How to set the quality index of number screening results? A practical evaluation guide

In the scenario of overseas customer acquisition, number screening is the first step to reach target users in batches. Many teams spend a lot of time screening tens of thousands or hundreds of thousands of numbers, but the messages sent out either go unanswered or are blocked. The problem often lies in the quality evaluation indicators of the screening results. In the past, people were used to looking only at the “activation rate” and thinking that a number that could be registered was considered qualified. However, on social platforms such as Telegram/WhatsApp, there is a big gap between “being able to connect” and “being able to effectively reach”. This article breaks down the quality definition of number screening from five core dimensions, gives threshold recommendations for different business scenarios, and attaches a checklist that can be used directly for benchmarking to help you spend your budget on numbers with real conversion potential.

What is the “quality” indicator in number screening?

The “quality” of number screening is not a single-dimensional value, but a comprehensive score composed of multiple dimensions such as effectiveness, activity, data accuracy, deduplication rate, and cost efficiency. Many people only focus on the “activation rate” and think that as long as the platform returns “activated”, it is a good number. This simplified thinking will lead to inefficiency in follow-up marketing - even if a user registers Telegram, if it has not been online for half a year, or the age or gender does not match your product at all, sending multiple messages will be a waste.

Why is it not enough to look at the activation rate alone?

Suppose you perform number screening on 10,000 numbers, and the activation rate is as high as 80% (8,000 are activated), but only 30% of them have been active in the past 30 days (2,400). You spend 100% of the screening fee to buy 8,000 “nominally valid” numbers, but only 2,400 can actually be reached - the effective cost has tripled. What’s worse is that if half of the numbers in this batch are men and you sell women’s makeup, there are only 1,200 that actually match. **The opening rate is just a starting point, not an end point. **

Five core evaluation dimensions of number screening result quality

The following table lists the five dimensions that need to be paid attention to when evaluating the quality of screen numbers, and their impact on subsequent marketing:

Dimensions	Description	Impact on marketing
Activation detection accuracy	Can it correctly identify whether a number has been registered on the target platform (such as Telegram, WhatsApp)	Low accuracy leads to the mixing of invalid numbers and wastes message costs
Activity window matching	You can specify the active time period (last 1 day/7 days/30 days/90 days)	The shorter the window, the higher the timeliness of contact, suitable for limited-time activities; the longer the window, the larger the number pool, suitable for long-term cultivation
Accuracy of additional fields (gender/age/avatar, etc.)	User portrait fields inferred by the screening platform based on public data and models	Although not 100% accurate, it can be used as a rough screening reference to greatly reduce the targeting scope
Cross-task deduplication rate	Whether there are duplicate detections between different batch numbers	The higher the deduplication rate, avoid repeated deductions and control costs
Average cost per line	Testing costs for each number under different test combinations	Directly affects the project budget and needs to be balanced between quality and cost

Dimension One: Activation Detection and Activity

“Activated” means that the mobile phone number has been registered on the target platform. “Active” means that the account has usage behavior (such as sending messages, logging in, posting) within the specified time window. Taking Telegram as an example, the platform usually provides three active windows: last 1 day (online), last 7 days (recently active), last 30 days (generally active). If you are doing community recruitment, it is recommended to give priority to numbers that have been active in the “last 7 days”. Such users are more likely to respond to invitation links; if you are doing system notifications, you can relax to the “last 30 days” because the messages will be pushed to the mobile phone notification bar.

Dimension 2: Availability of additional fields (gender/age/avatar, etc.)

Platforms such as Telegram, Line, and Zalo will provide some fields (such as display name, avatar, and personal profile) in public user profiles. The screening platform can identify gender based on this. Some platforms can also infer age groups through data such as language, time zone, and registration time. For example, KK-DATA’s Telegram gender detection results include the “age” field, which can be used to filter out people who are about 30 years old. Note: This type of inference is not official real-name data, and the accuracy is usually 70%~85%, but it is enough as a directional reference - in overseas marketing, if 70% of irrelevant users can be excluded, the cost of dialogue can be greatly reduced.

Dimension 3: Cross-task duplication rate

When a team continuously submits multiple batches of number screening tasks, if there is no duplication mechanism, the same number may be repeatedly tested twice and the fee will be deducted twice. KK-DATA provides data deduplication warehouse, the system will automatically compare the task history, and the same number will only be deducted once. When evaluating the deduplication rate, you can check the ratio of “actual number of charges vs. total number of tasks.” Ideally, the actual number of charges after deduplication should be 80% to 95% of the original number (depending on the duplication of number sources).

Dimension 4: Average cost per item

The more detection dimensions, the higher the cost per item. For example, only detecting “Telegram activation” may have the lowest unit price, but the cost will increase after adding “active + gender”. It is recommended to test different combinations on a small sample first and select the solution with the highest cost-effectiveness ratio. The specific unit price** please refer to the real-time price of the console** for details. The unit price is different for different platforms (Telegram/WhatsApp/Line/Zalo, etc.) and different detection types. Do not make a unified estimate.

How to set the quality threshold for number filtering based on business scenarios?

Different customer acquisition scenarios have very different requirements for number quality. The following are suggestions for threshold setting in three typical scenarios to avoid “one size fits all” wasting budget.

Scenario 1: Telegram community fans

• Detection type: activated + active in the past 7 days + tgid export
• Threshold recommendation: activation rate ≥70%, activity rate ≥40% (only the active part in the past 7 days is taken)
• Reason: The community invitation link requires users to actively click. Users who have been active in the past 7 days are more likely to check the notification in time and join. After the tgid is exported, it can be used for group addition or API calls without relying on mobile phone numbers.
• Cost Tradeoff: If the budget is limited, you can abandon the “tgid export” option and only do “activate + active”, which will reduce the cost by about 30%.

