Engineers have come a long way since the 18th century and the famous Mechanical Turk, a "robot" supposedly capable of answering the questions of its interlocutor. Although it was a hoax at the time, the dream of the conversational automaton is now a reality, thanks to chatbots. Let's take a look at how to make the most of their potential.
What will your chatbot be used for?
This first step is crucial to optimizing your project's chances of success. It involves defining a clear, precise strategy. Here are a few questions to help you define your robot's field of action:
- Which conversation channels will it use? Is it just chat, or would it be useful to connect it to your professional WhatsApp as well?
- Will it be able to handle your customers' frequently asked questions?
- Who will be responsible for updating the chatbot? And how often?
Some companies make the mistake of embarking on a chatbot project without taking these issues into account. They run the risk of having a messy, poorly-configured chatbot that aims to do everything and be able to answer any question, but fails to do so effectively.
In the end, it's the customer who pays the price. We've all heard the story of a friend, colleague or family member complaining about a chatbot that always sounds like a broken record. But don't worry, there's no reason for this to happen to you if your strategy is clear from the outset! It will help you avoid the catch-all project and move towards the success you're looking for: customer satisfaction.
Always keep in mind the fundamental purpose of the chatbot
Ultimately, this is the most important objective of your chatbot. It has to make your customers' lives easier, providing them with an extra helping hand, a useful additional channel for solving their problems.
We'd love our customers to contact us just to ask our opinion on the color of the table they'd like to order, or whether the bouquet of white flowers sold on our site will match the walls of our living room. Unfortunately, this isn't always the case...
At the end of the day, when a customer contacts a company's after-sales service, the company's crucial mission is to provide the right answer to the customer's question as quickly as possible. If a potential buyer wants to know how much it costs to deliver a piece of furniture to Orleans on a Saturday between 6 and 10 p.m., the nirvana of professional success will consist in providing this answer immediately to the future customer, within seconds of the start of the conversation.
Ideally, the answer will be found directly on the product page, or in the Frequently Asked Questions section. But our valued customers don't always take the time to read this information, and that's where the chatbot comes in.
How can I be sure that my chatbot will be useful?
Building a chatbot project will require taking a deep look into your customer service data. The data collected over the past few months can provide invaluable guidance on the best way to build your chatbot. Here are just a few avenues to explore:
- What are the main types of ticket for which customers contact you?
- Which of the first five typologies could easily be handled by a chatbot? Which don't require an exchange with your human agents?
- What information will the chatbot need to process requests?
- Will the knowledge base enable the chatbot to respond correctly to the most frequent requests?
The answers to these questions will have a decisive impact on the rest of your project.
In fact, there are two very different ways to program your chatbot, depending on the complexity of the questions most frequently asked by your customers.
Scenario 1:
If the majority of exchanges concern quick, easy questions that can be answered with a simple phrase, a simple transmission of information, then the chatbot 's architecture can be developed. We'll be able to program a large number of answers, without running the risk of providing the wrong answer. Here are a few examples of basic questions that can be safely delegated to the chatbot:
- How much does delivery cost?
- How do I cancel my order?
- How do I return my defective product?
The last thing you want is to fall into the trap of seeing chatbots as a way of reducing your teams' workload.
Don't get us wrong, this is an obvious advantage of a chatbot configured according to the rules of the art. However, we must never lose sight of the fact that, above all, it must add value for your customers. The chatbot must be an efficient, lightning-fast interlocutor, capable of answering your customers' queries in an instant. On the other hand, a chatbot is certainly not a tool for, if you'll pardon the expression, driving your customers round the bend in an endless spiral of vague and useless responses, unrelated to the initial request. In that case, it's better not to set up a chatbot!
Scenario 2:
If the majority of exchanges concern very complex, specific questions that could not easily be listed in a predefined database of answers, for example :
- I want to transport four cubic metres of goods from Monaco to Brest on Sunday. How much would that cost?
- I'd like a quote to redesign my garden, using only purple flowers.
- What is the power consumption of your Max Wind 9000 fan?
In this case, it would be preferable to set up a simple chatbot, whose aim will not be to answer all the questions by itself. The questions most frequently asked by your customers are too elaborate, and it's impossible to define all the variations of potential answers in advance.
So, instead of being a perfect know-it-all, your chatbot will play the role of assistant to your teams. Instead of answering questions itself, your robot will qualify your future customers' requests. To do this, make sure it gathers all the relevant information upstream, before transferring the request to your sales team, for example.
Focusing on the future
Having a chatbot that works and responds satisfactorily to your customers is great. But you also need to think about the future.
To use a floral metaphor, your chatbot is like a potted orchid. When you buy it, it's beautiful, it doesn't need watering immediately, you just have to admire it every day and enjoy its beauty. But the more days go by, the more it will need to be cared for, including watering, removing wilted petals... Well, little chatbots need attention too.
To ensure your chatbot's longevity, there are two key factors to consider.
The first is to design a clear, comprehensible architecture from the outset, right from the start of the project. It's easy to get lost in endless decision trees, literally going off in all directions. To avoid this pitfall, we try as far as possible to document the motivations behind each level of the chatbot. Why was such a change made? Was there a good reason not to ask for such essential information at the start of the journey?
The second is to update your chatbot at regular intervals. The questions posed by your customers can evolve over time, as can your products and your site, which are not fixed in time. For this reason, your knowledge base can become obsolete: some answers are no longer valid, or new issues have emerged, with their share of new questions.
To maximize your success, remember to monitor customer satisfaction with the robot's answers. This rate is measured by a simple question at the end of the conversation, such as "Did the answer given help you?", or "Are you satisfied with this answer?" If half of your customers tell you that the chatbot didn't hit the nail on the head at a specific stage of the conversation, you'll need to ensure that it's corrected as soon as possible.
