When a Chatbot is not the Best Start for Customer Service
It’s a tough time to be in customer service. Quality care has never been more important, yet savvy consumers have never been harder to impress. At the same time, satisfied customers aren’t necessarily loyal. If you are a startup, the brand is not there to begin with.
(Scroll down for a 2 minute video version)
A Tough Nut to Crack
As per a study published in Harvard Business Review, albeit a bit dated, customers are four times more likely to leave a service interaction disloyal than loyal. The same study found that:
- First, delighting customers doesn’t build loyalty; reducing their effort — the work they must do to get their problem solved — does.
- Second, acting deliberately on this insight can help improve customer service, reduce customer service costs, and decrease customer churn.
This is great news for customer service managers. Getting back to basics seems easy compared with “exceed all expectations, everywhere, all the time”. But customer needs change constantly. What’s the key to customer service excellence? It may seem counterintuitive, but there’s a case for full-scale digitalization. If done right, McKinsey reported that digitalization leads to higher customer satisfaction while lowering costs. Customer satisfaction jumped almost 20% when switching from traditional channels (phone, email, etc.) to digital.
Customers want what they’ve always wanted — easy access to high-quality information, and quick and efficient resolution to questions and problems. Now, they also expect to have their pick of medium — whether this means customer forums and chatbots, FAQs, trouble ticket systems, and phone calls. Taking advantage of digital mediums doesn’t let you off the hook for providing a personalized experience, however.
Digitalizing customer service is a big undertaking, but by tackling it in steps you can deliver the quality and consistent care that will create loyal customers.
Preparing the Ground
Let’s start with a plot. On one axis of this plot is the complexity of a customer service use case, which could range from activating a card sent over email (simple) to resolving engineering issues for a highly technical product like an integrated circuit chip (complex).
On the other axis is how much do you that the customer should solve this issue in self-service way. Contrast that to one of your team members (or your affiliate’s) interacting live with the customer to resolve the problem. It’s okay to say that you want your representatives to solve a certain use case for now, and may be you will think about making this self-served in future.
Now let’s take stock of the major issues your customer service team faces and sort them into three buckets:
- Simple use cases for customers
- Complex use cases for reps
- Complex use cases for customers
I have ignored the fourth bucket — Simple use cases for reps. You really should not have anything there.
Generally speaking, you want to tackle the simple stuff first, then move on to more complex cases that will only be handled internally — this allows you to experiment and work out the kinks away from customers. Finally, when you’ve got a working system that you feel comfortable with, roll out digital automation for more complex customer-facing issues.
Trending Bot Articles:
1. 8 Proven Ways to Use Chatbots for Marketing (with Real Examples)
2. How to Use Texthero to Prepare a Text-based Dataset for Your NLP Project
1. Simple Use Cases for Customers
Your simplest use cases probably take up a large amount of you customer service team’s time, but condensed they probably account for only a few different customer issues. Think refunds, shipment tracking, account questions.
Rules-based chatbots are going to be your go-to for these cases. If the question and solution can be mapped to a decision tree (see below) you’re golden. For more information about chatbots, you may want to look up ‘A Business Leader’s Guide to Chatbots.’
Pros:
- Chatbots are by far the quickest and easiest intelligent automation to deploy.
- They’re cost effective
- If done right, any deviation from the decision tree will be escalated to humans right away, so there’s limited chance for error.
- Customers are familiar with chatbots, so there’s no learning curve for your users.
Cons:
- Not scalable beyond a few very simple use cases. Decision trees quickly get out of hand for even remotely complex problems.
- No leverage from AI, which means your chatbot will never learn. Your chatbot will always perform exactly as programmed, and every change to your process must be carefully managed by your team.
2. Complex Use Cases for Representatives
Once your decision tree is working well for the simplest customer questions, you can move on to more complex tasks. Before you do, it’s important to understand the concept of intent as it relates to digitalization. Intent is the user’s intention, regardless of the language they choose to communicate. For example, a customer who has lost their password wants to get into their account. They may say “Lost password”, “forgot password” or even “I need to access my account”, but the intent is the same — to regain control of their account.
You can probably guess that this gets complicated very quickly. Forget simplicity for a moment — imagine you’re in charge of customer service at TurboTax or SAP. Think of all the details and combinations of customer entries you’d have to map on a decision tree if you wanted to automate. Dizzying, isn’t it? Thankfully, this is where Natural Language Search steps in.
Because it takes time to train an NLS-powered AI solution, we recommend that you bring NLS to your backend customer service processes first so that your customers don’t witness the adjustment and training period. Instead, your customer service agents will still answer customer calls, but instead of finding and delivering the right answers themselves, the AI will do the heavy lifting.
Depending on your data strategy, you’ll need the help of an AI vendor or your in-house team. Feed all relevant documentation into the AI, and introduce your customer service reps to the system. Using the AI should be very intuitive since NLS is built for human language and they won’t need to worry about keywords.
You’ll be surprised at how easy your employees take to a tool that speaks their language. This is called the virtuous cycle of AI — accurate, easy-to-use systems reward the user by making their jobs easier. This drives more traffic to the AI, which trains it better, improving it further. Don’t be surprised if you’re able to completely avoid the painful learning curve you’re probably used to when implementing new software.
Pros:
- NLS-driven AI scales to infinite complexity without any hand-holding. It learns through use, so it will stay in sync with your team without you programming in every little change to procedure.
- It personalizes for every user, based on use.
- Your CS rep is still in the driver’s seat at this stage, ensuring accuracy and a positive experience for your customers
- Automating backend processes first give your organization a chance to learn the ins and outs before end users gain access.
Cons:
- You’re not yet truly automated, as your reps manage the whole process. This won’t last for long, however.
Quick note here — although we do recommend rolling out AI to your agents first, we have had success in direct deployment as well. If your business is relatively simple or your on a deadline to digitalize, we can advise on whether you can skip this step.
3. Complex Use Cases for Customers
By now your AI is running smoothly and reliably — time to roll it out to your customers! Don’t be nervous — your team is still ready and waiting to assist when needed, but with the repetitive work offloaded, they’re fresh and ready to tackle more complex customer issues.
Final implementation is not much different than rolling out AI for your reps. We encourage you to be very clear with users about what to expect. Let them know that they’re not using a “dumb” search too — they can ask questions in full sentences, and in natural language. They won’t have to sift through huge documents to find answers hidden on page 126 — the AI will deliver the exact answer needed. If they still can’t find what they’re looking for, a human is always available to help them out.
Pros:
- Scales to infinite complexity while allowing for a personalized customer service experience with every use.
- Saves big on cost, as human involvement is minimal.
- Most importantly, your customers will like it. With 60% of customers already turning to digital self-service tools when they have an issue, making these tools more helpful and easier-to-use will make your customers happier and improve NPS.
Cons:
- Needs sufficient training and investments (this can be minimized through deployment on the backend first, and training time can run as low as four weeks).
Digitalization isn’t a nice-to-have, it’s critical for 21st century organizations. NLS and AI-based solutions may seem futuristic now, but they’re already taking intelligent workflows by a storm at companies large and small. As a startup it’s especially important to institutionalize the right mindset early on.
If you take one thing away from this (longish) article, let it be that you must roll out AI driven solutions internally before expanding them to your customers.
Don’t forget to give us your 👏 !
