Benchmarking Proxies: 12 Metrics to Evaluate Quality Before Purchase and Use

Purchasing proxies without prior testing is a common mistake that leads to wasted money, blocked accounts, and failed campaigns. Arbitrageurs lose advertising budgets due to slow proxies, SMM specialists get client Instagram accounts banned due to poor anonymity, and sellers cannot scrape competitor prices due to blocks.

In this guide, we will cover all critical metrics for proxy benchmarking, show how to properly test providers, and provide specific tools for quality checks—regardless of whether you are working with Facebook Ads, managing 50 accounts in Dolphin Anty, or scraping Wildberries.

Why Proxy Benchmarking is Critical for Business

Low-quality proxies create problems that directly impact profits. An arbitrageur running a campaign on Facebook Ads through slow proxies experiences delays in loading creatives—the platform's algorithm detects suspicious behavior and bans the ad account along with the entire budget. An SMM specialist using proxies with WebRTC leaks loses client Instagram accounts due to the exposure of their real IP. A seller on Wildberries cannot promptly track competitor prices because their proxies get blocked after 10 requests.

Benchmarking is a systematic check of proxies against objective metrics before starting work. Instead of relying on the honesty of the provider, you get concrete numbers: speed of 50 Mbps or 5 Mbps, uptime of 99.9% or 85%, DNS leaks present or not. This allows for data-driven decisions rather than marketing promises.

A real example: an arbitrage team tested three proxy providers for Facebook Ads accounts in the pharma sector. The first provider promised "premium quality," but benchmarking revealed a speed of 3 Mbps and an uptime of 80%—accounts were banned within 2-3 days. The second showed a speed of 40 Mbps, but a leak check revealed an IPv6 exposure—resulting in bans as well. The third provider passed all tests: speed of 50+ Mbps, uptime of 99.8%, no leaks—the survival rate of accounts increased from 30% to 85%. The difference in one month of work was $15,000 in saved advertising budget.

12 Key Metrics for Proxy Success

Professional benchmarking covers several categories of indicators. Below is a complete list of metrics to check for different tasks.

Category 1: Performance

1. Download Speed
Measured in Mbps. Critical for working with ad accounts, uploading creatives, and scraping large volumes of data. Minimum for comfortable operation: 10 Mbps for residential proxies, 50+ Mbps for data center proxies. For mobile proxies, the norm is 5-15 Mbps (limited by mobile network speeds). Checked via Speedtest.net or Fast.com through the proxy connection.

2. Upload Speed
Important for mass posting content on Instagram, TikTok, and placing ads on Avito. If you are uploading videos to Stories or posting 50 times a day, slow upload creates delays that platforms perceive as bot activity. Norm: at least 5 Mbps for residential, 20+ Mbps for data centers.

3. Latency
The server's response time in milliseconds. High latency (over 200 ms) creates noticeable delays when working in anti-detection browsers—pages load slowly, forms are submitted with delays. For Facebook Ads and Instagram, the optimal is: up to 100 ms for local proxies, up to 200 ms for foreign ones. Checked using the ping command or via Pingdom.

Category 2: Reliability

4. Uptime
The percentage of time the proxies are available and operational. Measured in percentages over a month. Providers promise 99.9%, but reality often differs. For critical tasks (advertising campaigns, multi-accounting), uptime should be no less than 99%. Checked through monitoring over 7-14 days using UptimeRobot or custom scripts.

5. Disconnection Frequency
How many times the proxies lose connection per day/week. Even with high uptime, frequent short disconnections kill sessions in anti-detection browsers—you get kicked out of Facebook Ads Manager in the middle of setting up a campaign, losing progress. Norm: no more than 1-2 disconnections per week lasting up to 5 minutes.

6. IP Stability
How often the IP changes when using sticky sessions. For multi-accounting, it is critical that one account always operates through one IP—changing IP is perceived as suspicious activity. Checked by querying IP check APIs (e.g., ipify.org) every 10 minutes over a day. Norm: IP does not change during the paid session period (usually 10-30 minutes for rotating, indefinitely for static).

Category 3: Anonymity and Security

7. IP Leaks (WebRTC, DNS, IPv6)
Even if you are connected through a proxy, the browser may expose your real IP through WebRTC, DNS requests, or IPv6 connections. This is the main reason for bans on Facebook and Instagram during multi-accounting. Checked through specialized services: ipleak.net, browserleaks.com, whoer.net. Norm: ZERO leaks across all categories.

