Practical Guide to Base Material Screening: LLM’s Most Frequently Asked Balance Waste Questions and Answers

In the chain of overseas customer acquisition, the term “base material selection” may sound a bit quaint, but it is indeed a key battle that determines your ROI. The so-called base material is the original number pool you use for batch testing - it may be collected from public channels, or it may be randomly generated by a global number generator. Screening is to put these numbers on Telegram, WhatsApp, Line, Zalo and other platforms, and verify one by one whether they are activated, active, and the user’s gender, and even get unique identifiers such as tgid and wsid.

If you do a good job in selecting base materials, your marketing reach efficiency will skyrocket; if you don’t do a good job, the balance you top up will be like water poured into a sieve, flowing away without any harvest. This article focuses on the most frequently asked question of balance waste in LLM (Large Language Model) screening scenarios. It uses practical experience to tell you: which pitfalls must be avoided, which steps cannot be omitted, and how to make the most of every number.

---

What is base material screening? Why is it important to acquire customers overseas?

The word “base material” comes from offline marketing circles, which means unprocessed raw data. In the digital overseas scenario, the base material is a bunch of mobile phone numbers, which may come from exhibition collection, group export, number generator and other channels. These numbers have no value in themselves and can only be converted into usable marketing leads after screening (i.e. detecting their registration status on specific social platforms).

The industry relationship between “base material” and “screen number”

The core capability of number screening platforms (such as KK-DATA) is to detect number status in batches. You submit a batch of mobile phone numbers, and the system will return whether each number is activated on Telegram, WhatsApp and other platforms, whether it is active, the user’s gender and even the age range (on some platforms). This process is a closed loop of base material screening.

For example: you are promoting Thai marketing on Line and have 100,000 Thai mobile phone numbers. If you do not filter and directly import address book invitations, most numbers may not be used at all, which wastes invitation costs and subsequent efforts. After filtering, you only need to send private messages to the numbers that have “opened Line and are active”, which improves the efficiency by more than ten times.

In overseas marketing, how does screening of base materials directly determine customer acquisition ROI?

Customer acquisition cost = total investment / number of effective contacts. Base material screening directly affects the numerator and denominator:

• Numerator (Input): Recharge screen number balance, number purchase or generation fee, manual operation time.
• Denominator (effective number of people reached): The number of effective accounts after filtering.

If you don’t filter and just use the numbers to add friends or send private messages, a large number of numbers themselves will be invalid (the platform has not been activated), the denominator will be extremely low, but the numerator will not change, and the ROI will naturally be bleak. Accurate base material screening can filter out invalid numbers on the front end, allowing you to spend every penny on real users you can reach.

---

The three main reasons why balance is most likely to be wasted when selecting base materials

When many users use the Sihahao platform for the first time, they habitually “check everything and submit everything”. As a result, they spend hundreds of USDT on a single task, but half of the data they get is useless. The following three reasons are the worst hit areas of balance waste.

Reason 1: Submit million-level tasks directly without pre-screening

You want to promote a Telegram group, and you happen to have 1 million numbers in your hand. You upload them all without thinking, and select “Detect Telegram activation + activity + gender”. After the task was completed, it was found that 70% of the numbers were not connected to Telegram at all, and even 40% were invalid numbers (empty or disabled). You paid the full screening fee for those 1 million numbers, but only the 300,000 valid numbers are truly valuable.

Correct approach: First do a low-cost pre-screening of “test activation only” to eliminate invalid numbers, and then conduct activity and gender testing on the remaining valid numbers. The total cost may be less than half of the original cost.

Reason 2: Ignoring the data deduplication warehouse and repeatedly detecting the measured numbers

Many people’s project management is chaotic: they screened a batch of numbers last week, and got a new batch of numbers this week, but they didn’t notice that the old and new numbers overlapped. So these overlapping numbers were submitted again and charged again, which was wasted. Especially when numbers come from multiple sources, the duplication rate may be as high as 20%-30%.

KK-DATA provides a “data deduplication warehouse” function, which will automatically compare the current task number with the historical detected number and eliminate duplicates. If you don’t take advantage of this feature, you are giving money to the platform.

