Build Telegram Bot to manual customer service system from scratch: automatic transfer SOP and missed connection troubleshooting guide

When your Telegram community or Bot user base grows, relying solely on Bot automatic replies can no longer satisfy complex inquiries. When users encounter after-sales, technical problems or personalized needs, their most direct need is “I want to find a real person”. However, when many teams build Telegram Bot to manual systems, they often encounter problems such as agents not receiving notifications after users are transferred, multiple people grabbing orders at the same time, or users waiting for a timeout and being lost.

This article will use TG-Staff as an example to guide you step by step in building a standardized Telegram Bot customer service system from scratch, covering automatic transfer SOP, session diversion rules, diversion link configuration, and troubleshooting methods for common missed connection problems. Whether you are an overseas team, a Web3 project party, or a B2B SaaS operator, you can directly apply this process.

Why do we need a standardized set of Telegram Bot to manual SOP?

The following problems usually occur when transferring to a manual process without SOP:

• Missed call: After the user clicks “Transfer to manual”, there is no prompt on the agent side. The user thinks that no one is in charge and leaves directly.
• Duplicate reception: Multiple agents reply to the same user at the same time, resulting in confusing information.
• Response Delay: Users don’t know how long they have to wait, and agents don’t know who is in line.
• Difficulties in Attribution: From which channel (advertising, community, official website) do users come? Untraceable.

A standardized set of manual transfer SOP can solve all the above problems: it defines the complete path of the user from the bot to the agent, the agent allocation rules, and the cover-up strategy in abnormal situations. TG-Staff is designed around this scenario - through the visual process editor and flexible diversion rules, you can build a usable customer service system in 30 minutes.

Step 1: Bind Bot in the console and complete basic settings

Before starting the transfer configuration process, you need to complete account registration, Bot binding, and agent permission assignment.

Register and bind Bot

1. Visit TG-Staff Console to register an account (free trial for 3 days).
2. Create or get your Bot Token in @BotFather.
3. Enter Token on the “Bot Settings” page of the TG-Staff console and click Bind.
4. After the binding is successful, the system will automatically synchronize the basic information of the Bot (the avatar and name can be edited later).

Configure agent and project permissions

• Add Agent: Invite team members in “Agent Management”, and each agent will receive an independent login link.
• Assign project permissions: If you manage multiple Bot projects, you can set “only view a certain project” permissions for different agents to avoid information leakage.
• Confirm seat quota: The free trial period supports 1 seat; the standard version supports 3; the professional version supports 5; higher packages support 20 (see the official website package page for details).

Step 2: Build Bot → Manual automatic transfer process

This is the core step. You need to use TG-Staff’s Visual Command Process editor to build a user interaction menu and set the “to manual” trigger node.

Design welcome message and multi-step interactive menu

1. Enter the “Visual Process” editor in the console.
2. Drag a Welcome node and set the Bot’s first reply content, for example: “Hello, welcome to consult! Please select your question type: pre-sales consultation / technical support / transfer to manual customer service.”
3. Add a Button Node (Inline Keyboard or Reply Keyboard) for each option. For example:
   • “Pre-sales consultation” → Automatically reply to product introduction
   • “Technical Support” → Jump to FAQ process
   • “Transfer to manual customer service” → Connect to the agent allocation node

Configure transfer to manual nodes and diversion rules

1. In the flow chart, connect the “Transfer to manual customer service” button to the Agent Assignment Node.
2. Configure Session Diversion Rules in the project settings:
   • Rolling (default): Poll authorized agents in order, suitable for teams that want load balancing.
   • Online Priority: Prioritize allocation to currently online agents; if all agents are offline, fall back to rotational allocation.

3. Set Project Customer Service Scope: You can choose “All Customer Service” or “Specified Customer Service”. If you have a senior customer service representative on your team who only handles complex issues, you can limit the scope.

Step 3: Use diversion links to achieve advertising attribution and multi-channel traffic drainage

Many teams ignore the question “where do users come from?” TG-Staff’s Diversion Link function can help you accurately guide external traffic such as social media, advertising, and official websites to Bot, and track conversion effects.

• Generate diversion link: Create a short link on the “Diversion Link” page of the console, such as https://app.tg-staff.com/{code}.
• Set parameters: You can append custom parameters (such as utm_source=facebook) to the URL, and the system will automatically capture visitor IP and browser information.
• Configuration Jump: After the user clicks the link, he will first jump to the TG-Staff page (capture data), and then immediately redirect to your Telegram Bot.
• Attribution Analysis: Check the source channel of the user in the user portrait, and cooperate with the manual agent to form a complete traffic drainage → consultation → conversion link.

