tgbot automatic reply workflow design: complete tutorial from welcome language, keyword matching to multi-level menu

#tgbot Auto-reply workflow design: complete tutorial from welcome message, keyword matching to multi-level menu

When your Telegram community or customer service bot receives hundreds of repeated questions every day, manual responses are not only inefficient, but also easy to miss important messages. Design a set of tgbot automatic reply workflow that allows you to use zero-code to automatically answer common questions 24/7, guide users to operate, and seamlessly transfer to human agents when necessary.

This article will start with the disassembly of the core module, taking TG-Staff’s visual command process editor as an example, and guide you step by step to build a complete workflow that includes welcome, keyword matching, and multi-level menus.

Why we need to design tgbot automatic reply workflow

A Telegram Bot without automatic replies is essentially just a “message relay” - the user sends a message and you (or the agent) reply. This can quickly become a bottleneck in the following scenarios:

• 7×24 hours response: Users may inquire late at night or on holidays, and manual labor cannot cover all periods.
• Frequently repeated questions: 80% of inquiries focus on several issues such as “price”, “how to purchase” and “customer service hours”. It is a waste of manpower to reply manually every time.
• User Guidance: New users don’t know what to do after entering the Bot and need a clear guidance path.

The core value of visual workflow design is: without writing a line of code, the interaction logic of Bot can be defined through drag-and-drop and configuration. You don’t need to hire developers, the operators can complete the setup and iteration themselves. This is exactly the core problem that platforms like TG-Staff solve.

Core modules of automatic reply workflow: welcome, keywords and multi-level menu

A complete tgbot automatic reply workflow usually consists of three modules that cooperate with each other. Understanding the role and design principles of each module is the prerequisite for building an efficient workflow.

Welcome - the user’s first interactive experience

The welcome message is the message the user sees when they first launch the bot (usually by sending the /start command). It determines the user’s first impression of the Bot.

Design principles:

• Short and Friendly: Introduce the functions of the Bot in one sentence, such as “Welcome to XX official customer service. I can help you check orders, learn about products, and contact human resources.”
• Provide clear next steps: Guide users through buttons or instructions, such as “View Menu”, “FAQ” and “Contact Customer Service”.
• Avoid information overload: Don’t cram all the features into one paragraph, otherwise users won’t remember it.

Keyword matching - accurate answers to frequently asked questions

Keyword matching is the “brain” of automatic replies. When user input contains specific words, the bot automatically returns canned responses.

Design principles:

• Cover high-frequency FAQ: Sort out the 5-10 questions most frequently asked by users, and set 2-3 synonyms or related keywords for each question (such as “price”, “cost”, “how much”).
• Distinguish between exact and fuzzy matching: Exact matching requires the user to input exactly the same keyword; fuzzy matching can identify the situation where the user input “contains” the keyword (such as “price” in “What is your price”), and has a higher hit rate.
• Keep your responses concise: Answer the question directly, you can include a link or next action button, but don’t write a long paragraph.

Multi-level menu - guides users to complete complex operations step by step

When users need to complete a multi-step process (such as submitting a work order, selecting a product category, and checking the order status), the multi-level menu makes the operation intuitive through button navigation.

Design principles:

• No more than 3 layers: Users can easily get lost if there are more than 3 layers. It is recommended to split processes above 4 layers into multiple independent workflows.
• Buttons per layer ≤6: Too many buttons will crowd the screen and affect selection efficiency. It is recommended to limit the number to 4-6.
• Each level of menu provides “return” and “manual” exits: ensuring that users can return to the main menu or transfer to a manual agent at any location.

Design your tgbot automatic reply workflow step by step: take TG-Staff as an example

Below we use TG-Staff’s visual command process editor to build a sample workflow from scratch that includes welcome words, keywords and multi-level menus. You can register for a 3-day free trial at TG-Staff official website and follow the steps.

Step 1: Create a project and enter the command process editor

Log in to TG-Staff Console and create a new Bot project (you need to have a Telegram Bot Token first). After the creation is successful, find “Command Process” in the left menu and click to enter the editor interface. You’ll see a blank canvas with the Node Toolbox at the top.

Step 2: Build the welcome node

Drag a Start Command node from the toolbox to the canvas. Double-click the node to edit the welcome text:

欢迎来到 XX 产品官方客服！ 👋
我可以帮你：
- 了解产品功能和价格
- 查询订单状态
- 联系人工客服

请点击下方按钮，或直接输入问题。

In the “Button Settings” on the right side of the node, add two buttons: 常见问题 and 联系客服. Each button can be linked to subsequent nodes. After saving, click “Publish” in the upper right corner, and your Bot will have its first automatic reply.