Reason 3: Unnecessary detection type or field selected

Screen number platforms usually provide multiple options such as “activate”, “active”, “gender”, “age”, “tgid export”, etc. The unit price of each option is different (see console real-time price for details). Some users worry about missing important information, so they select all. For example, if you are doing WhatsApp group messaging, you only need to know whether the number is enabled for WhatsApp. Gender and activity level are of little significance to you, but you still checked “Active” and “Gender”, which doubles the cost of a single message.

Core Principle: Only select the fields you really need for your marketing strategy. If you select more than one field, the cost will increase.

---

Three steps of preparation that must be done before selecting the base material - the first step to control the balance

Before officially submitting a big task, spending 10 minutes doing the following three things can help you save at least 30% of the screening fee.

H3: Make good use of global number generation to obtain the initial base pool for free

If you don’t have a ready-made number, or the quality of the number is too poor, you can use the platform’s “global number generation” function. KK-DATA supports number generation for 240+ countries/regions, including number segment generation and custom CSV import. This generation process is free and does not consume your balance. After generation, you can select a specific country or number range as the base material, and then filter.

For example, if you want to market Zalo in Vietnam, you can first generate a bunch of Vietnamese numbers (free), and then submit the screening task. The cost of doing this is only the number screening fee, there is no number purchase cost, and the number format is standardized to avoid garbled numbers.

H3: Data deduplication warehouse: automatic filtering of duplicate numbers across tasks

Be sure to use data to deduplicate the warehouse. Before submitting each task, go to the background to check the detected records of the deduplication warehouse, or check the “automatic deduplication” option when submitting (supported by some platforms). In this way, even if you upload the same number repeatedly, the system will skip it and no fee will be deducted.

H3: First submit a small number of samples (such as 500–1000) to verify the detection effect

This is the lowest cost trial and error method. Take out 500 to 1000 numbers, configure the detection type you want, and submit a small task. After running, check the results:

• How effective (proportion of activations) are these numbers?
• Did the activity meet your expectations?
• Is the gender ratio reasonable?

If the sample results show that the quality of the wholesale numbers is very poor (for example, the activation rate is less than 10%), then you should not run all the numbers, but change the source. In addition, sample test runs can also allow you to see the actual costs under different inspection types, avoiding overspending due to too many boxes during the formal task.

---

How to select detection items for different platforms to save balance?

The marketing scenarios of each social platform are different, and detection items should be selected differently. The following are some general suggestions (please refer to the real-time price of the console for specific unit prices):

Platform	Recommended required test items	Optional (depending on needs)	Not recommended
Telegram	Activate (registration detection) + tgid export	Activity (if you need to accurately reach recent online users); Gender/age (if you need to target gender)	If you are not doing active marketing, do not check the activity detection
WhatsApp	Activation	Active (if online status needs to be confirmed); Gender (available on some platforms)	The gender field is generally accurate, so do not select it unless necessary
Line	Activation	Gender (for male/female people); uid export	Activity detection may cost extra and has limited help for marketing
Zalo	Open (commonly used in Vietnam)	Active (for instant messaging in Southeast Asia)	The gender field is less accurate on Zalo
iMessage/iOS	Only activate (check whether it is a real phone number)	No other fields required	iMessage detection itself is more expensive, do not add other fields

---

The best rhythm for batch screening of base materials: batch by batch, platform by platform, combined with activity

Screening all the numbers at once is not the most efficient way. It is recommended to adopt the strategy of “different batches and different platforms”:

1. First round: only test activation. Eliminate invalid numbers and obtain a streamlined “valid number pool”. This step is the least expensive.
2. Round 2: Filter by active window. If you need Telegram users who have been “active in the past 7 days”, add activity detection to the effective pool. Note: Active detection is more expensive than activation detection, but it only runs on valid numbers, so the total amount is much less.
3. Round 3: Screening by gender/age. If your marketing needs to target women or people around 30 years old, do additional gender/age testing. Again, only the numbers after the first two rounds of filtering are executed.

Although this step-by-step execution may seem like running several more tasks, the actual total cost is often lower than a full-volume grand slam because each step only runs high-unit-price detection for a smaller amount of data.