8. Proxy Type by Anonymity (Transparent/Anonymous/Elite)
Transparent proxies pass the X-Forwarded-For header with your real IP—completely useless for multi-accounting. Anonymous proxies hide the IP but inform the server that you are using a proxy. Elite (High Anonymous) completely mask themselves as a regular connection. For working with Facebook Ads, Instagram, TikTok, only Elite are needed. Checked via whoer.net in the "Anonymity" section.

9. IP Reputation
Even a quality proxy can have an IP from "dirty" subnets—if previous users spammed or committed fraudulent actions, the IP gets blacklisted. Such addresses are instantly banned on Facebook, Google Ads, and marketplaces. Checked through: ipqualityscore.com, scamalytics.com, abuseipdb.com. Norm: Fraud Score below 50%, IP not listed in spam databases.

Category 4: Geographical Accuracy

10. Geolocation Accuracy
The provider promises an IP from Moscow, but the geolocation database shows Vladivostok—this is a problem for targeted advertising and regional tasks. If you set up Facebook Ads for Moscow through a "Moscow" proxy, but the platform sees an IP from another city, the campaign will not launch or will get banned. Checked through: ipinfo.io, ip-api.com, maxmind.com. Norm: city and region match those stated by the provider.

11. Timezone and Locale Matching
Advanced anti-fraud systems (Facebook, Google) check the correspondence of the IP address, browser timezone, and language settings. If the IP shows New York, but the timezone is set to Moscow—that's a red flag. Anti-detection browsers (Dolphin Anty, AdsPower) automatically set the timezone by IP, but if the proxy's geolocation is inaccurate, desynchronization occurs. Checked manually: compare the timezone on browserleaks.com/geo with the real geolocation of the IP.

Category 5: Compatibility

12. Protocol Support (HTTP/HTTPS/SOCKS5)
Some tasks require specific protocols. SOCKS5 is mandatory for working with Telegram, messengers, and some anti-detection browsers (GoLogin, Incogniton). HTTP/HTTPS is sufficient for browser tasks. If the provider offers only HTTP, you will not be able to set up the proxy in applications that require SOCKS5. Checked in the provider's documentation and through test connections.

Speed and Ping Testing

Speed is the first metric users check, and the easiest to fake by providers. Some show high speeds on test proxies, but after purchase, the real figures drop by 5-10 times. Proper testing requires several checks at different times of the day.

Step-by-Step Speed Check Instructions

Step 1: Set up the proxy in your browser or anti-detection browser (Dolphin Anty, AdsPower). In the profile settings, specify the proxy type (HTTP/SOCKS5), server address, port, username, and password from the provider's data. Save the profile and launch the browser.

Step 2: Open Speedtest.net or Fast.com through the proxy connection. Click "Start Test" and wait for the results. Record three metrics: Download (download speed), Upload (upload speed), Ping (latency).

Step 3: Repeat the test at least 5 times at different times: in the morning (9-11), during the day (14-16), in the evening (19-21), at night (23-02). Providers often overload servers during peak hours—speed drops by 50-70%. If in the morning the proxy shows 50 Mbps, but in the evening 5 Mbps—that's a sign of overloaded servers.

Step 4: Compare the results with your base internet speed (without proxy). Residential proxies typically reduce speed by 30-50% due to additional hops through home networks—this is normal. Data center proxies should show 80-90% of your base speed.

Speed Norms for Different Tasks

An important point: test speed using the same tool you will use for work. If you plan to work through Dolphin Anty, set up the proxy in Dolphin and test from there—the results may differ from a regular browser due to the specifics of anti-detection profiles.

Anonymity and Data Leak Checks

IP leaks are a critical issue for multi-accounting. You can perfectly set up Dolphin Anty, use expensive proxies, but one WebRTC leak will expose your real IP, and Facebook will link all your accounts into one network. The result is a mass chain ban.

Types of Leaks and How to Check Them

WebRTC Leak
WebRTC is a technology for video calls in the browser (Zoom, Google Meet). It creates a direct connection between devices, bypassing the proxy, and reveals your real local and public IP. Most anti-detection browsers block WebRTC by default, but some proxy providers do not protect against leaks at the server level.

Check: open browserleaks.com/webrtc through the proxy. If you see your real IP (not the proxy IP) in the "Public IP Address" or "Local IP Address" section—there is a leak. Norm: only the proxy IP should be displayed, or the message "WebRTC is blocked."

DNS Leak
When you open a website, the browser sends a DNS request to resolve the domain to an IP address. If DNS requests go through your internet provider rather than the proxy, the server sees your real region. Example: you are using a proxy from the USA, but DNS requests go through Rostelecom—Facebook sees a mismatch.

