Practical guide for 100,000-level TG filtering tasks: the whole process of batch splitting, deduplication, notification and export

When the overseas marketing team gets a list of hundreds of thousands or even millions of numbers, and the goal is to quickly filter out Telegram active users, identify gender or export tgid, tg filtering is no longer a simple “upload→wait→download” action. The larger the amount of data, the higher the requirements for task batch management, deduplication strategies, notification mechanisms, and result export capabilities. This article takes the KK-DATA platform as an example to detail the entire process from batch splitting to final export, helping you obtain the most accurate TG filtering results with the lowest detection cost.

What is tg filtering? Why do high-volume tasks require special planning?

tg filtering (or Telegram number detection) refers to using technical means to verify whether a batch of mobile phone numbers have been registered with Telegram and are active within a specified time window, and to extract auxiliary fields such as gender, age, tgid, etc. This operation is extremely common in cross-border e-commerce community operations, private message promotion, airdrop marketing and other scenarios - first filtering out real and active target users and then reaching them can greatly increase ROI.

For small tasks with hundreds of items, manual processing is acceptable; but when the amount of data reaches 100,000 or more, you must solve three essential problems:

1. Number Batch Management: Submit 100,000 numbers at a time. Once an error occurs in some link (such as format problems, network interruption), the cost of retrying is extremely high.
2. Waste of repeated testing: The same number is submitted multiple times, which means paying several more testing fees, which is meaningless.
3. Task Monitoring and Export: One hundred thousand detections may take several hours. You cannot keep an eye on the progress bar. Automatic notification and batch export solutions are required.

Below we break down these pain points one by one and provide practical solutions.

Common pain points of batch tg filtering: splitting, duplication and waiting

Batch splitting is difficult - how to feed 100,000 numbers to the screening system?

Many users directly stuff 100,000 numbers into a CSV file and feel uneasy after clicking submit. If the file encoding is incorrect, the column names do not match, or the number format contains spaces, the task may directly fail and have to be restarted. What’s more common is that there is no feedback from the system for a long time after submission, and you can’t tell whether it is being processed or stuck.

Misunderstanding: It is easiest to submit all the numbers at once, thinking that “one-time transaction” is the easiest thing to do.
Reasonable approach: Split 100,000 numbers into multiple small batches according to certain rules (a batch of 50,000 to 200,000 numbers is recommended), and submit each batch independently. In this way, even if an error occurs in one batch, you only need to rerun the batch without affecting the others.

Data duplication detection - money spent twice

If there is a lack of awareness of duplicate removal, the same batch of numbers may be submitted multiple times. For example, lists of numbers collected from different sources overlap, or the same document is submitted twice by mistake. Under the pay-per-item model, repeated inspections directly become a waste of real money. According to actual customer feedback, the inspection fee may cost 30% to 50% more without deduplication.

I don’t know when it will be completed and how to export it.

One hundred thousand level tasks often take several hours to complete, and manually refreshing the console page is extremely inefficient. In addition, many novices are confused about how to interpret the result export fields (tg activation, active, gender, tgid) and how to cooperate with subsequent use.

How to scientifically split the 100,000-level TG filtering task? (batch strategy)

Split by number source or number segment

If your numbers come from multiple channels (such as different country number segments and different CSV files exported from the global number generation module), it is recommended to keep the channels independent and create a tg filtering task for each. The advantage of this is: when the activity rate in a certain batch of results is abnormally low, you can quickly locate the data quality problem of which channel, instead of rummaging through a pot of porridge.

Example of file naming convention: tg_filter_US_20250301.csv, tg_filter_BR_20250301.csv to facilitate follow-up.

Split by detection type (opened vs active vs gender)

Unit prices may vary for different test types and may have different business objectives. If you only need to compare the activation rate, just select “Activation Test”; if you also want to screen male users who have been active in the past 7 days, you can do it in two steps:

1. First do the “tg activation” test to filter out unregistered numbers;
2. Resubmit the number whose test result is “activated” for “tg active + gender test”.

Although this “split into multiple steps” method requires one more operation, it can avoid the activity detection of unregistered numbers (the unit price of the latter is generally higher than the activation detection), and is overall more cost-effective.

Split by time window or notes

When the requirements of active windows are different, it is recommended to handle them in separate tasks. For example, create a task for the number based on “Active in the past 7 days” and “Active in the past 30 days”, because the active window parameters will affect the detection time and cost. In addition, if the number has business labels (such as “Customer A”, “Customer B”), the labels should be kept independent and can be used directly after exporting without secondary classification.

How to avoid duplicate tg filtering and deduction for data deduplication?

