Chatbots are incredible automation tools that can streamline a variety of business workflows saving you both time and resources. Still, regardless of how much they can do, there are situations when human intervention is necessary. This is when a well-managed chatbot-human handoff comes into play.
Human takeover is one of the key aspects of bot implementation into your business processes. Hence, it’s to manage the transfer as smoothly as possible.
Who is the article for?
What is the article about?
Using a chatbot for automation allows you to cater to a greater number of leads and customers more efficiently and in a personalized way, 24/7. Bots help businesses reduce waiting times and reduce costs by taking care of routine tasks and questions.
Despite all the good work, there are key scenarios in which transferring the conversation to a human agent is essential or extremely advantageous. There are times you need that human touch to reach the best possible outcome.
Below are 5 pivotal scenarios when you should consider implementing chatbot to human handoff:
Chatbots, while quite smart, are far from omniscient.
After all, they are only as smart as we train them to be. Which, let’s be honest, is just not enough to deal with everything users are capable of throwing at them.
Once in a while, they will come across a task that is simply way too complex and requires creative problem solving only a human is capable of.
When it comes to this, your bot can try to understand the user’s intent by asking them to rephrase. However, if it’s too frequent or repetitive people get frustrated and the bot can do more harm than good.
Still, you can’t train your chatbot to be ready for every possibility.
Therefore, a much more reliable option is to train your bot to recognize its limits. That way, instead of failing to understand and help the user, it will be able to assess when a specific user request requires bot-human handoff. In that instance, the bot can leisurely say:
“I’m not able to solve this issue on my own. Please wait while I transfer you to one of our human agents who is sure to be of help!”
In the meantime, your agent will receive a notification and gain access to the previous conversation so the user doesn’t have to repeat the query.
Some situations are more critical than others. To be more precise, there are situations where a bot can speed things up and others where it can be a serious deterrent to issue resolution.
Often, users need to go down a specific conversation funnel and move through a sequence of dialogues to reach a resolution. However, in some cases, the interaction the bot needs to assess the situation can be counterproductive and a waste of time.
A great example is a situation when a customer is late for or has an issue shortly before the departure of a plane, train or a bus and the issue needs to be resolved promptly.
Chatbots are just perfect for answering common questions about travel rules, routes, timetables as well as booking tickets and sending user updates and changes to their schedule.
They are not so great when there’s an emergency when the plane is about to take off. Back and forth with a bot will only upset the customer and lose precious time. So, to prevent this from happening, you might teach your bot to trigger human takeover when the departure time is less than 20 minutes away (e.g. by booking code or by identifying some language markers in their input).
Customers in distress will be handed off to your human employees who will be able to assist them quickly and efficiently. And the bot? The bot will be taking care of the less urgent cases.
At times, people simply don’t feel like chatting to a bot. No matter how good your chatbot is in natural language understanding and processing, there’ll always be people who feel uncomfortable sharing information with an artificial construct. Hence, “user preference” is our chatbot-human handoff scenario number three.
Giving your users/customers the choice between a bot and a human is a great way to introduce your new chatbot to your customer base. You present the option but don’t force it on people ensuring, you don’t stir up some bad blood implementing sudden changes.
If you think that such a choice will mean no one will try to speak to the bot, you are very wrong. You would be surprised what people are willing to do to avoid waiting 😁.
To introduce the idea of a bot, the trick lies in NOT giving people a choice on a classic static interface in the form of a button BUT inside the chatbot conversation flow. That way, your users HAVE TO meet and interact with the chatbot YES or YES and so are more likely to give it a try.
Let the user have a couple of interactions but do place the option to speak to a live agent close the start of the conversation.
Offering such a choice will make for happier customers and better customer service.
First of all, if they make a choice to speak to a human agent over a chatbot, they are less likely to be frustrated with the wait because, after the all, waiting was their choice. The same goes for talking with a bot. That’s the magic of user-driven experiences. Secondly, you’ll also avoid users faking serious problems. Some smarty-pants out there might figure how your bot work and misuse it to trigger bot to human handoff when it’s not necessary.
This couldn’t be a real chatbot article if I didn’t talk about leads. Every business is hungry to not only get a substantial quantity but also quality. Hot leads are the bomb!
Hot leads mean consumers who are ready to convert. It’s exactly the kind of leads you want to hold on to when they visit your site or try to get in touch via WhatsApp or Messenger.
The great thing about chatbot is that they can react to evolving situations in real-time. Therefore, you can condition your chatbot to initiate chatbot-human handoff when the user fits the profile of a hot lead.
It’s been proved over and over that quick response time is the key to improving lead conversion. The famous Harvard Business Review study found out that companies that responded to a potential client within one hour of their inquiry were 7x more likely to qualify the lead.
Naturally, it’s impossible to respond to all leads so quickly.
BUT, thanks to your bot, you can assess user interacting with your bot in real-time and, when they check all the boxes of an ideal lead, send them off to chat with an agent INSTANTLY. No need to make them wait even that one hour!
Chatbot to human handover will prevent – or at least significantly decrease – the loss of the most qualified leads and so, increase your conversion rate.
The smartest AI lead generation chatbots are the ones that work hand in hand with your human team!