Scenario 2: WhatsApp one-on-one private message

• Detection type: activated + active in the past 30 days + avatar detection
• Threshold recommendation: activation rate ≥ 85%, activity rate ≥ 60%, proportion of avatars ≥ 50%
• Reason: WhatsApp private messages are charged per message (even paid channels are required), and high opening rate and high activity rate are the basis for responses. Accounts with avatars are generally more trustworthy and less likely to be labeled as “spam accounts.”
• Cost Trade-off: The cost of avatar detection is usually very low, and it is recommended to be a must; age/gender depends on the product, and is not required.

Scenario 3: Line Marketing in Southeast Asia

• Detection type: activated + active + gender (such as male) + region matching
• Threshold recommendation: activation rate ≥75%, activity rate ≥50%, gender targeting accuracy ≥80% (manual verification of small samples)
• Reason: Line has a large user base in Thailand, Indonesia, Taiwan and other places, but its online identities are relatively scattered. Through the “gender + region” combination, the target group can be quickly identified to avoid sending it to irrelevant users.
• Cost Tradeoff: Region detection relies on number ownership (non-GPS positioning) and is only used as a rough screening. If the product is not limited to regions, this item can be skipped.

Common misunderstanding: Why is it not enough to only look at the “opening rate”?

Use a comparative case to illustrate:

Indicators	Plan A (only look at activation)	Plan B (activation + activity + gender)
Total number	10000	10000
Opening rate	80% (8000)	80% (8000)
Activity rate (last 30 days)	Not detected	60% (4800)
Gender match (male 60%)	Not tested	60% (2880)
Actual Effective Number	8000 (but actual active & matched unknown)	2880
Single test cost	Low	High (but the effective cost per test is lower)
Subsequent conversion rate (assumed)	2% (160)	6% (173)

It can be seen that although the detection cost of option B is higher, the number of effective conversions in the end is more, and the single user acquisition cost is lower. **Looking only at activation rates will hide a lot of invalid spending. **

How to configure high-quality number screening tasks on the KK-DATA platform?

Taking the KK-DATA screening platform as an example, it demonstrates how to improve the quality of results by combining multiple detection types (the operation logic of other platforms is similar):

1. Import numbers: You can upload the target country/region number segment through CSV upload or use the global number generation function (free).
2. Select detection type: Check “Enable detection + activity window (for example, last 7 days) + gender recognition + tgid export” on the task configuration page. The system automatically calculates the estimated cost based on the combination.
3. Enable data deduplication: Make sure “Cross-task deduplication warehouse” is enabled to avoid repeated deductions for the same number.
4. Submit task: Select the notification method (Telegram message notification), and then confirm the submission.
5. Export results: Download the CSV/TXT file after the task is completed. The fields include number, platform status, activity level, gender, age (some platforms), tgid/wsid/uid, etc.

Number screening quality checklist (can be saved)

The following checklist is for the team to check before each screening task to avoid missing key quality dimensions:

• Is the target platform clear (Telegram / WhatsApp / Line / Zalo / iMessage, etc.)?
• Is the activity window set (such as the last 7 days/30 days)?
• Are additional fields such as gender/age/avatar enabled based on product audience?
• Is the cross-task data deduplication warehouse enabled? Do you know the deduplication ratio?
• Have you first selected a small sample (such as 1,000 items) for testing tasks to compare the unit price and effective output of different testing combinations?
• Are the deduction details of this task recorded to review the trend of each effective cost?
• Have you verified the accuracy of the screening number results (manually extracted 50~100 numbers for secondary confirmation)?

FAQ

**Q: What type of test has the best quality number screening results? ** Answer: There is no absolute “best”, it needs to match the business goals. If you are doing a one-time mass posting, it is recommended to select “Activate + Active in the past 30 days”; if you are doing crowd targeting, it is recommended to add “gender” and “age fields”. You can refer to the real-time price of the KK-DATA console to compare the costs of different combinations.

**Q: Which is more important, activity or activation rate? ** A: It depends on the marketing method. Automatic system messages can tolerate lower activity levels, but manual private messages or high-cost channels should prioritize ensuring high activity rates. It is generally recommended that the activity rate is no less than 30%.

**Q: How to judge whether the results of a screening platform are reliable? ** Answer: A small sample test can be used: take a batch of known valid numbers to see if the platform can correctly mark them as “activated”; take another batch of invalid numbers to see if they are misjudged as valid. Repeated detection results from different platforms can also be compared.

**Q: Do I need to enable all detection options in the number screening task? ** Answer: No. The more detection dimensions are enabled, the higher the cost per item will be. It is recommended to selectively enable it based on the target customer profile. For example, if you are only targeting men, you can only enable gender detection, not age.

**Q: After number filtering is completed, what fields are included in the data export? ** Answer: Supports CSV, TXT and other formats. The exported fields include number, platform activation status, activity level, gender, age (some platforms), tgid/wsid/uid, etc. The actual export version of the KK-DATA console shall prevail.

---

Higher quality number screening results can directly reduce customer acquisition costs and improve conversion efficiency. If you are preparing to enable or optimize the screening task, you can log in to the KK-DATA console now to experience the complete detection combination configuration, or you can directly contact two-way customer service for one-on-one configuration suggestions.

👉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/