Step 3: Configure keyword matching rules

Add a “keyword node”, drag it to the canvas, and connect it to the welcome node. In node settings:

• Trigger keyword: Enter “price”, “cost” and “how much”, one per line.
• Matching Mode: Select “Fuzzy Match”, so that the user can also trigger when inputting “What is your price?”
• Response content: Enter a standard reply, such as “Our standard version package is 8.99 per month, and the professional version is 16.99. For details, please visit the XX page. If you need manual consultation, please reply to “Customer Service”.”

You can create multiple keyword nodes to cover different FAQs. For example, another node matches “refund” and “cancel” and replies to the refund policy.

Step 4: Design multi-level menu branches

Suppose you want to design a product consultation menu: the first-level menu displays “Product A”, “Product B” and “Return to the main menu”. After clicking “Product A”, the secondary menu displays “Price”, “Function” and “Trial Application”.

Operation steps:

1. Add a “Button Menu” node and set the first-level menu buttons: 产品 A, 产品 B, 返回主菜单.
2. Create a child node (another “Button Menu”) for the “Product A” button, and set the secondary menu buttons: 价格, 功能, 试用申请, 返回上一级.
3. Connect the “Price” and “Function” buttons to the corresponding reply nodes (can be text nodes or keyword nodes).
4. Make sure that each submenu node has a “Return to Previous Level” button connected to the parent node; and a “Contact Manual” button connected to the manual transfer node.

After completing all node connections, click “Publish”. Your Bot now has a complete auto-reply workflow: the user sends /start → see the welcome message and menu → click the button to enter the multi-level menu → or enter a keyword anywhere to get the auto-reply.

Best practices and considerations for automatic reply workflows

In actual operation, there are several experiences worth noting:

• Test each branch: After publishing, use a test account to go through all possible paths to ensure that each button and keyword can be triggered correctly. TG-Staff’s editor provides a “preview” mode that can simulate user interaction without publishing.
• Avoid infinite loops: Do not connect two nodes to each other to form a loop of A→B→A, otherwise the user will be trapped. Every circular path must have an “exit” exit.
• Handling Exception Input: The content entered by the user may not match any keywords or menus. Be sure to add a “Default Reply” node at the end of the process to send a guidance message, such as: “Sorry, I didn’t understand what you meant. Please reply to ‘Help’ to see the available options, or contact human customer service.” This can prevent users from being lost due to unresponsiveness.
• Regularly updated keywords: As business changes, the FAQ will be updated. Review user consultation records once a month and add new high-frequency questions to keyword nodes.

How to combine automatic reply with conversation diversion to improve customer service efficiency

Autoresponders can’t solve every problem. When the user’s problem exceeds the preset range, or the user explicitly requests to “switch to a human agent”, it needs to be smoothly switched to a human agent.

TG-Staff’s Session Offload feature comes into play here. You can set a “transfer to manual” node in the command process. When the user triggers this node, the system automatically creates a session and assigns it to the corresponding agent according to the diversion rules you configured (rotating allocation or online priority).

Recommended configuration strategy:

• Place a “Contact Human” button at each level of the auto-reply workflow (welcome, keyword reply, bottom of menu).
• Select “Online Priority” in the diversion rules to ensure that users can be taken over by the agent on duty as quickly as possible.
• Combined with Diversion Link, you can also divert users from different channels (such as advertising, social media) to Bot through exclusive links, and automatically assign different welcome words or diversion rules to achieve refined operations.

In this way, your tgbot automatic reply workflow forms a complete closed loop: automatically answer common questions → guide users to self-service operations → seamlessly transfer complex questions to manual work, which not only improves response speed but also reduces labor costs.

FAQ

**Q: Does tgbot’s automatic reply support fuzzy matching? ** Answer: TG-Staff supports both exact matching and fuzzy matching modes. Fuzzy matching can identify situations where user input contains keywords (such as “price” in “What’s your price”), improving the hit rate.

**Q: How to set up a multi-level menu in the auto-reply workflow? ** Answer: In the TG-Staff command process editor, just drag the “Button Menu” node and connect the sub-nodes. Each menu button can jump to another node, supporting unlimited levels (recommended ≤3 levels).

**Q: Can regular expressions be used for keyword matching? ** Answer: The current version does not support regular expressions, but supports multi-keyword phrases and fuzzy matching. Please pay attention to the official documentation for future updates.

**Q: After the automatic reply workflow is published, can it be modified at any time? ** Answer: Yes. After editing the process in the TG-Staff console, click “Publish” to take effect without redeploying the Bot. Sessions running during the modification are not affected.

**Q: What will the Bot do if the user input does not match any keywords? ** Answer: It is recommended to add a “default reply” node at the end of the process to send a guidance message (such as “Sorry, I didn’t understand what you meant. Please reply to ‘Help’ to see the options.”) to avoid unresponsiveness from the user.

---

If you also want to quickly build a tgbot automatic reply workflow in a visual way, you can try TG-Staff for free for 3 days: https://app.tg-staff.com/. For detailed operation instructions, see Official Document. If you have any questions during use, please directly contact the customer service Bot @tgstaff_robot.