2 Methods for Customer Support Chatbots
For overworked, understaffed customer support teams, chatbots are an excellent solution. Chatbots can be an incredible tool for offloading commonly asked questions and saving your team’s valuable time. It’s important to be realistic about the resources required. We have created two bot methods and the time/resources you should expect for each method.
Method 1: Decision Tree
Effort required: 1-2 hours
Technical expertise: None required
Resources needed: Any Instabot subscription
Ideal for 5-20 FAQs/Support topics
1. Open Customer Support template
Use the Instabot “Customer Support” template to quickly build a decision tree bot. You can preview it here and even select “Use this Template” to start an Instabot trial using this template if you don’t have an account. If you do have an account, create a new bot from a template and select “Customer Support”.
2. Edit FAQs
By default, this bot has fields for six FAQs across two categories but you can easily add or delete options depending on how many topics you want to answer.
Method 2: Dialogflow
Effort required: 1-2 hours (initial bot) + 15-30 minutes per Intent in Dialogflow (topic)
Technical expertise: Minimal
Resources needed: Instabot subscription Standard and above, Dialogflow account
Ideal for 20+ FAQs/Support topics
Accuracy: 75-90% depending on time spent training bot
1. Setup Dialogflow
Create your Dialogflow intents and answers. More info on how to do that here! In order to get around an 80% accuracy rate, we recommend entering at least 150-200 training phrases per topic you want the bot to be able to detect. If many of your questions sounds similar you may need even more training! If you’ve been using Method 1 start by exporting all of the data you already have so you don’t have to start from scratch!
Connect your Dialogflow bot to Instabot
Here’s how. It’s easier than ever
Train train train!
Your bot will never be 100% accurate in understanding questions, so it’s crucial to understand that there will always be a margin of error. If you work in an industry where that doesn’t work, stick with Method 1! You can always log back into Dialogflow to improve its accuracy.