TG Diversion Link A/B Testing Guide: Optimize Telegram Customer Support Conversion with Different Codes and Welcome Messages

If you’re using a Telegram Bot to handle customer inquiries, you’ve likely faced this problem: How many users who click your link actually start a conversation? Do ad-driven leads drop off immediately, or do they get guided to a human agent by your welcome message? The answer often depends on how you design that “first impression”—the first message your Bot sends after the user clicks a diversion link.

This article uses TG-Staff’s diversion link feature as the core to walk you through a complete TG diversion link A/B test: from creating links with different codes, to designing welcome message variables, to analyzing data and selecting the winning version. Whether you’re a SaaS operator, cross-border customer support manager, or Web3 project owner, this method helps you replace guesswork with data and improve your Telegram customer support conversion rate.

What Is TG Diversion Link A/B Testing and Why Is It Important?

A TG diversion link is an official domain short link provided by TG-Staff (like https://app.tg-staff.com/{code}). When a user clicks it, their IP, browser info, and URL parameters are captured first, then they are redirected to your Telegram Bot, which automatically sends a welcome message and, if needed, transfers them to a human agent.

The core idea of A/B testing is simple: Create multiple diversion links with different codes, pair them with different welcome messages or Bot flows, and compare which combination yields higher user engagement or agent handoff rates.

Why is this crucial for Telegram customer support operations?

• Precise attribution: Each code corresponds to a channel or campaign, so you know exactly where traffic comes from and how it performs.
• Reduced trial and error: Instead of changing welcome messages based on gut feeling, use data to verify which version works better.
• Optimized traffic funnel: From ad click → Bot welcome → human agent, each step’s conversion rate becomes measurable.

Core Elements of A/B Testing: Code, Welcome Message, and User Behavior

A standard A/B test involves three variables:

Element	Description	Example
Code	Unique identifier for the diversion link, used to distinguish different channels or versions	?code=ad1, ?code=ad2
Welcome message	The first message sent by the Bot after the user clicks the link	Concise version vs. detailed version
User behavior	Key actions tracked	Click rate, conversation initiation rate, agent handoff rate

The code is the “ID card,” the welcome message is the “storefront,” and user behavior is the “report card.” All three are indispensable.

Suitable Scenarios

A/B testing of TG diversion links isn’t a silver bullet, but it works well in the following scenarios:

• Ad campaigns: Use different codes for different channels (Google Ads, Facebook, Twitter) to compare which channel’s users are more willing to interact with the Bot.
• Promotions: Pair limited-time offers with tailored welcome messages (e.g., “Enter promo code for 10% off”) and test different copy for conversion rate.
• Multilingual testing: Compare Chinese vs. English welcome messages to find the language style preferred by target users.
• Customer support routing optimization: Different codes correspond to different agent groups (pre-sales vs. post-sales) to test which grouping improves response time and satisfaction.

Preparation: Configure Diversion Links and Bot in TG-Staff

Before starting the test, ensure you have completed the following prerequisites:

1. Log in to TG-Staff Console (https://app.tg-staff.com/），注册后享有 3-day free trial.
2. Create or select a Bot project: Add your Telegram Bot in “Project Management” (you need a Bot Token, obtainable via BotFather).
3. Confirm your plan: The diversion link feature is available in Standard plan and above. It works during the free trial as well.
4. Prepare multiple codes: Use short, unambiguous English or numeric combinations, such as ad1, ad2, promo_a, promo_b. Avoid Chinese characters or special symbols, as they may affect URL parsing.
5. Configure basic welcome message: In TG-Staff’s “Visual Command Flow,” create at least one start node to define the first message the Bot sends after the user clicks the link.

Step 1: Create Multiple Splitting Links with Different Codes

TG-Staff’s splitting link generation entry is located in the “Splitting Links” section of the console. Follow these steps:

1. Go to the “Splitting Links” page and click “Create New Link.”
2. Fill in the following information in the popup:
   • Name: A label for easy identification, e.g., “Ad A Version.”
   • code: A unique identifier, e.g., ad1. The system will automatically generate the full link: https://app.tg-staff.com/ad1.
   • Associated Project: Select the Bot project you want to test.
   • Associated Flow (optional): If you want different codes to correspond to different Bot interaction flows, specify here. Leave blank to use the Bot’s default flow.
3. Repeat the above steps to create a second link with the code set to ad2.

Once created, you will see two links and their corresponding codes in the list. Be sure to record which code corresponds to which channel or variable, otherwise later analysis will be chaotic.

Step 2: Design Welcome Message A/B Variables

The welcome message is the first message users see after clicking the link, and it directly determines whether they are willing to continue interacting. When designing variables, it is recommended to change only one dimension for proper attribution.

Variable 1: Welcome Message Style Comparison

• A Version (Concise Prompt): “Hello! Welcome, how can I help you?” — Open-ended question encouraging user input.
• B Version (Menu Prompt): “Click the menu below to quickly learn about services → 1. Product Introduction 2. Price Inquiry 3. Customer Service” — Provides clear options to reduce user thinking cost.

In TG-Staff’s visual command flow, you can bind different starting nodes for each code. For example, bind ad1 to a node with only text messages, and ad2 to a node with menu buttons.

Variable 2: Guidance Path Comparison

• A Version (Keyword Trigger): After the welcome message, prompt “Reply ‘help’ for assistance,” leading users into a self-service flow.
• B Version (Human First): After the welcome message, immediately prompt “For customer service, reply ‘agent’,” guiding users to quickly transfer to a live agent.

