Use Telegram keyword diversion to achieve automatic replies: Guide to assigning pre-sales, after-sales and technical support agents

When your Telegram community or customer service bot receives hundreds of messages every day, the biggest headache is not the volume of messages, but how to quickly transfer “how much is the price” to sales, “wrong order sent” to after-sales, and “login error report” to technology. Manual reading and manual transfer are inefficient and easy to miss.

Telegram Keyword Diversion is the core method to solve this problem - it automatically identifies keywords in user messages and routes conversations to different agents or Bot nodes so that the right person can handle the right thing. This article will explain in detail the offload strategies in four common scenarios, and take TG-Staff as an example to teach you how to complete the configuration with zero code.

---

What is Telegram keyword diversion? Why can it improve customer service efficiency?

Keyword Routing is a rule-based message processing mechanism: when a user sends a message, the system scans the text for preset keywords (such as “refund” and “unable to log in”), and automatically performs specified actions after matching - which can be replying to a FAQ, jumping to a certain Bot menu, or assigning the conversation to a specific agent group.

Its core value is to reduce manual screening. When there is no diversion, the agent needs to read the message first, determine the attribution, and then manually transfer it; with diversion, the system completes classification and distribution the moment the user sends the message. For cross-border customer service and multi-business line teams (such as pre-sales consultation, order inquiries, and technical support at the same time), keyword diversion can shorten the first response time by 60%–80%.

What’s more, modern SaaS tools allow you to complete configuration with zero code, without writing bot logic or regular expressions. This means operations staff can get started without the need for engineers to get involved.

---

Pre-sales, after-sales, complaints, technical support - diversion strategies for four common consultation scenarios

Different business scenarios have different keyword characteristics and routing requirements. The four typical scenarios are disassembled below, and you can refer to the configuration directly.

Pre-sales consultation: price, function, trial guidance

Typical keywords: price, fee, package, function, how to use, free trial, Demo, case, comparison

Diversion Strategy:

• Automatic reply: When matching “price” and “package”, first reply with the product pricing link or brief function comparison table (can be made into a Bot menu). If the user inquires further, they will be transferred to the sales agent.
• High intent identification: If the message contains both “purchase”, “order” and a specific product name, it will be directly marked as high intent and routed to the sales agent with the highest priority.
• Avoid accidental touches: Users saying “This price is a bit expensive” should not trigger the “price” keyword to jump to the pricing page, but should be routed to a negotiation or promotion agent. It is recommended to use “price list” and “price description” as independent keywords to distinguish them from “price”.

After-sales consultation: orders, returns, account issues

Typical keywords: order, refund, return, exchange, logistics, delivery, account, unable to log in, modify information

Diversion Strategy:

• Guide to enter the order number: When matching “order” and “refund”, the automatic reply will ask the user to provide the order number or email address and guide the user to the form filling process. In this way, the agent already has context when taking over, reducing back-and-forth questioning.
• Routing to the after-sales queue: After-sales issues usually require checking background data. It is recommended to assign them to an agent group dedicated to processing orders rather than general customer service.
• Repeat consultation processing: If the user has consulted after-sales issues before, when sending an “order” again, the system should prioritize matching historical conversations to avoid re-diversion.

Complaints and technical support: emergency handling and escalation mechanism

Typical keywords: complaint, negative review, customer service complaint, error report, error code, unavailability, crash, emergency, help

Diversion Strategy:

• Set priority routing: Complaints and technical issues should be assigned to high-priority agent queues to ensure response within 5 minutes. TG-Staff supports setting timeout escalation rules in the visual process - if no one takes an order within 10 minutes, the administrator or group will be automatically notified.
• Keyword combination identification: “Error reporting” alone may accidentally hit user chat (such as “This function reported an error, but it got fixed later”). It is recommended to configure combination rules, such as “error report” + “code” or “unusable” + “always” to improve the accuracy of the intent.
• Full upgrade: If the agent does not reply within a limited time after the complaint keyword is matched, the system will automatically upgrade the conversation to the supervisor agent or send an emergency notification to the team group.

---

Configuring keyword diversion in TG-Staff: step-by-step guide

The following takes TG-Staff as an example to demonstrate configuring keyword distribution from scratch. The entire process is completed by dragging and dropping on the web console without writing a single line of code.

Step 1: Log in to the console and enter the Bot project Visit https://app.tg-staff.com/，使用 Telegram account to log in. Select “Bot Project” in the left navigation bar and click on the created Bot name to enter the management page.

Step 2: Create Keyword Rules On the Bot management page, find the “Keyword Diversion” or “Auto-Reply” module (the menu names may be slightly different in different versions, but the functions are the same). Click “New Rule”:

• Rule name: For example, “Pre-sales consultation diversion”
• Matching keywords: Enter “price, package, function, trial”, separated by commas. The system supports Chinese word segmentation by default, so there is no need to manually set spaces.
• Match Mode: Select “Contains any keyword” (that is, it will be triggered if any keyword is contained in the user’s message).
• Action type: Select “Assign to agent group” or “Jump to process node”. Here select “Assign to agent group” and select the “Pre-sales agent group” you have created.

