More unique IPs than any network on the planet, in the world's largest residential pool. That's Shifter.More unique IPs than any network, in the world's largest pool. That's Shifter.

See the numbers

Proxy Detection Tests: How Sites Flag Proxy IPs, and How to Test a Provider's

The signals websites use to flag proxy IPs, what an online proxy check can and cannot tell you, and a proxy detection test you can run on any residential provider's exits.

Chris Collins Engineering · Updated November 9, 2026 · 6 min read
checkpassedflagged
Requests meet a detection check. Amber ones were flagged.
Authenticity
On this page
  1. How sites flag proxy IPs
  2. What an online proxy check tells you
  3. The target is the real test
  4. How to test a provider’s IPs
  5. Reading the results
  6. Doing it yourself
  7. Frequently asked questions

A proxy detection test is anything that tries to answer one question about a visitor: is this connection coming through a proxy? Sites run these tests silently on every request, and IP lookup tools run a simplified version when you paste in an address. If you are buying residential proxies, what you want to know is how a provider’s exits fare against the real versions, not the simplified ones.

This guide covers the signals sites actually use, what an online “am I using a proxy” check can and cannot tell you, and a test you can run on any provider in an afternoon. It is about the IP and the connection. For the configuration mistakes on your side that get good IPs flagged, see the common residential proxy mistakes that trigger detection.

How sites flag proxy IPs

Detection works in layers. No single layer decides; each one adds to a risk estimate.

  • The address itself. The site, or an IP intelligence vendor it pays, looks up who owns the address. A hosting company’s range is the easiest flag there is. A consumer ISP range passes this layer.
  • Known proxy exits. Vendors build lists of addresses they have seen acting as proxy exits, from honeypots, traffic partners and their own testing. Residential exits can land on these lists, and how many do varies a lot between vendors.
  • Reputation. Past abuse from the address: spam, scanning, credential stuffing. Because residential exits are shared, reputation follows the address, not you.
  • The connection. Headers a proxy adds, such as Via or X-Forwarded-For, a TLS fingerprint that matches a client library rather than a browser, DNS lookups that resolve in a different country from the IP, or an IPv6 route that bypasses the proxy.
  • Behaviour. Request rate, timing, navigation patterns, and how many identities one address carries. This layer catches clean IPs used badly.
  • Consistency. Whether the IP’s location agrees with the browser’s timezone, language and other signals.

A good residential pool should pass the first layer almost everywhere and the connection layer every time. The rest depends on how the pool is used and how widely it is shared.

What an online proxy check tells you

Paste an address into a proxy check site and you get one vendor’s view of the first two or three layers: owner, connection type, proxy flag, perhaps a risk score. That is useful, within limits:

  • It is one vendor. Services disagree, sometimes wildly, on whether a residential address is a proxy. Run the same residential exits through two detection services and it is common for one to flag a large share of them and the other almost none.
  • It is one address. A pool hands you thousands. One lookup says nothing about how much of the pool would be flagged.
  • It is not your target. The site you care about may use a different vendor, its own model, or nothing at all.
  • It tests the IP, not the connection. Most lookup tools never see your headers, TLS handshake or DNS path.

Treat a lookup tool as a quick screen for hosting addresses, not as a verdict.

The target is the real test

The only proxy detection test that matters is the one run by the site you need data from, and the only way to see its result is to send real requests and classify what comes back. A block page is a detection. So is a CAPTCHA, and so is an empty results page served with a normal status code.

This is how Atlas approaches it. Every network gets the same requests to the same set of targets, scored site by site, and a request only counts as a success if the content is real; challenges count as failures (how outcomes are classified). Hard targets separate networks that easy ones cannot.

On the authenticity side, Atlas counts the share of each pool that resolves to hosting networks, and checks the connection itself: DNS, WebRTC, proxy headers, TLS fingerprint and IPv6 fallback, asserted from the target side. Those are the layers a provider controls. Reputation and behaviour depend partly on other customers and partly on you.

How to test a provider’s IPs

A test that tells you something has three properties: many exits, more than one signal, and your real targets.

  1. Collect exits. Send a few hundred requests through the provider to an endpoint that echoes the IP and headers. Record each distinct exit.
  2. Screen the addresses. Look up every exit in two or three IP intelligence services. Count the share listed as hosting, and the share each vendor flags as a proxy.
  3. Check the connection. Confirm that no proxy headers reach the echo endpoint, that DNS resolves in the exit country, and that no IPv6 or WebRTC path reveals your own address.
  4. Hit your targets. Send a sample of your real workload. Classify every response as success, block, challenge or soft block by its content, not its status code.
  5. Repeat it. Run the whole thing again a few days later. Pools rotate and reputation drifts.

Reading the results

  • High hosting share: the pool is not what it is sold as. This is the most serious finding, and it holds whichever vendor you use.
  • High proxy flags, low hosting share: normal for residential exits that a good vendor has learned about. Look at target outcomes before worrying.
  • Connection leaks: a provider or client configuration problem. Fix it before testing anything else, because it skews every other result.
  • Clean lookups, failing targets: check your own request patterns and fingerprint first, then how the provider behaves under load.
  • Two providers flagged on the same addresses: they may share inventory. The proxy pool overlap guide explains how to check.

Doing it yourself

A short checklist for any trial:

  • A few hundred distinct exits at least, not a handful.
  • Two or more lookup vendors, compared side by side.
  • Echo endpoint checks for headers, DNS and IPv6.
  • Your real targets, with challenge pages counted as failures.
  • A second run on a different day.

If a provider looks good on lookups and bad on your targets, the targets win. Detection tests are a means to an end, and the end is pages that load.

Frequently asked questions

How do websites detect proxy IP addresses?

They combine several signals: IP reputation, hosting and ASN data, locations that do not match, connection patterns and unusual request behaviour. Many also read browser and device signals to spot automated traffic.

Can residential proxies be detected by websites?

Yes. Residential proxies are not undetectable. Their addresses belong to home networks, but a site can still flag a poor reputation, unusual traffic patterns, inconsistent locations or automated behaviour.

What is the difference between proxy detection and IP reputation checks?

An IP reputation check asks whether an address has been linked to abuse or earlier blocks. Proxy detection asks whether a connection appears to come through a proxy, VPN, hosting provider or other intermediary. Sites often use both together.

Why do some proxy IPs get flagged even when they are residential?

Because of earlier misuse, activity from other users on the same network, unusual request volumes or suspicious behaviour. Being residential does not guarantee a clean reputation or acceptance by every website.

How can I check whether my proxy IP is detected?

Online proxy checks show how lookup databases classify an address: its ASN, connection type, location and reputation. They are a useful first look, and they often disagree. The test that matters is the target's own response, so send real requests to the sites you need and count blocks, CAPTCHAs and empty pages.

Get started

Run your own detection test

Start a free trial on Shifter's residential network, screen the exits in the lookup tools you trust, then point them at the sites you actually need.

3-day money-back guarantee. Cancel anytime.