Customer Service Chatbot KnowledgeBase
Getting Started
Configuring the website widget
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Widget Settings`.
Step 3: Change the various fields in the `Customize Widget` pane as you like. As you make changes, you will see the changes reflected immediately in the `Preview` pane on the left.
Step 4: Click `Update` and test how the widget will look like by clicking on the test widget on the bottom right hand corner.
Step 5: Once customized, click `Copy Script` and paste the widget script into the <Body> or <Head> tags in your website.
Embedding the website widget into your website
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Widget Settings`.
Step 3: After configuring the website widget as you would like, click `Update`. Then click `Copy Script`.
Step 4: Paste the widget script into the <Body> or <Head> tags in your website.
Deployment on Facebook and Telegram
The chatbot can be deployed on simultaneously on your website, on Facebook Messenger and Telegram. The chatbot will respond the same way on each platform.
Contact us through singaporechatbots.sg and we will set this up for you.
Difference between Manager and Staff logins
There are 2 different portals to manage the chatbot.
The Manager portal shows all tabs including `Broadcasts`, `Statistics` and `Configuration` tabs and allows the manager to configure the chatbot.
The Staff portal shows the `Customer` tab only and allows staff to respond to chats and requests.
Configuring the Chatbot
Setting up Welcome / Restart Message
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent Info`
Step 3: Change the `Restart / Welcome Message` field as required. This is the first message that the chatbot will send when the user first engages the chatbot. It is also the message that the bot will send if the user types “Restart” or similar phrases.
Step 4: The `Restart / Welcome Message` is always followed by `Restart Quick Replies`. These are buttons that appear after the message that guides the user on what he might want to enter next. Quick Replies should be separated by the `|` separator.
Step 5: Click `Update`. The bot will now respond as updated.
Setting up Follow On Message
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent Info`
Step 3: Toggle `Follow-On Message` to “On”. Change the `Follow-On Message` field as required. This message will appear whenever there are bot responses without any quick replies. For good bot design, we encourage bots to always provide a follow on message and follow on quick replies. This helps guide the user on what he might want to enter next.
Step 4: The `Follow-On Message` should be followed by `Follow-On Message Quick Replies`. Toggle `Follow-On Message` to “On” and change the quick replies as required. These are buttons that appear after the message that guides the user on what he might want to enter next. Quick Replies should be separated by the `|` separator.
Step 5: Click `Update`. The bot will now respond as updated.
Setting up the Chat with Staff pre-built feature
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent Info`
Step 3: The `Chat with Staff` feature allows the bot to understand when the user types “Chat with Staff” or similar phrases. This will cause the bot to create a new Request ticket that the user wants to chat with a human rather than a bot. Emails and SMSes can also be sent to alert customer service officers when such a request is triggered. Toggle this feature to `On` if desired.
Step 4: Change the `Chat with Staff Message` as required.
Step 5: Click `Update`. The bot will now respond as updated.
Setting up the Call Back pre-built feature
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent Info`
Step 3: The `Call Back` feature allows the bot to understand when the user types “Arrange Call Back” or similar phrases. This will cause the bot to create a new Request ticket that the user wants somebody to give him a call back. Emails and SMSes can also be sent to alert customer service officers when such a request is triggered. Toggle this feature to `On` if desired.
Step 4: Change the `Call Back Message 1` as required. This message should prompt the user to enter his phone number.
Step 5: Change the `Call Back Message 2` as required. This message will appear after the user has entered a phone number.
Step 6: Click `Update`. The bot will now respond as updated.
Setting up the Small Talk pre-built feature
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent Info`
Step 3: The `Small Talk` feature allows the bot to understand various small talk messages that the user may send, including “Who are you?” and “How old are you?”. The responses are already pre-built in the bot. Toggle this feature to “On” if desired. If you wish to cope with such messages in a customized way, set this feature to “Off” and enter each potential small talk question as a separate FAQ in the bot.
Step 4: Click `Update`. The bot will now respond as updated.
Setting up FAQs
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent FAQs`.
Step 3: Click `New` and you can enter a new FAQ that the bot will understand.
