How to Calculate Cost per Reply (ROI) Using Number Filtering Data? A Complete Tutorial from Filtering to Cost Reduction

In B2B SaaS overseas customer acquisition, many teams only focus on the “activation cost” – how much they spent per valid number – while ignoring a more critical metric: Cost per Reply. Even if a number is valid and active, if it doesn’t generate a reply, your promotional budget is essentially wasted. This article starts by defining Cost per Reply, then walks you through the full practical workflow of “filtering → sending → calculating ROI,” helping you optimize your filtering strategy with data and truly reduce customer acquisition costs.

What is Cost per Reply? Why Must Overseas Marketers Pay Attention to This Metric?

Cost per Reply is the total cost you pay to obtain one effective reply, including filtering fees and promotional send costs. The corresponding activation cost only measures the success rate of number validation but cannot reflect subsequent conversion efficiency. In instant messaging scenarios like Telegram and WhatsApp, users may pass activation checks but remain inactive for long periods, never viewing messages, resulting in extremely low reply rates. Therefore, Cost per Reply is the true measure of customer acquisition efficiency.

Cost per Reply vs Activation Cost: Which is More Informative?

Metric	Calculation	Significance
Activation Cost	Total cost ÷ Number of valid numbers	Measures number acquisition efficiency, ignores subsequent interaction
Cost per Reply	Total cost ÷ Number of valid replies	Directly reflects conversion efficiency, includes full chain of filtering and promotion

For example: You spend 1000 yuan to filter out 10,000 activated numbers, giving an activation cost of 0.1 yuan each. But after sending, you only get 200 replies, making the cost per reply 5 yuan. Alternatively, if you filter by activity, keeping only 3,000 active numbers, the filtering cost might rise to 0.3 yuan per number (total 900 yuan), but replies increase to 900, lowering the cost per reply to 1 yuan. Clearly, a lower cost per reply is where real savings lie.

The Amplifying Effect of Reply Rate on Cost Analysis

Reply rate directly amplifies or reduces the final cost. Assuming a fixed total cost of 1000 yuan for filtering + sending, the cost per reply at different reply rates is as follows:

• Reply rate 2% → Cost per reply 50 yuan
• Reply rate 10% → Cost per reply 10 yuan
• Reply rate 20% → Cost per reply 5 yuan

A seemingly small improvement in reply rate can exponentially reduce costs. And number filtering is the most direct way to increase reply rate.

How Does Number Filtering Help Reduce Cost per Reply? Four Core Steps

Number filtering is not just about “verifying whether a number exists.” It directly impacts reply rates through the following four steps, thereby reducing cost per reply:

1. Eliminate Invalid Numbers: Phone numbers may be canceled, suspended, or vacant. Excluding them before sending avoids wasted message costs.
2. Filter Active Users: Telegram supports detecting 7/15/30-day activity; WhatsApp can detect valid online status. Sending only to active users can boost reply rates by 2-5 times.
3. Filter Non-Target Gender/Region: For specific products (e.g., women’s skincare, men’s games), use avatar recognition to filter by gender, further improving reply match.
4. Deduplicate to Avoid Repeated Charges: Repeatedly validating the same number across tasks wastes balance. A cross-task dedup warehouse ensures you save every penny.

The combined effect of these steps: Under the same budget, the number of valid replies increases significantly, and cost per reply drops dramatically.

Practical Steps: From Filtering to Calculating Cost per Reply (ROI)

Below is a complete workflow. We recommend following these steps and checking against the checklist at the end.

Step 1: Prepare a High-Quality Number Pool (Generate + Deduplicate)

• If you don’t have existing numbers, use the platform’s Global Number Generation feature to randomly generate numbers from 240+ regions by country and number segment.
• If you already have a number list (CSV/TXT), import it directly.
• Critical Action: Go to the Data Dedup Warehouse, upload or paste existing numbers. The system will automatically mark historically validated records to avoid repeated charges later.

Step 2: Filter Target Audience (Activity, Gender, Region)

In the console, create a new filtering task, select the target platform (Telegram / WhatsApp / iMessage, etc.) and validation type:

• Basic Filtering: Activation check (verify registration status)
• Intermediate Filtering: Activation + Activity check (recommend choosing “7-day active” or “15-day active”)
• Advanced Filtering: Activation + Activity + Gender check (based on avatar recognition, with some error, but can greatly improve reply rates for specific categories)

The estimated cost will be displayed before submitting the task. Confirm and execute.

Step 3: Calculate Costs and Export Results

After the task completes, go to “Task Records” to see actual deductions and export the filtered numbers (CSV or TXT). Record two data points:

• Total filtering cost: Amount deducted from balance
• Number of valid numbers: Count of exported numbers

Then, send promotional messages to these numbers (using bulk messaging tools or manual DMs). Record the sending cost (message unit price × number of messages sent) and the number of valid replies received.

