploy an omnichannel experience to provide the most impact, understand what your highest trafficked channels are to create a high-impact roll out plan.

AI-powered customer service is the future. Let’s chat and see how Netomi can help you transform your customer experience.

For more information on customer service, visit:

Do Consumers Want AI Agents?

Written by Emily Cummins  on   Oct 4, 2018

It’s easy to wonder what consumers want and expect out of AI.

Are they terrified that AI is taking over the world, as some media reports would have you think? Or are they mystified with Hollywood’s portrayal of AI in movies like Her and i, robot?

With AI Agents, we’re creating a new consumer behavior. We’re asking humans to trust an AI to help them in a moment of need. We’re asking them to spend their time engaging with an AI Agent, carrying on a conversation with an AI, talking to it as they would a friend.

We wanted to understand exactly what your customers think about and expect from AI.  Here’s what we learned from 1,500 real people.

What do consumers think about AI for customer service?

Takeaway 1: Consumers are visionaries

It turns out that consumers are really excited about AI and foresee many ways it can impact their day-to-day lives, from providing a “simplified life” as it can “handle mundane tasks” to empowering them to make “better decisions.”

Action – Your AI Agents need to have meaningful interactions with customers: help your customers accomplish a task or get a resolution quicker; proactively alert them of issues, or provide more personalization at the moment to aid in better decision making.  

Takeaway 2: Consumers are tuned in and have clear expectations

We found that consumers are educated about AI.  They have a smart and clear picture of what they want and what they don’t want. Above anything else, they expect an AI to be accurate, effective and authentic.

Action Consumer expectations are not unreasonable. Ensure you’re leveraging NLU to effectively understand what your customers are saying in their natural dialogue, be it text, images or emojis (today’s “digital slang”).   If you’re reaching a global audience, have localized AI Agents trained to understand local slang and nuances.

Train the AI with existing data sources like call center logs, historic emails,  social media logs, etc. so the AI learns from real exchanges and even the top agent performers.  Training an AI with FAQs is important, but by also having it learn from past tickets, it can pick up on conversational distinctions and grow its neural network.  

Takeaway 3: Consumers feel short-changed by most AI interactions

Consumers feel like most AI interactions have been unhelpful to date as they are not as good as humans at accomplishing things or have not been given the authority to make decisions.

Action AIs need to have the authority to resolve tickets, not just respond to customers. Empower the AI to approve or reject a return request, upgrade someone on a flight or refund someone based on pre-trained parameters. If AI always has to rely on a human to close a ticket, the consumer is going to believe the AI has no value and is only causing a delay as a middle-man.

Takeaway 4: Consumers want to be heard

Consumers are driven by emotion when they reach out to brands. Being heard and listened to is the most important attribute for customer service interactions, and resolution and a human voice help deliver this.  Customers do not want a relationship with brands. Quality and consistency in a company’s service and experience is what matters.

Action Empower an AI Agent to take the next best action in a particular circumstance, whether that is resolving the issue autonomously or escalating to a human (say, if it’s a complex or unique situation). An AI Agent should also be able to understand tone and sentiment and tailor a conversation to show empathy when needed. Integrate with various systems like CRM to provide the most helpful resolution to the individual based on their unique history with the brand.

AI is transforming customer service, making it more convenient and resolving issues faster than ever. Consumers are excited about how AI will impact their daily lives, but the AIs need to meet consumer expectations. Empower agents to listen to them and solve their problems, personalize to the individual when necessary, and have an empathetic ear.

To learn more about AI-enabled customer service and how msg.ai can transform your customer support, let’s chat. For more information on how to help agents work faster and smarter, check out our Zoho chatbot.

For more information on customer service, visit: