All locations active · 99.99% uptime
Proxy Guide

How Do You Test a Proxy Trial?

A trial period is short and usually comes with limited traffic. Spending that time randomly opening a few pages means leaving the purchase decision to chance. A good trial test should produce five measurable metrics and those metrics should come from your real targets.

In this article we present a test protocol you can apply step by step.

Before You Test: The Measurement Plan

FIGUREThe stages of a trial test
PROTOCOL01Baselineverification5 minConnectionis it established,is the exit IP correct02Geography andanonymity10 minIs the country correct, is therea header leak?03Pool breadth15 minHow many different IPs, how manydifferent blocks?04Real target test40 minSuccess rate, CAPTCHA, latency05Session stability20 minDoes sticky duration reallyhold?~90 minutes in total

Allocate most of the time to the real target test. The first three stages only verify that the service is configured correctly; the fourth stage determines the purchase decision.

Stage 1 — Basic Verification

First verify that the connection is established and that the exit is the address you expect:

FIGUREConnection and exit IP check
Terminal01# Read the exit IP through an HTTP proxy02curl -x http://kullanici:sifre@gateway.example.com:8000 -s https://ornek-ip.example/json0304# For SOCKS5 (resolve DNS on the proxy side too)05curl --socks5-hostname kullanici:sifre@gateway.example.com:1080 -s https://ornek-ip.example/json0607# See the response time and status code together08curl -x http://gateway.example.com:8000 -s -o /dev/null \\09 -w "kod:%{http_code} dns:%{time_namelookup}s baglanti:%{time_connect}s toplam:%{time_total}s\\n" \\10 https://example.com

--socks5-hostname with --socks5 the difference between them is critical: the former resolves DNS on the proxy side and prevents leaks.

If you prefer a tool our proxy check page performs the same verification in bulk and detects the protocol automatically; you can view the exit IP through My IP Address as well.

Stage 2 — Geography and Anonymity

If you requested country targeting, verify that the exit IP really belongs to that country. Then measure the anonymity level — see whether the proxy adds extra headers.

  • Country accuracy: Does the country you requested match the country reported? If not, the targeting parameter may have been written incorrectly.
  • City accuracy: If you bought city targeting, measure the deviation rate; 100% accuracy is not a realistic expectation.
  • Header leakage: The anonymity test determine the elite/anonymous/transparent level.
  • DNS leakage: DNS leak test to check where name resolution is being done from.
  • WebRTC leakage: If you are going to use a browser the WebRTC test before you start.

Stage 3 — Pool Breadth

If you are trialling a rotating service, measuring the real breadth of the pool is far more informative than the figure on the sales page:

FIGUREMeasuring the pool repetition rate
Python01import requests, collections0203PROXY = "http://kullanici:sifre@gateway.example.com:8000"04proxies = {"http": PROXY, "https": PROXY}05gorulen = []0607for i in range(200):08 try:09 r = requests.get("https://ornek-ip.example/text",10 proxies=proxies, timeout=15)11 gorulen.append(r.text.strip())12 except Exception:13 pass1415benzersiz = set(gorulen)16bloklar = {".".join(ip.split(".")[:3]) for ip in benzersiz}17print("istek :", len(gorulen))18print("benzersiz IP :", len(benzersiz))19print("benzersiz /24:", len(bloklar))20print("repeat rate :", round(1 - len(benzersiz) / max(1, len(gorulen)), 3))

190+ unique IPs out of 200 requests points to a strong pool. If you are seeing 40–50 unique IPs, the pool is not as broad as claimed.

Why block distribution matters subnet diversity article in detail.

Stage 4 — Real Target Test

This is the stage that determines the decision. On general test sites almost every proxy succeeds; the real question is what will happen on your target.

FIGUREMetrics to record in the real target test
METRIC2xx %Success rateOver at least 300 requestsCAPTCHA %Verification rateRequests returning a block pagep50 / p95LatencyPercentiles, not the averageError distribution403 / 429 / timeoutTo distinguish the source

The average latency is misleading; a few very slow requests distort the average. p50 (the median) shows the typical experience, while p95 shows the worst case.

Tip

Run the same test over a direct connection (without a proxy) as well. The difference between the results with and without the proxy shows the proxy's real contribution and cost.

Stage 5 — Session Stability

If you bought a sticky session, verify that the promised duration actually holds. The method is simple: using the same session key, send requests at regular intervals throughout the promised duration and watch whether the exit IP changes.

FIGUREVerifying a 10-minute sticky session
MEASUREMENT00:00Session opened — IP: A02:30Same key — IP: A ✓05:00Same key — IP: A ✓09:30Same key — IP: A ✓10:30TTL expired — IP: B (expected)

If the IP changes before the time is up, the session commitment is not holding. In jobs that require logins, this turns directly into an account security problem.

Results Comparison Table

If you are trialling multiple providers, collect the results in the same table. The decision should be made not on "which is cheaper" but on cost per successful request instead.

MetricProvider AProvider BNote
Success rateHigher wins
CAPTCHA rateLower wins
Median latencyp50 value
p95 latencyTail behaviour
Unique IPs / 200Pool breadth
Sticky accuracyDoes the commitment hold
Cost per successful requestThe real decision metric

What Not to Do During a Trial

The right approach

  • Testing against real targets, at a rate similar to production.
  • Recording every metric numerically.
  • Comparing against a direct connection.
  • Putting a technical question to the support team.

What to avoid

  • Burning through the entire quota in the first half hour.
  • Looking only at general test sites.
  • Deciding based on a single request.
  • Applying load far above production rate and getting the wrong result.

Summary

A trial period is a decision tool; used correctly, it grounds a choice that will last for months in solid data. Apply the five stages in order, record the results numerically and make the comparison on cost per successful request. To speed up your tests proxy checker, anonymity test and ping test you can use our tools together.

Frequently Asked Questions

01How many requests are enough for a trial?

At least 300 requests against a real target, and 200 requests for pool measurement, provide a meaningful sample. With fewer requests, random deviations distort the result.

02What happens if I use up my quota during the test?

Most trials end when the quota is exhausted. So split the test into stages and allocate most of the quota to the most valuable stage (the real target test).

03Should I look at average latency or the median?

The median (p50) better reflects the typical experience. The average is distorted by a few very slow requests. Add p95 alongside it to see the worst case as well.

04Why might my sticky session be ending early?

The most common causes are: the session key being generated differently on every request, the provider's upper limit being shorter than what you requested, and the exit IP dropping out of the pool. First verify on your own side that the key is being sent unchanged.

05How do I run the same test fairly across multiple providers?

Use the same target list, the same request rate, the same time window and the same number of requests. Run the tests simultaneously if possible; the target site's behaviour over the course of the day can affect the result.

Related Articles and Pages

NEXT STEP

Strengthen your proxy setup today.

Get started in minutes with a paid plan, or try our free proxy list first.

FREEPROXY.TR

Looking for a free proxy? You're in the right place

A complete proxy platform where you can browse up-to-date free proxy addresses, compare HTTP and SOCKS proxy types, and check your proxy connections with free tools.