Step 4: Enter the following fields
- `Action`: This is simply a short identifying name for the FAQ. Don’t repeat Action names
- `User Says`: These are some phrases (separated by the `|` separator) that express the same user intent. Enter a few permutations that mean the same thing. With these few permutations, the bot will (using Natural Language Processing) understand similar phrases.
- `Answers`: Enter how the bot should respond when the user enters the phrase above. If you wish, use the `|` separator to provide different answer possibilities. The bot will rotate amongst these answer choices.
- `Quick Reply`: These are buttons that guide the user as to what he might enter next.
- `Button Title`: Sometimes, the bot should just provide a button as part of the answer (e.g. to link to a website or map). In this case, enter a title for the button here. If a button title is entered, the Quick Replies will not appear.
- `Button URL`: Enter a URL that the user will go to when he clicks on the button.
Step 5: Click `Update`. The bot will now respond as updated.
Setting up a Conversation Flow
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Agent FAQs`.
Step 3: Set up an FAQ and enter a quick reply that guides the user what he might enter next.
Step 4: Set up another FAQ in response to the earlier quick reply. For example, in the above image, after the user enters “About”, the “Main Menu” quick reply will appear. When he clicks “Main Menu”, the bot will respond based on the “Main Menu” FAQ.
Step 5: Repeat the above and you can create a tree of FAQs which act guide users along various conversation paths.
.
Setting up Email and SMS Alerts
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Configuration` -> `Alert Settings`.
Step 3: Edit the email address and SMSes for appropriate request type
- request.types: Emails and SMSes will be sent to these addresses and number for all requests.
- sentiment: Emails and SMSes will be sent to these addresses and number whenever any customer displays low sentiment as measured using the Automated Sentiment analysis of the bot.
- back: Emails and SMSes will be sent to these addresses and number whenever any customer requests for a `Call Back`.
- to.agent: Emails and SMSes will be sent to these addresses and number whenever any customer requests to `Chat with Staff`.
Working with the Chatbot
Replying to Customers
Step 1: Log in to the Manager or Staff Portal using the credentials provided.
Step 2: Navigate to `Customers` -> `Conversations`.
Step 3: From this screen, you can see conversations that you have not read. Click on the chat icon to select the chat that you want to respond to.
Step 4: In the individual conversation screen, you can manually send messages to users (regardless of the platform with which they engaged the bot). Remember to click `Pause Bot` to stop the bot from functioning, so that the bot will not jump in to try to respond to questions while you are conversing with the user.
Step 5: Of course, remember to `Unpause Bot` after you are done chatting with the user, so that the bot can respond to any new messages from the user.
Handling Requests
Step 1: Log in to the Manager or Staff Portal using the credentials provided.
Step 2: Navigate to `Customers` -> `Requests`.
Step 3: Requests are tickets that are automatically created when the user makes a request that requires human intervention such as “Chat with Staff” or “Call Back”.
Step 4: Chat with the user using the chat icon if necessary.
Step 5: Remember to update the status of the request according to the below
- Pending: All requests start off as `Pending`.
- Processing: Update a request to `Processing` if it is being dealt with but is still work in progress.
- Cancelled: Update a request to `Cancelled` if there was a mistake or the request cannot be properly dealt with.
- Completed: Update a request to `Completed` once it has been properly dealt with.
Step 6: The time taken to deal with requests, based on status changes, is recorded in `Statistics`.
Training the Bot by Updating FAQs
As you populate the bot with more FAQs, hopefully, the bot will be able to deal automatically with most questions by users without human intervention.
To train the bot, you should add more FAQs.
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make configuration changes).
Step 2: Navigate to `Statistics` -> `Messages Not Understood`.
Step 3: This is a list of messages that the bot did not understand. Look through the messages and decide which user messages you want to provide an answer for.
Step 4: Enter a new FAQ to deal with that user message. The next time a user enters a similar message, the bot will reply based on the new response.
Sending Broadcasts
Step 1: Log in to the Manager Portal using the credentials provided (Note: If you log in to the Staff Portal, you will not be able to make broadcasts).
Step 2: Navigate to `Broadcasts`.
Step 3: Click `New` and enter the Broadcast text, or follow the instructions within to send images and videos.
Step 4: Select `Broadcast to Active Customers` or `Broadcast to All`. Active customers are those who have engaged the bot in the last 48 hours.