It would be great if this didn’t overlap another Front Event I plan on attending!
https://front.com/resources/new-features-that-improve-logistics-agent-productivity 

How do you use the branch by keywords function for chat? Can the keywords directly link to the knowledge base?

What is the best way to use the chat bot during and after hours? Is there a way with rules to use the same chatbot for both situations?

What are some recommended uses for a chatbot?

Can customers restart the chatbot flow after they’ve gotten a question answered? What does it look like for the customer?

Are there analytics for the chatbot flows that customers click through? I’d want to know what people are interested in so I can bring those improvements to our service.

Is the chatbot AI-powered?

@JohnW I hear you! This doesn’t typically happen, but the scheduling panned out this way. You can still register for the chatbot webinar and the recording will be sent to you even if you don’t attend 

Thanks for your question! Using the branch by keywords step after an open-ended question is most helpful for identifying certain topics / inquiry types and classifying your conversations accordingly. For example, you could search for keywords “urgent” “important” “ASAP” to identify inquiries that are urgent in nature, and can choose to end the chatbot flow immediately for those customers to route them to a chat agent. Or, we have some customers who will search for expletive keywords and choose to end & archive the chatbot flow at that point.

Once you’ve identified the relevant keywords, you can certainly provide knowledge base article links to help resolve their issue. For example, if you are a travel company and you identify the keywords “cancel” “remove booking” “refund” etc., you can send them a link to a knowledge base article about cancelling & getting a refund on a travel booking.

There are two ways to use the chatbot during/after hours!

1. You can choose to remove the chat widget from your website entirely during offline hours using our offline hours feature in Front Chat channel settings. This will prevent your customers from chatting in when your team is offline.
2. However, if you’d like to keep the chatbot up when your team is offline to deflect certain inquiries, you can create an auto-reply rule. You can configure it so it only triggers when your team is offline. I recommend responding with a message template along the lines of “Our chat team is currently offline but we’ll get back to you via email once we’re back online!” This auto-reply will be sent after the chatbot flow is complete.

With both the above methods, you won’t need to manage a separate chatbot for online vs. offline hours.

Are there new features coming for chatbot?

Great question! Here are a few recommended uses for a chatbot:

1. Reducing repetitive questions your chat agents ask each customer: There are certain questions your team probably asks each chat visitor repetitively (e.g., what’s your name/email, what’s your order ID, etc.) You can save time for your team by automating this information collection so they can focus on resolving other inquiries
2. Triaging, classifying, and routing incoming chats: You can use the chatbot to automatically tag incoming chat messages based on topic (e.g., tagging based on inquiry type, product, etc.) This enables you to create routing & assignment rules so the chat is picked up by the right teammate based on the topic. You can also use tags to track commonly asked topics in Analytics.
3. Deflecting FAQ-type inquiries: Chatbots are great at deflecting commonly-asked questions (things like “How can I reset my password?” “How do I update my billing information?”) You can include information and resources directly in your chatbot flow to help resolve these types of inquiries. I recommend including Help Center links for commonly-asked questions. And, you can take advantage of our AI Answers feature (coming next year!) to make deflecting these inquiries even easier!

Currently we don’t have out-of-the-box chatbot analytics. However, our team is actively working on building out a new Chatbots report in Analytics. You’ll be able to measure the below metrics, and this what the upcoming report will look like:

1. How many chatbot flows were initiated
2. How many chatbot flows were completed
3. How many conversations the chatbot deflected from your team
4. Deflection rate (%)
5. Drop-off rate (%)
6. Average time to completion

In the meantime, before this is live I’d recommend including tags in your chatbot flow so you can tag inquiries by topic and track popular topics using the Tags Analytics report.

Great question - we’re currently working on a very exciting new AI-powered chatbots feature called AI Answers! AI Answers trains on your knowledge base content and automatically responds to your customer inquiries based on your knowledge base articles. This will be a powerful tool for deflecting conversations from your team so you can focus on resolving more complex, higher-touch inquiries.

Your customers can always let the chatbot know if AI’s response is helpful or not. If AI didn’t fully resolve your customer’s inquiry, you’ll be able to pass the chat to a human agent who can close out the conversation.

Beta testing for this feature will kick off in the new year - if you’re interested in participating, reach out to community@front.com and we will reach out with more resources once the beta program is live!

Our Support team at Front keeps our chatbot online 24/7, so we created an auto-reply rule during the weekends (outside of normal business hours) to let our customers know! Here’s what that looks like in Front: 

Yes! Our team is investing heavily into chatbots so we have a number of upcoming features & improvements coming soon:

1. Chatbot analytics: You’ll be able to measure key metrics for your chatbot including chatbot flows initiated and completed, deflection rate, dropoff rate, and average time to completion.
2. Chatbot subflows: You’ll be able to build and maintain your chatbot more easily by using subflows to easily reuse steps in your chatbot flow. Set up a subflow (which consists of a series of chatbot steps) once and reuse it multiple times in your chatbot flow. Edits made to your subflow will automatically be applied everywhere it appears in your flow.
   
3. Chatbot flow previewer: Soon you’ll be able to test your chatbot flow to make sure your steps are configured correctly without having to set your chatbot live on your website.
4. Transition conversations from chat to email: We’re building a much easier way to transition a conversation that began on Front Chat to email. This is helpful if your team needs more time to look into a complex chat inquiry and wants to follow up over email, or if you want to keep your chatbot online during offline hours but respond to any questions that come in via email once your team is back online the next day.
5. Dynamic variables: You’ll be able to define variables (like customer_name or order _id) to further personalize your chatbot flow. Once you’ve defined these variables, you’ll be able to conditionally branch your chatbot flow based on them. For example, in the flow below there’s different chatbot behavior based on which website URL your customers are chatting in from:
   
   Note: These designs are not finalized.
6. AI Answers: As mentioned in another response, AI Answers is coming to chatbots soon! You’ll be able to leverage AI to train on your knowledge base content and automatically resolve chat inquiries. 

You can also keep track of upcoming feature releases in our Public Roadmap (“In the works” tab)!

At this time, chat visitors (customers) do not have the ability to restart the chatbot flow on their end, but this will be possible in early 2024! Here’s a sneak peek of what that might look like in the future, stay tuned 👀



Also mentioned here, if the chat visitor is actively responding to the conversation after 6 hours, the chat widget will not refresh until there is a 30-minute window without a response. The chat widget on your website will automatically refresh 6 hours after the conversation is initiated as long as 30 minutes have elapsed without a response from the chat visitor. If the chat visitor is actively responding to the conversation after 6 hours, the chat widget will not refresh until there is a 30-minute window without a response.

Here’s what the customer would see when the conversation has ended or has been auto-archived (when the “Archive and end chat” option is selected at the chatbot’s end path)

Our Support team at Front leverages the branch by keyword step by asking the chat visitor to input the exact error message they see!

If we detect specific keywords in the error message, the chat visitor will be taken down various paths depending on the error message they receive. Some paths have a message that links the visitor to reference the relevant knowledge base article on how to resolve the error in question. For other paths, they will be sent a message the chat visitor unique troubleshooting steps that we outline in the chatbot message.