This comparison is suitable for testing user preference between self-service and human intervention. If your team has limited agents, you may want users to self-service first; if conversion is more important, you may want users to talk to a real person quickly.

Step 3: Deploy Links and Start Testing

After creating the links, deploy them to the corresponding channels:

• Place the ad1 link on the Google Ads landing page button.
• Place the ad2 link in the pinned comment of a Facebook post.
• If using email marketing, randomly assign the two links within one email group (or send in two separate batches).

Testing Period Recommendation: Run for at least 3 days to cover weekday and weekend user behavior differences. If traffic is low, extend to 7–14 days to ensure at least 100 clicks per variable.

Tips for avoiding confusion: Besides the code itself, you can also append UTM parameters to the URL (e.g., ?code=ad1&utm_source=google). Although split links automatically capture basic information, UTM parameters allow for more detailed attribution in Google Analytics.

Step 4: Analyze data and select the winning version

After the test ends, view the click data for each code on the “Split Links” page in the TG-Staff console. You’ll see statistics similar to the following:

code	Clicks	Conversations Started	Conversation Start Rate	Agent Transfers
ad1	150	45	30%	20
ad2	160	72	45%	38

In this example, ad2 has a significantly higher conversation start rate (45%) compared to ad1 (30%), indicating that version B’s welcome message (menu guidance) is more effective. Agent transfers also increased notably.

But don’t just look at click-through rates. You should also pay attention to the following metrics:

• Conversation retention rate: Did users interact again within 24 hours after starting the conversation?
• Agent handover rate: The proportion of conversations that transition to a human agent.
• Final conversion: Did users complete target actions like purchases or registrations? (Requires backend data integration.)

If the difference in metrics between two versions is small (e.g., conversation open rate differs by only 3%), use an online significance calculator (such as Optimizely’s A/B test calculator) to determine the p-value. When p < 0.05, the difference is generally considered significant. If the difference is not significant, it is recommended to extend the test period or increase the sample size.

Best Practices: 5 Tips to Optimize A/B Testing

1. Change only one variable at a time: Do not change both the welcome message and the code simultaneously. If you want to test the welcome message, keep the code naming logic consistent (e.g., ad1 and ad2 both come from the same ad channel); if you want to test the channel, keep the welcome message the same.
2. Ensure even traffic distribution: TG-Staff does not provide automatic traffic distribution; you need to manually distribute links with different codes to different channels or time periods. If using the same channel, consider using random distribution tools (such as third-party A/B testing platforms) or manual rotation.
3. Record test hypotheses and results: Before starting a test, write down your hypothesis (e.g., “Menu-guided welcome message can increase conversation open rate by 20%”). After the test, compare actual data and record the results for future reference, regardless of the outcome.
4. Focus on long-term metrics, not just click-through rate: A high click-through rate does not necessarily mean good conversion. Some welcome messages may attract clicks, but complex follow-up processes can lead to user churn. It is recommended to track user retention or paid conversion within 7 days.
5. Combine with TG-Staff’s conversation routing rules: Assign different agent groups to different codes. For example, ad1 points to the pre-sales group, ad2 points to the after-sales group, to test the impact of different agent groups on conversion rate.

Frequently Asked Questions

Q: What plan is required for A/B testing of TG diversion links? A: TG-Staff’s Standard plan and above support diversion link functionality, allowing you to create multiple codes for testing. The free trial (3 days) also allows you to experience this feature. The Professional plan also supports more detailed user profiles and data statistics, which is helpful for in-depth analysis.

Q: How to ensure even traffic distribution between two diversion links? A: TG-Staff does not provide automatic traffic distribution; you need to manually distribute links with different codes to different channels or time periods. It is recommended to use random distribution tools (such as third-party A/B testing platforms) or manual rotation. If traffic comes from the same channel, consider alternating the release of the two links within a day.

Q: What sample size is needed for welcome message A/B testing? A: It is recommended to obtain at least 100 clicks per variable before analyzing data; the larger the sample size, the more reliable the results. If traffic is low, you can extend the test period or merge low-traffic variables. For high-value conversion scenarios (such as paid subscriptions), it is recommended to have a sample size of 500 or more.

Q: How to determine significance in A/B test results? A: Use online significance calculators (such as Optimizely or AB Test Guide) to compare click-through rates or conversion rates. When the p-value < 0.05, the difference is generally considered significant. If the difference is small, continue testing or adjust variables. Note: Even if statistically significant, consider practical business significance (e.g., whether a 1% improvement is worth the resource investment).

Q: After the test ends, how to apply the winning version? A: In TG-Staff, set the diversion link corresponding to the winning code as the default link, and update the Bot’s welcome message to the winning version. If using multiple channels, unify all links to point to the winning code. Remember to delete or disable failed links in the console to avoid future misuse.

Summary and Next Steps

A/B testing of TG diversion links is not a one-time task but a continuous optimization cycle. From creating codes to designing welcome messages, from deploying links to analyzing data, each step helps you better understand users and use Telegram customer service resources more efficiently.

If you haven’t started yet, now is the best time. Sign up for a free trial of TG-Staff (https://app.tg-staff.com/），创建你的第一个分流链接，选择两个不同的欢迎语版本，投放出去，然后看看数据告诉你什么。

If you have any questions during testing, feel free to contact customer service Bot @tgstaff_robot, or refer to the official documentation (https://docs.tg-staff.com/）了解分流链接的详细配置方法。

Remember: Don’t guess, test. Your next best welcome message may be in the results of this A/B test.