Check: open ipleak.net through the proxy. In the "DNS Addresses" section, only the proxy provider's DNS servers or public DNS (Google 8.8.8.8, Cloudflare 1.1.1.1) should be displayed. If you see your internet provider's DNS—there is a leak.

IPv6 Leak
Many proxies only work with IPv4, but your provider may support IPv6. If the site requests an IPv6 address, the browser bypasses the proxy and reveals your real IPv6. This is a rare but dangerous leak—it is difficult to detect without special tests.

Check: open test-ipv6.com through the proxy. If the test shows "IPv6 address detected" and the address does not match the proxy—there is a leak. Solution: disable IPv6 in your operating system settings or use proxies that support IPv6.

Comprehensive Anonymity Check

Use whoer.net—a service that provides an overall anonymity score in percentage and shows all issues. Open whoer.net through the proxy and check:

• IP Address: should match the proxy IP from the provider's data
• Anonymity: should be "High" (Elite)—if "Low" or "Medium," the proxy reveals headers
• DNS: should be green—if red, there is a DNS Leak
• WebRTC: should be green—if red, there is a WebRTC Leak
• Timezone: should match the geolocation of the proxy IP
• Language: if using a proxy from the USA, the browser language should be en-US

Norm: overall anonymity percentage of 90-100%. If below 80%—there are serious issues, such proxies should not be used for multi-accounting.

Connection Stability and Uptime

Providers promise 99.9% uptime, but reality often differs. An arbitrageur sets up a campaign in Facebook Ads, the proxy drops for 5 minutes—the session resets, the campaign is not saved, and you have to start over. An SMM specialist publishes Stories on Instagram through 10 accounts, one account's proxy drops—the publication fails, and the client is unhappy.

How to Measure Real Uptime

The only way is to monitor for at least 7 days (better 14-30). Use UptimeRobot (a free service): create an "HTTP(s)" or "Keyword" monitor, specify any site (e.g., google.com), but in the monitor settings, specify the provider's proxy server. UptimeRobot will check availability every 5 minutes and log disconnections.

An alternative for technical users is a simple Python script that sends a request through the proxy every 5 minutes and logs the result. If the request fails—it logs downtime. After a week, you will get accurate statistics: how many disconnections occurred, their average duration, and when they happened.

A real example: the provider promised 99.9% uptime (that's 43 minutes of downtime per month). Monitoring showed: 8 disconnections in a week lasting 10-15 minutes each—that's about 2 hours of downtime per week, or 8 hours per month. Real uptime is 98.9%, not 99.9%. For advertising campaigns, this is critical—every disconnection can derail a launch.

IP Stability with Sticky Sessions

For multi-accounting, it is important that one account always operates through one IP. Providers offer sticky sessions—the IP does not change for a specified time (10-30 minutes). But some providers change the IP earlier or issue different IPs upon reconnection.

Check: set up the proxy in an anti-detection browser (Dolphin Anty), open the profile, and record the IP (via whoer.net). Close the profile, reopen it after 5 minutes—the IP should remain the same. Repeat 10 times over an hour. If the IP changes—the sticky sessions are not working correctly.

Geolocation Accuracy of IP Addresses

Inaccurate geolocation of proxies creates problems for regional tasks. An arbitrageur buys a proxy for targeting Moscow, but Facebook's geolocation database shows an IP from St. Petersburg—the targeting does not work correctly, and ads are shown to the wrong audience. A seller uses a proxy "from Yekaterinburg" to monitor prices on Wildberries, but the marketplace identifies the IP as Novosibirsk—showing different prices and availability of goods.

How to Check Geolocation Accuracy

Use several services because different platforms use different geolocation databases:

• ipinfo.io — shows city, region, country, coordinates. Used by many services
• ip-api.com — provides detailed information: city, region, postal code, provider
• maxmind.com — the database used by Facebook for determining ad geolocation
• 2ip.ru — shows geolocation based on Russian databases (important for Wildberries, Ozon)

Open each service through the proxy and compare the results. If the provider promised "Moscow," but three out of four services show "Moscow Region" or another city—the geolocation is inaccurate. For critical tasks (targeted advertising, regional pricing), discrepancies are only acceptable within the same city.

ISP Check

Residential proxies should show real home providers (Rostelecom, MTS, Beeline for Russia; Comcast, AT&T for the USA). If ipinfo.io shows the provider as "Hosting Provider LLC" or "DataCenter Inc"—these are not residential proxies but repackaged data center proxies. Facebook and Instagram easily detect and ban them.