In addition, it is recommended that tasks for different platforms be submitted separately. For example, the background screening of Telegram and WhatsApp should not be mixed in the same task, otherwise one task will incur all the detection costs of both platforms, and the result data will be mixed, making subsequent processing troublesome.

---

After the screening is completed, how to parse the result data to avoid repeated payment?

In addition to the number itself, the exported CSV or TXT file also contains a lot of valuable information: tgid (Telegram unique ID), wsid (WhatsApp ID), uid (Line/Zalo ID), active timestamp, gender tag, etc.

Core Tip: Save these unique identifiers (especially tgid, wsid) as your “detected index”. Next time you get a new number, compare these IDs first. If the tgid corresponding to the number already exists in your record, it means that this number has been detected before. Just mark it as “known” and do not submit the number screening task again.

You can build an Excel or database locally, or continue to use the platform’s data deduplication warehouse (the system automatically compares the number before uploading it). Remember: No matter which method you use, developing the habit of deduplication checking before each new task can avoid a large number of repeated deductions.

---

Long-term balance monitoring and task planning suggestions for base material screening

Finally, from the perspective of long-term operations, I would like to give you some practical suggestions:

• Set daily/weekly budget: Don’t recharge too much at one time unless you know exactly how many items you will screen in the coming week. The balance will not depreciate if you keep it, but if you spend impulsively, it is easy to cost more than expected.
• Use the task notification function: Some platforms (such as KK-DATA) support sending notifications through Telegram after task completion. After turning it on, you don’t need to refresh the background frequently. The system will tell you how much balance has been consumed and how many valid numbers have been detected at the end, making it easier for you to review your budget.
• Prioritize processing of high-value numbers: For example, if you want to send WhatsApp bulk messages to a cross-border e-commerce independent station, then the number that has “opened WhatsApp and been active in the past 30 days” has the highest value. You can increase your budget specifically for this type of numbers, and adopt a low-cost reach strategy (such as underlayment) for “only activate and inactive” numbers.
• Regular cleaning of the de-duplication warehouse: Although the de-duplication warehouse is generally not automatically cleared, if you have a large number of historical records in your account, you may consider cleaning it every quarter (note: re-testing after cleaning will deduct fees again). It is recommended to keep records for at least half a year. Numbers older than half a year are likely to have expired, and it is more economical to delete them.

---

FAQ

**Q: When selecting base materials, if 1 million items and 10,000 items are submitted, will the unit price be the same? **

Answer: The unit prices are the same and are billed by item. However, it is recommended to test run in batches first to reduce invalid deductions caused by poor number quality. You can first submit 5,000 numbers to verify the quality of the numbers. If the quality is good, then expand the scale.

**Q: Can the data deduplication warehouse be shared across users? **

Answer: No. The deduplication warehouse belongs to a single account. It only deduplicates the numbers submitted by itself and does not affect other users. This means that the numbers you screen will still be retested (and charged) when others screen them.

**Q: How accurate is the gender field in base material screening? Can it be used for precision marketing? **

Answer: The gender field is the result of model inference and can be used for population orientation (such as male/female), but it is not an ID card-level accuracy. It is recommended to combine it with other fields for comprehensive judgment. For example, it can be used together with the age field to filter “men around 30 years old”, but do not rely entirely on it to make high-precision decisions.

**Q: Can the free generated number be used directly for screening? Will there be repeated deductions? **

Answer: Yes. Generating a number is free, and fees will be deducted only when the number screening task is submitted. It is recommended to go through the deduplication warehouse and then filter after generation to avoid repeated detection. If the generated number coincides with the historical record, the deduplication warehouse will help you skip it and save related costs.

**Q: Can I still cancel and get a refund after the task is submitted? **

Answer: No. Once the task is submitted and detection starts, the corresponding balance will be consumed; please be sure to confirm the settings before submitting. Therefore, it is strongly recommended to run a small number of samples first to confirm that the configuration is correct before officially submitting the large task.

---

After mastering the core skills of base material screening, immediately go to the console to experience the free number generation and data deduplication functions. If you need one-on-one consultation or operation demonstration, please contact two-way customer service.

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