Apr 16, 2020 9:59:21 AM
Thinking about getting a customer service chatbot? Read this first.
It’s no secret that customers get annoyed when they have to wait in line. Whether it’s waiting 12 minutes before a phone agent picks up their call, or 12 hours for a reply to their email, waiting just isn’t fun.
Customer service chatbots are a trending topic in customer service organisations around the world—and for a good reason. They significantly reduce both resolution and first response times. Time that most customer service teams would gladly spend on important requests that require empathy.
And frankly, chatbots are great. We have full faith in this technology and the potential it has to positively transform customer service for both the end customer and the company. That being said, chatbots aren’t suitable for every use case and not all customer service organisations need to invest in them. The ROI of a customer service chatbot can vary greatly, depending on the size of the company, its support structure and business model.
Like any business project, bots need to solve predetermined business problems, improve KPIs and fit into the overall business strategy.
However, if your customers are not reaching out to you through digital channels, or your customer base only consists of highly individualised use cases that cannot be standardised, you’ll have a hard time implementing a chatbot that’s worth the money.
So, our recommendation is this: before even discussing the type of chatbot you might need, you should evaluate whether you actually need one at all. We’ve outlined the following questions to help you do just that.
1. Do your customers want to talk to you?
How big is the inbound contact volume? Chatbots don’t reach out to customers on their own (yet)—they are better for replying to incoming requests. So if your customers aren’t contacting your organisation directly, there is no real need for a chatbot.
If, however, your customers like to get in touch via email, contact forms or chat, a chatbot could be the right option for you. Chatbots yield the best results when they’re on the frontline and where your customers are<link?>. Depending on your business, that can be a service chatbot beacon across your website, but also on Facebook, WhatsApp, or within your mobile app, if you have one.
Customers are most likely to give your chatbot a shot if it’s easily accessible and visible on the channels they use to interact with your brand.
It’s worth noting that even if most of your inbound contacts come via “traditional channels”, such as the phone, you might still be able to redirect them to the bot. To do that, make sure your customers know about this service. Ways to raise awareness about your chatbot include having the chatbot beacon on your FAQ pages and app, mentioning it to customers while they wait in the phone queue, or linking to the chatbot in auto-replies.
2. Are your customer requests repetitive?
Chatbots learn and work best when their tasks are repetitive. If the questions and answers vary a lot, it will be hard to successfully train a chatbot to resolve those requests.
If you are working in direct-to-consumer industries, such as banking, consumer services or an e-commerce company, a vast majority of the requests will most likely revolve around a few, select topics. Unless the products or services you offer are incredibly “made-to-measure”, both the questions and answers should follow clear patterns.
With our customers, we’ve typically seen the following: 20% of customer problems account for 80% of the requests. In short, the 80/20 rule also applies to customer service. So, in cases like these, a chatbot will bring significant value to your service department—and your overall business.
3. Are you able to categorise those requests?
The last question is the hardest. Even if your answers to the questions above were affirmative, that’s still no guarantee that a chatbot will improve your service experience. Ultimately, the ability of a chatbot to yield a positive ROI depends very much on the type of customer requests you want it to solve.
The “Customer Service Request” matrix
The matrix shows two variables that significantly impact the ability of chatbots to deliver value:
- “Get-it-done” vs “Troubleshooting” requests, and
- “Self-service” vs “Need-agent” cases
Most chatbots work best in the get-it-done self-service category. Typical requests in this category are things like “What are your delivery times?” or “How do I change the cookie settings?”. They are basically questions that could be answered by a FAQ catalogue, albeit much more slowly.
For a chatbot to solve need-agent cases, you have to provide it with extra information. Issues might include “Where is my order?”, “Cancel my booking”, or “I need to block my credit card”. From a technical perspective, these types of requests can be automated with relative ease, as they tend to be very repetitive and usually come with a standardised solution or course of action (e.g. retrieving the shipping status when provided with a tracking number).
In such cases the chatbot also needs to be connected with an API on your site. Here, the technical setup can become a limitation. If your organisation does not have the relevant APIs or the bot cannot be connected to them, these types of request cannot be automated. Instead, they will have to be handed over to the service team.
Troubleshooting requests are generally the hardest ones to automate. They could be something along the lines of “I am experiencing a bug. How can I fix it?”. Commonly, the problem is not articulated very clearly, as customers usually simply ask “What is wrong with my device?” or “Why isn’t the service working?”. Getting to the bottom of such vague requests typically requires human intervention.
If your service team routinely receives lots of troubleshooting requests, you have to evaluate internally whether a chatbot would really be able to help your customers. The key here is to ask enough clarifying questions, so that you can get down to the root of the problem as fast as possible. Given the complexity of many troubleshooting issues, however, it might not be feasible to automate these cases.
If you answered “yes” to the three key questions listed above, then there’s a good chance your organisation would greatly benefit from a customer service chatbot. If implementing a chatbot is part of your customer service strategy, the next steps would be to outline and validate the chatbot project goals. As with any business project, you need to know exactly why your service team requires a chatbot and what benefits it will deliver.
To help you get started, we’ve put together a “Beginner’s guide to customer service chatbots”. It’s full of valuable tips that will enable you to choose the right type of technology and vendor, as well as define your project roadmap—so your chatbot will be a success.
Anna is part of Solvemate’s marketing team. She enjoys being creative, but has always had a weak spot for science and nerdy stuff, too. In her free time she loves expanding her horizon through travelling and many other opportunities that come her way.