Cost per Reply = (Filtering cost + Sending cost) ÷ Number of valid replies

Cost Analysis Example: Comparing Cost per Reply Under Different Filtering Combinations

Assume you have 10,000 raw numbers for Telegram marketing. The following table demonstrates the impact of different filtering depths on cost using hypothetical unit prices (actual prices are for illustration only; refer to the console for real costs).

Filtering Combination	Filtering Unit Price (yuan/number)	Total Filtering Cost	Retained Numbers	Sending Cost (0.01 yuan/message)	Total Cost	Expected Reply Rate	Valid Replies	Cost per Reply
Activation only	0.015	150	8,500	85	235	4%	340	0.69
Activation + Activity (7 days)	0.025	250	3,200	32	282	18%	576	0.49
Activation + Activity + Gender	0.035	350	1,500	15	365	30%	450	0.81

Interpretation: Activation-only filtering has the lowest filtering cost but also the lowest reply rate, resulting in a cost per reply of 0.69 yuan. Adding activity filtering reduces the cost to 0.49 yuan, a 29% decrease. Adding gender filtering increases the reply rate further, but the filtering unit price rises and the retained number shrinks, making the total cost slightly higher. This shows that deeper filtering is not always better; you need to balance cost and reply rate based on your target audience.

How to Judge if Your Cost per Reply is Reasonable? ROI Baseline

Two key factors for setting an ROI baseline:

1. Average Order Value (AOV): If your product AOV is 500 yuan, a cost per reply below 500 yuan (or even 100 yuan) can be profitable. If AOV is only 50 yuan, the cost per reply must be controlled below 10 yuan.
2. Industry Average Reply Rate: Based on experience, activation-only filtering yields a reply rate of about 3%-8%. Adding activity filtering can increase it to 15%-30%. You can work backwards from your actual data to determine the maximum acceptable filtering fee.

Suggestion: Track cost per reply weekly, compare different filtering combinations, and continuously optimize. If the cost rises for two consecutive weeks, it may indicate declining number quality or a need to adjust the target audience.

Hidden Costs Often Overlooked When Calculating Cost per Reply

Many teams only calculate the “unit price per filtering number” but ignore the following hidden expenses:

• Wasted repeated validations: Validating the same number in different tasks wastes money for no reason.
• Time spent on inactive numbers: Sending 10,000 messages to invalid numbers wastes not only message fees but also manual/system time.
• Manual cleaning of invalid data: After receiving many non-replies, manual removal and tagging add operational burden.

Solution: Use the platform’s Data Dedup Warehouse to avoid repeated charges, and prioritize activity filtering options. You can learn more about the dedup warehouse in the Documentation.

Checklist & Best Practices

• Estimate the cost before each filtering task, ensuring sufficient balance
• Enable the Data Dedup Warehouse and import historically validated numbers
• Choose the activity window based on your product (short cycle like 7 days for high-frequency interaction products, long cycle like 30 days for B2B with longer decision cycles)
• Record filtering cost, sending cost, and valid replies for each task
• Calculate cost per reply weekly and compare different filtering combinations
• If reply rate consistently drops below 10%, check messaging or target audience for deviations
• When using gender filtering, be aware of avatar recognition errors (about 10%-20%); combine with activity filtering for better results

Frequently Asked Questions

Q: How exactly is cost per reply calculated?
A: Total cost (filtering fee + sending/promotion fee) ÷ number of valid replies. Filtering fees are deducted from your balance based on the number of validations; promotion fees include message costs or manual labor. You can view each deduction detail in the “Task Records” section of the Console.

Q: If I only use activation checks (Telegram valid) for number acquisition, what is the typical reply rate?
A: The reply rate for activation-only numbers is generally below 5%, because many numbers are registered but long inactive. Adding activity filtering (e.g., 7-day active) can boost the reply rate to 15%-30%, making the cost per reply even lower.

Q: Can gender filtering really improve reply rates?
A: Yes. For specific products (e.g., women’s beauty products), filtering only female accounts can increase reply rates by 2-3 times. However, gender recognition is based on avatars and has some error; it’s recommended to use it alongside activity filtering.

Q: Do I need to re-filter every time? How can I avoid wasting balance on repeated validations?
A: Use the Data Dedup Warehouse. KK-DATA supports cross-task deduplication; numbers already validated in previous tasks will not be charged again, effectively controlling costs. See the Documentation for details.

Q: My budget is very small (e.g., fewer than 5,000 numbers). Is filtering suitable for me?
A: Yes. Filtering charges per number, with no subscription threshold. Even with a small amount of data, accurate filtering can improve reply rates, ensuring every penny is well spent.

---

Now you can reassess your customer acquisition costs using this method. Log in to the console to create your first filtering task, or contact our online customer service for free consultation.

👉 Log in to the console to start filtering
🤖 Two-way customer service: https://t.me/kkdata_robot
📖 Detailed user guide: Documentation
💰 View billing details: Billing page

Note: The price data in this article are logical examples. Actual deduction rates are based on real-time display in the KK-DATA console.