Mobile proxies should show mobile operators (MTS, Megafon, Tele2, Beeline for Russia; Verizon, T-Mobile for the USA). If you see a wired provider—it is not a mobile proxy.

Compatibility with Platforms and Anti-Detection Tools

Proxies may work well on regular sites but get banned on Facebook Ads, Instagram, Google Ads due to specific anti-fraud checks of these platforms. Or proxies may be configured correctly in Chrome but not work in Dolphin Anty due to protocol specifics.

Testing on Target Platforms

Before purchasing, test the proxies specifically on the platforms where you will work:

For Facebook Ads arbitrage: Set up the proxy in Dolphin Anty or AdsPower, create a test profile, log into your personal Facebook account (not an ad account!). Check: does the feed open, do Stories load, can you like and comment? If Facebook requests identity verification or shows a captcha on the first login— the proxy is blacklisted.

For Instagram/TikTok: Log into a test account through the proxy, try to publish a Story, like several posts, and follow accounts. If the platform blocks actions ("try again later") or requests phone number verification—the IP is blacklisted.

For scraping Wildberries/Ozon: Open the marketplace through the proxy, search for a product, and open 20-30 product cards in a row. If a captcha appears or a message "suspicious activity" shows up—the proxies are not suitable for scraping.

Compatibility with Anti-Detection Browsers

Check the proxy setup in your working tools:

• Dolphin Anty: supports HTTP, HTTPS, SOCKS5. Ensure that the proxies pass the built-in check (green checkmark after entering data)
• AdsPower: similar to Dolphin, but has stricter validation—some proxies may not pass validation
• Multilogin: requires SOCKS5 for maximum anonymity, HTTP may work unstably
• GoLogin: supports all protocols but recommends SOCKS5 for critical tasks

If the anti-detection browser cannot connect to the proxy or shows an authentication error—the problem is either in the data format (check for spaces, special characters in username/password), or the provider blocks connections from data centers (some providers allow connections only from home IPs).

Tools for Comprehensive Benchmarking

Manual testing of each metric takes 30-40 minutes per proxy. If you are testing 10 proxies from different providers—that's 5-6 hours of work. Use specialized tools to automate benchmarking.

Ready-Made Proxy Checking Services

ProxyCheck.io — a comprehensive service for checking proxies. Upload a list of proxies (up to 1000), the service automatically checks: speed, anonymity, geolocation, type (residential/data center), presence in blacklists. Results are presented in a table with ratings for each proxy. Free—100 checks per day, paid subscription—from $10/month.

ProxyScrape Checker — a free tool for bulk checking. Supports HTTP, SOCKS4, SOCKS5. Checks availability, response speed, anonymity. Does not check geolocation and IP reputation—additional services are needed for that.

2Captcha Proxy Checker — a specialized tool for checking proxies for captcha handling. If you are involved in scraping or automation where Google reCAPTCHA appears—this tool will show which proxies successfully pass captchas and which are instantly blocked.

Built-in Tools of Anti-Detection Browsers

Dolphin Anty and AdsPower have built-in proxy checks. When adding a proxy to a profile, click the "Check Proxy" button—the browser will check availability, speed, geolocation, and show the result. This is a quick way to filter out non-working proxies, but it does not replace full benchmarking—the check is superficial.

Minimum Check Checklist (5 Minutes per Proxy)

If you don't have time for full benchmarking, use this minimum checklist:

1. Open whoer.net through the proxy → check the anonymity percentage (norm: 90%+)
2. Check the IP on ipqualityscore.com → Fraud Score should be below 50%
3. Run Speedtest.net → speed should meet your task requirements (see table above)
4. Open the target platform (Facebook/Instagram/Wildberries) → check for captchas or blocks
5. Compare geolocation on ipinfo.io with that stated by the provider → city should match

If the proxy passes all 5 points—it can be used for work. If at least one point fails—the proxy is of poor quality.

Comparison Table of Proxy Types by Metrics

Different types of proxies show different results in benchmarking. Residential proxies are slower but safer for multi-accounting. Data center proxies are fast but easily detected by platforms. Mobile proxies are the golden mean for working with social media.

The choice of proxy type depends on priorities. If speed is critical and the budget is limited—data centers. If account security and low ban risk are important—residential or mobile proxies. For Facebook Ads and Instagram, mobile proxies show the best results in account survival.

Pre-Purchase Proxy Testing Checklist

Use this checklist for each new provider. Save the results in a table (Excel/Google Sheets) for comparison.