Last but not least, we have the unfortunate case of your bot failing more times that can be tolerable.
Now, ideally, you want to avoid chatbot fails. We all know that if a 100% success rate was possible, the chatbots would be already our overlords.
Therefore, every responsible bot owner should think of cases when the bot fails to understand what the user actually wants. I want to point out that this is different from the handover because of the complexity of the request.
Bot fails are either caused by users asking/saying:
The second group is the one that should interest you. There’ll always be people messing around with your bot but your energy should stay focused on those valid requests you didn’t think about.
For example, when a bot fails repeatedly with the same user, the failed user intents can be sent to your human team as a Slack or email notification. Then, it’s up to one of your support agents to determine if the intent is just a silly game to be ignored or a relevant request that requires agent engagement.
After each such incident, the agent will write up the case and retrain the bot (or send it to the dedicated bot-creator so he or she can do it). The request can fall under the simple queries category where the chatbot takes care of its resolution from start to end or to the complex query category. In the latter case, instead of being confused, your bot identifies the task at hand and facilitate chatbot-human handoff as in the #1 Complexity Scenario.
For human takeover to work like… well, clockwork, you don’t just need to master the “WHEN” but also the “HOW”. There are three different ways to ensure a smooth transition from the bot agent. It all depends on which of these works the best for you and your use case.
With Landbot, all you need to do to set up a chatbot to human handoff is drag arrow from the last conversation block and select to implement the “Human Takeover” block.
Below are the three different ways you can configure it.
When you open to configure the Human Takeover block, you are asked to “Choose the Assignation Type”.
The first on the list is “Simple Assignation”.
In this case, no special setup is necessary. The human takeover will happen without the bot asking any additional questions. Your human agent is free to jump in and carry on the conversation thread.
The only thing you can customize in this assignation scenario is which agent(s) will be responsible for taking over the conversation of this particular bot.
The second option is the Default Assignation Type.
When selected, before transferring the user to a human, the bot will ask a series of default questions. The questions are designed to capture the contact of the user and clarify the issue at hand. Thanks to such measures, the transfer can be smooth and the agent can help as quickly as possible.
Here, you can select to send an email notification to specific agents when the takeover is triggered. Simply enter their email addresses into the designated field and click save.
The third and last option is the Custom Assignation flow that allows you to personalize the takeover process in every bot.
It offers several different options that make it easy to adjust the handoff to a particular use case. Thus, your lead generation bot-to-human handoff, for example, will be different from the customer support bot-to-human handoff.
You can customize whether or not you want:
Similarly to choosing your takeover management style, you can also choose the way you inform yourself and your team about the human handoff trigger being set off in real-time.
Landbot chat manager is a section in the Landbot app where you can see any conversation happening on any of your active chatbots. You can access it by clicking on the chat icon on the left side of the Landbot dashboard.
From the Landbot chat manager, you can take over any of these ongoing conversations, directly from the chat manager interface (even if the human takeover wasn’t activated).
In any case, when the handoff is actually triggered, designated agents receive an alert.
This way, it’s much easier to monitor the conversations users are having with your chatbot on the web or WhatsApp and enable your agents to perform a chatbot-human handoff as frictionlessly as possible.
In the chat manager, the list of all past and live conversations appear on the left. In the middle, you’ll be able to see a particular conversation. On the right, there’s a field with all user data collected which you are free to edit if you collect any additional information during the live chat.
And don’t worry! When you opt to respond to a chat, the chatbot that the user was previously interacting with will be paused.
Ok, Landbot’s human takeover block allows you to send email notifications just by clicking a button. It cannot get simpler than that.
BUT, if you browse through the builder options, you’ll come across the “Send an Email” block.
This is a more sophisticated email alert. It lets you customize the content of the email and can be used to send emails to your team members as well as the customer.
For instance, it can be useful if you have more bot-to-human handoffs going on and you want to know the type of query coming in. Or, maybe, even if you are not the one taking care of the user, you want to be informed about certain handoffs taking place.
Last but not least, you can also set up a quick Slack notification to let you know a user is need of human attention. If your team breathes and lives through Slack communication, this is definitely the type of alert to go for.
To set it up, just create a Slack block just before the Human Takeover section.
Click add to Slack.
A new tab will open.
There, select the Slack account you want to connect with and then assign a person or channel to receive the notification about this handoff.
Then, go back to the Landbot builder. The channel you just connected with Landbot, will appear in the “Choose an Option” selection; Under the “ADD TO SLACK” button.
Lastly, define the message to be displayed as part of the notification. Using variables, you can pull user name, email or even a specific message they sent.
The lesson is simple. Humans and chatbots both have their strengths and weaknesses. Your strength will be learning which situations require a human touch and arranging for chatbot-human handoff to cater to that need.
There’re many different reasons to implement chatbot to human takeover but the art lies in finding the golden balance and leaving only the most complex and critical to your human employees.
It’s a win-win. Bots will take care of all the mundane banalities while your employees get more interesting cases that require thinking only humans can do.
In a way, having the perfect handoff procedure matters even more than having a perfect bot… at the very least, the perfect handoff procedure is attainable whereas the latter hardly so.