How the deduplication warehouse works

KK-DATA has built-in data deduplication warehouse function. When you upload a batch of numbers, the system will automatically compare it with the numbers in all historical detection tasks. Numbers that have been detected will be skipped, and only undetected numbers will be filtered by TG. This means that even if you upload data that is duplicated from a previous task by mistake, you will not be charged again.

Best practice for deduplication: merge first and then submit

Although the platform has a deduplication mechanism, it is recommended that users do it themselves before submitting. Use Excel or a text deduplication tool to merge the same numbers in multiple files into one file. This can reduce the amount of comparison calculations when deduplicating the system, and avoid overly large task files due to too many duplicate numbers. Tip: When removing duplicates, pay attention to retaining the international format of the number (such as +8613812345678) to avoid the same number being regarded as two due to different formats.

How to view the details of money spent and saved?

In the “Recharge Flow” and “Task Consumption Report” of the console, you can see the actual deduction amount, the number of detections, and the number of numbers skipped by deduplication for each TG filtering task. Reviewing this data regularly will allow you to visually evaluate the cost savings resulting from deduplication.

Task notification and monitoring: When will you know that tg filtering is completed?

You don’t need to stare at the console all the time. After submitting the tg filtering task, you can bind Telegram notifications in the task settings. When the task is completed (or partially completed), the system will automatically send a message to your Telegram, including the task ID, a summary of the detection results (such as the number of activations and activities), and the export link.

If you need to manage multiple batch tasks at the same time, you can view the real-time status of all tasks (queued, in progress, completed, failed) on the “Task List” page of the console. It is recommended to keep records in the order of submission, such as adding serial numbers when naming: batch_01_US, batch_02_US to facilitate quick location.

Export suggestions: interpretation and subsequent use of result fields

After tg filtering is completed, the exported CSV file usually contains the following key fields:

Field name	Meaning	Usage examples
phone	Original number	For subsequent contact
tg_open	Whether to register for Telegram (activate)	Filter out unregistered numbers
tg_active	Whether active within the specified window	Filter active users in the past 30 days
tg_sex	Gender (Male/Female/Unknown)	Target male or female
tg_age	Age (to be filled in with user profile, for reference only)	Screening for people around 30 years old
tgid	Telegram User ID	Used for API group import, etc.

Practical steps:

1. Download the results CSV file.
2. Use Excel or Python to filter out rows with tg_open=是 and tg_active=是 and tg_sex=男.
3. Use these numbers for private message promotion or group invitations to ensure that the targets you reach are real active users.

If you need a more detailed portrait (such as avatar, recent online time, etc.), you can combine it with additional fields in gender detection. The specific fields are subject to the console export results.

FAQ

**Q: What fields can tg filter detect? ** Answer: It supports detection of whether the number is registered with Telegram (activated), active within the specified time window, gender identification (including age, avatar and other fields, age is for reference only and not accurate at the ID level), and exporting tgid. Please refer to the console export results for specific fields. Different detection packages may differ.

**Q: How long does it take to filter 100,000 numbers? ** A: The processing time depends on the number of numbers, platform detection queue load and detection type selected. Usually 100,000 items can be completed within a few hours in the ultra-fast filtering mode. It is recommended to set aside half a day to a day and turn on Telegram notifications to avoid paying attention all the time.

**Q: Can multiple tg filtering tasks be submitted at the same time? ** Answer: Yes, the system supports submitting multiple tasks in parallel. Each task runs independently and does not affect each other. However, before submitting, please ensure that the account balance is sufficient to cover the estimated costs of all tasks. Tasks with insufficient balance will not be submitted.

**Q: If you find that a certain batch number is detected incorrectly after the batch is split, can you try again? ** Answer: Completed tasks cannot be modified, but you can resubmit the batch number. Note: The deduplication warehouse will be recognized as detected, and duplicate numbers will be skipped unless you manually clear the history or switch to a different detection type (such as switching from “Active Detection” to “Active Detection”). It is recommended to do a trial run of 100 to 500 items in a small batch, and then submit the full quantity after confirming the effect.

**Q: Is the age field in the tg filter results accurate? ** Answer: The age field comes from personal information actively filled in by Telegram users. It is unofficial authentication information and is for reference only. It is suitable for crowd portrait analysis (such as screening users around 30 years old), but cannot be used in scenarios that require precise age verification.

---

After reasonable batch splitting, deduplication, notification and export planning, the 100,000-level TG filtering task can be completed efficiently and at low cost. Log in to the console immediately to start your batch Telegram screening, or contact customer service in both directions for one-on-one guidance.

👉Log in to the console to start screening numbers Two-way contact customer service: https://t.me/kkdata_robot Learn more: Official website https://kkdata.cc/ · Documentation https://docs.kkdata.cc/