Step 4: Enable content risk control and internal control management (Professional version)

For compliance-sensitive teams such as Web3, cryptocurrency exchanges, and NFT projects, agents mistakenly sending payment addresses or sensitive information may bring huge risks. TG-Staff Professional Edition provides content risk control (internal control management) function.

• Create risk phrase: Create a new phrase in “Content Risk Control”, such as “sensitive wallet address”, add TRC20, ERC20, BTC and other addresses or address fragments.
• Associated Project: Bind the risk phrase to a specific project, and only agents under that project will be monitored.
• Set trigger action: Select “Pop-up window secondary confirmation” or “Directly block sending”.
• Audit Log: All trigger records will be saved, including agent information, session content, and trigger time, to facilitate subsequent tracing.

Troubleshooting Guide for Common Missed Connection Problems

Even if a complete manual transfer SOP is configured, users may still be unresponsive after transfer in actual operations. Here are the troubleshooting steps in order of priority:

The agent is offline or not logged in

• Check agent account status: Confirm whether the agent is online (green icon) in the console “Agent Management”.
• Confirm whether the agent is logged in to the web console: The agent must keep the web page open and logged in to receive messages. If it is inconvenient to open the browser all the time, you can pay attention to whether TG-Staff will launch mobile notifications in the future (it is not currently supported).
• Check project customer service scope: Confirm whether the agent is assigned to the current project.

Distribution rule configuration conflict

• Confirm rule effective scope: Project-level diversion rules will override global settings. Check whether “designated customer service” is mistakenly selected but does not include online agents.
• Fallback logic for online priority mode: If all agents are offline, the system will fall back to turn-based allocation. However, if the agent is still offline after the rollback, the message will be entered into the queue and will not be displayed until the agent comes online. It is recommended to set up an automatic reply to inform the user that “the agent is temporarily busy and customer service will contact you later.”

The Bot process node did not trigger the manual transfer correctly.

• Check the visual process: Open the process editor and confirm whether the “Transfer to manual customer service” button is connected to the agent assignment node and not to other automatic reply nodes.
• Test the complete path: Use another Telegram account to simulate user operations and go through the complete process from clicking the button to the agent receiving the notification.

Best practices and operational suggestions

1. Set waiting words: After the user triggers manual transfer, Bot automatically replies with a message, such as “You have entered the manual customer service queue, the current number of people in line: {queue_count}, please wait.” This can significantly reduce users’ waiting anxiety.
2. Use session tags: When the agent is receiving, tag the session (such as “pre-sales”, “after-sales”, “complaints”) to facilitate subsequent statistics and user grouping.
3. Regular audit of content risk control logs: Check risk phrase trigger records once a week to understand high-frequency scenarios where agents mistakenly send messages, and adjust keywords or strengthen training in a timely manner.
4. Use the mass sending function for return visits: For users who have ended the session, you can use TG-Staff’s batch mass sending function to send return visit messages or promotional information in groups according to tags.

FAQ

Question: Why did the user click the “Transfer to Manual” button, but no agent received the notification?

Answer: Please troubleshoot in the following order: (1) Check whether there is currently an agent online and logged in to the Web console; (2) Confirm whether the “all customer services” or “specified customer services” range in the distribution rule configuration includes online agents; (3) Check whether the “transfer to manual” node in the visualization process is correctly connected to the agent allocation node.

Question: How many agents does TG-Staff support at the same time?

Answer: The free trial period supports 1 agent; the standard package supports 3 agents; the professional version supports 5 agents; and the more advanced package supports 20 agents. Please refer to the official package page for details.

Question: What is the difference between diversion links and ordinary Bot links?

Answer: The diversion link is a short link to the official domain name of TG-Staff (such as https://app.tg-staff.com/{code}). After the user clicks and jumps to the Telegram Bot, the system will automatically capture their IP, browser information and URL parameters to facilitate advertising attribution and multi-channel tracking. Regular Bot links do not have these tracking capabilities.

Question: Which chains are supported by wallet address monitoring in content risk control?

Answer: It supports configuring addresses or address fragments of mainstream chains such as TRC20, ERC20, and BTC as risk keywords. The system will detect the outbound message before the agent sends it. If it hits the message, a pop-up window will ask for a second confirmation or directly block the sending.

Question: Is the missed connection problem related to the Bot Token expiration?

Answer: It is possible. If the Bot’s Token is reset in BotFather or the Bot is deleted, all transfer processes will be invalid. Please check whether the Bot status is normal in the “Bot Settings” of the console. If it is abnormal, you need to rebind it.

---

The above is a complete guide to building a Telegram Bot to manual customer service system from scratch. If you want to try this process right away, you can go to TG-Staff official website to register for a free trial, or start configuration directly in the console app.tg-staff.com. If you encounter any problems, you can also contact the customer service Bot @tgstaff_robot for help. If you want to learn more about the diversion rules and internal control configuration, you can check the Official Document.