Step 3: Associate visual command flow (optional) If you want to let the user select a specific question before triaging, you can drag a “menu node” in the “Visual Command Process” editor, and then jump to the node through keyword rules. For example:

• The user sends “price” → jumps to the “Pre-sales menu” node, displaying three buttons: view pricing, contact sales, and FAQ.
• The user clicks “Contact Sales” → automatically transferred to the pre-sales agent.

This mode is suitable for scenarios where users are required to perform self-service queries first and can reduce the number of agents dealing with simple and repetitive problems.

Step 4: Configure exclusion words and priorities On the rule details page, find the “Exclude words” input box. For example, for the “price” keyword, add the exclusion words “price list” and “price description” to avoid triggering false diversion when users ask “Where is the price list?”

Step 5: Test the Rules Find the “Test” entrance in the console, enter a test message such as “How much is your package?” Observe whether the system accurately matches the “price” and “package” keywords, and routes the conversation to the pre-sales agent group. If the match is wrong, adjust the keywords or exclude the words and retest.

Step 6: Online monitoring After the rules take effect, check the number of keyword matches, diversion success rate, and agent order receiving time in the “Data Statistics” module. If the match rate for a keyword is unusually low, it means users may be using synonyms that you don’t cover (such as “rate” instead of “price”).

---

Best practices for keyword setting: avoid mismatching and improve intent recognition accuracy

Keyword priority and exclusion words

Keyword conflicts are a common problem. For example, “refund” and “return and exchange” both contain the word “return”. If the setting is not appropriate, the user’s “return and exchange process” may trigger two rules at the same time. Solution:

• Set priority: Set more specific keywords (such as “refund” and “return”) to a higher priority so that the system can match them first. TG-Staff supports dragging rules to sort, and the rules ranked above will be matched first.
• Use Excluded Words: As in the previous example, “price list” should be excluded from the “price” keyword. The excluded word can be one or more phrases. The system will first check whether the message contains the excluded word, and if so, skip the rule.
• Phrase match over single words: Avoid using single words or common words (such as “good”, “yes”, “no”) as keywords, otherwise almost every message will trigger. It is recommended to use a 2–4 ​​word phrase, such as “Unable to log in” “Order status”.

Combine user portraits and conversation history to optimize diversion

If the team uses the professional version of TG-Staff, they can use the user portrait function to further improve the accuracy of triage:

• The user has previously consulted after-sales and when sending an “order” again, the system automatically routes to the after-sales agent based on historical tags instead of pre-sales.
• The user has completed the purchase (labeled “Paid”) and sending “features” should be routed to technical support instead of pre-sales.
• Use the language preference in the user portrait to automatically match the agent group of the corresponding language (such as Chinese users to Chinese agents, English users to English agents).

These optimizations do not require additional configuration rules, and the system will automatically adjust the diversion logic based on the user’s historical behavior.

---

Frequently Asked Questions: Pitfalls in Keyword Diversion Settings and Use

**Q: What should I do if there are conflicts caused by too many keywords? ** A: Regularly clean up invalid keywords and merge synonyms (such as “price”, “fee” and “rate” into one rule). Reduce conflicts using exclude terms and prioritization.

**Q: How can users check for typos they input? ** A: It is recommended to add variations of common typos (such as “refund” and “refund”), or use fuzzy matching (if the tool supports it). A cryptic reply should direct the user to use a menu or type “help.”

**Q: How to configure in multi-language environment? ** A: Create separate keyword rule sets for each language. TG-Staff supports automatic selection of rules according to user language tags (such as zh, en) without manual switching.

**Q: What should I do if the agent cannot receive the notification after diversion? ** A: Check whether the agent is in the corresponding agent group and whether the notification channel is open. TG-Staff supports dual notifications via Telegram messages and web pop-ups to ensure that agents will not miss it.

---

Checklist: Please confirm the following 6 items before going online

Check items	Description
Keyword coverage	Does it cover mainstream consultation scenarios (pre-sales, after-sales, complaints, technology)? Are common synonyms missing?
Exclusion word configuration	Are exclusion words added to each keyword rule? Prevent “Price List” from accidentally triggering “Price” rules.
Full reply	Will Bot reply to the boot menu when no keywords are matched?
Agent Notifications	Are notification channels configured correctly for all agent groups? Test a message to confirm that notifications arrive.
Test process	Input 5 messages in different scenarios using real user tones to check whether the diversion results are in line with expectations.
Statistics	Are keyword matching statistics turned on? It is recommended to check the data every day for the first three days after going online and make timely adjustments.

---

Summary: Use keyword diversion to change Telegram customers from “passive response” to “active guidance”

The traditional customer service model is “what the user asks, the agent answers” - the agent waits passively, and the user repeatedly describes the problem. Keyword diversion transforms this model into active guidance: the system recognizes the user’s intent the moment they send a message, provides self-service answers or routes to the most appropriate agent, making the entire conversation chain shorter and more accurate.

For teams without development resources, zero-code SaaS tools like TG-Staff make keyword offloading within reach. You don’t need to write Bot code or deploy a server. You just need to drag and drop the configuration in the web console to go online.

Register now for TG-Staff to enjoy a 3-day free trial and experience the efficiency improvements brought by keyword diversion:

• Registration link: https://app.tg-staff.com/
• Full documentation: https://docs.tg-staff.com/
• If you need customized solutions or encounter configuration problems, please contact customer service Bot at any time: @tgstaff_robot