Last week, our CEO, Dan, had the pleasure of participating in Zendesk’s Startup Central event where he talked about AI, Customer Experience and how the two are related.
During our session, we also covered the following topics:
Since we’re big believers in the concept that sharing is caring, we’ve decided to bring you the complete transcript of the event.
Have additional questions? Join the conversation on Startup Central.
Let’s dive right in!
After leading a team at OrCam, I realized - we have this amazing technology that can help the blind see. Yet, when I call my bank, I have to be on hold for 20 minutes to understand I was waiting for the wrong representative.
The gap between the available technical capabilities and the experience customers are getting was mind-blowing. It still is. That’s why I felt that I have to use my knowledge to improve the CX field. That's how Plantt was born.
Even before Covid, many companies struggled to match customers' demands and scale their support organization fast enough, which led to violating SLAs and a drop in customer satisfaction.
Many companies are searching for a solution that can help them scale the team quickly without having to waste time training new employees - a process that can take a few weeks or even months.
Many companies implemented AI in chatbots to try and handle the increasing demand from the customers. AI and chatbots can, ideally, help customers navigate and understand the product better.
However, adding AI is the first step in the process of improving the Customer Experience.
The key is data, data, and data. We at Plantt have seen chatbots that weren't based on customers' data vs. ones that were tailor based - and it's a clear cut. Custom chatbots work 10x better. You should always keep customer data in the loop.
You know your customers. You and your team interact with them daily, and to make your AI work, you need to leverage that data and implement it in your AI.
As far as I’m concerned, it’s the only way in which you can accelerate your development and boost the end user experience.
How conversational do you want your human reps to be?
You probably want them to be friendly, pleasant, and helpful towards the customer. And there’s no reason your AI shouldn’t act the same - over time.
Adding a name, some fun gestures, or even emojis can go a long way for some companies. Others might want to provide a quick and to-the-point response to satisfy their customers.
In both cases, you want to provide an excellent communication experience while keeping a consistent tone of voice - whether customers communicate with an AI or a human.
So the short answer is yes, it should be human-like. But there’s no need to sugarcoat it, let customers know they are interacting with a bot.
You need to understand it's not about "when" you should add AI, but rather "why" and "how."
For your AI to be impactful and useful, you need to answer the simple questions of why do you want to add it - aka - what are you optimizing for.
You need to know which issues you want your AI to solve before spending time, money, and company resources implementing it. We see it in many companies - they have a fancy AI that can't answer the most common questions.
Every company has its timeline, demands, and customers. I believe it's essential first to get to know your audience and clients before you can set up an AI, so you'll know what they're expecting of you and your product.
Once you have the data, you can understand which issues you want your AI to handle, the best way to do so, and how much it can help you and your team.
To make a long story short - each company has its own "correct" time, but usually, it all comes down to the data you have. The key is always to start small and grow with your customers.
There are many different AI types available when the most common one is probably the chatbot.
Behind the scenes, AI can provide additional tools and abilities, such as identifying and predicting intent, NLP analysis, sorting and routing of issues, conversational designer, etc.
As I mentioned, each company has its own definition of “a helpful tool”, and it’s essential first to understand why you want to add the AI in the first place and what issue you’re trying to solve.
Obviously, generic FAQs are the widespread use case for AI, but it’s important to mention that today AI can handle far more complex conversations than FAQs.
AI can fully understand user intention during the conversation and perform conversational tasks, such as getting and updating account information.
However, not all issues can be solved with the help of AI. Some queries or issues require a more complicated process, which is best handled by a human being.
I would go with the 60-40 percent rule (when 60% of issues can be solved with AI).
Here’s an image from Plantt that shows this principle. We can see the AI can deal with tasks like delivery status, issues with items and receipts very well, while more complex matters (missing or wrong items) need to be handled by human reps:
As part of our day-to-day at Plantt, we can see a need to implement AI in different areas, such as the sales department, product teams, etc.
The data that customers are currently providing through their conversations can be valuable to these departments as well, whether in finding issues inside the product that needs attention, understanding what areas sales teams should focus on in their initial training, and so on.
There’s no doubt AI will advance beyond the customers; it’s just a matter of time & data.
I think it’s best to start small and grow with your customers.
A good AI might be able to take the load off your team, allowing you to focus not only on hiring people, but letting your existing team focus on the issues and tickets that are critical.
First of all - AI is all about managing the expectation of your users. I've seen companies trying to hide from their users that they are talking with a chatbot, and you end up with a bad customer experience.
Always be transparent with your users about what the chatbot can and cannot do.
Also, I think that AI does not excel in "exploration" tasks, such as shopping or browsing an e-commerce website without a clear goal. Sometimes it's better to give the users time to explore the website and options in it.
Once the users know what they want or what they're looking for, AI can step in and provide guidance, help them complete their order, and give them any additional help or support.
I think I learned (and still am learning) a lot from all of the people I worked with, past and present.
As for my management style, I believe in a flat organization. Everyone can contribute to each team, and we often have brainstorming sessions with the heads of all of the departments about our roadmap, goals and future initiatives. It leads to the newest and imaginative ideas, and each team can have a meaningful impact on our product.
Shout out to the awesome Greg Geibel who made this Q&A happen!
If you have any other questions, we’d love to answer them! Join the conversation on Startup Central, or reach out to us directly at firstname.lastname@example.org