Here's the thing most survey guides won't say plainly: survey panels are supposed to catch duplicate accounts. Their entire product is one-response-per-human data. When a brand pays for 1,000 responses, it's paying for 1,000 different people — and the panel's fraud team is the only thing standing between that promise and garbage data.

So they built detection that would make a bank jealous. If you run more than one account — or a whole household does surveys — you need to understand the four layers, because guessing wrong gets accounts banned with balances forfeited.

Layer 1: IP reputation

Every IP carries a history. Panels score IPs on: is it residential or datacenter, has it been linked to bans before, how many accounts use it, does its geolocation stay stable. A home IP with two accounts and clean history is fine. A datacenter IP with forty accounts is a bonfire. This is also why VPNs get flagged on sight — shared VPN exits have the worst reputations on the internet.

Layer 2: Browser and canvas fingerprinting

Your browser broadcasts dozens of signals — canvas rendering quirks, WebGL renderer strings, installed fonts, screen dimensions, timezone, language, audio context. Combined into a hash, it's a device ID more reliable than a cookie, because wiping cookies doesn't change it. Two "different" accounts on one browser are one device in the panel's graph. Panels increasingly use commercial fingerprinting vendors, not homegrown scripts, so the coverage is thorough.

Layer 3: Device and network signals

On mobile: device IDs, app-install fingerprints, carrier data. On desktop: TLS/JA3 fingerprints (your browser's handshake has a signature), WebRTC leaks that reveal your real IP behind a proxy, DNS patterns. These are the signals that catch people who fixed layers 1 and 2 but missed the plumbing.

Layer 4: Answer-pattern analysis

This is the clever one. Panels compare how accounts answer, not just who they are technically:

  • Speed: finishing a 20-minute survey in 4 minutes flags the response and the account.
  • Straight-lining: picking "C" down a whole grid, or identical answers across related questions.
  • Demographic drift: you're 34 in January and 29 in June; you live in Texas on one panel and Ohio on another. Panels re-ask screener questions months apart specifically to catch this.
  • Cross-account correlation: two accounts with different names giving byte-identical open-ended answers, or completing the same surveys in the same order at the same times.

Layer 4 is why technical isolation alone isn't enough — behavior has to be human too.

How the layers connect

No single layer bans you. The system builds a graph: accounts sharing an IP and a fingerprint are linked with high confidence. Add matching answer patterns and it's case closed. Panels typically don't ban on the first signal — they watch, accumulate, then ban in waves, which is why "it worked fine for three months" stories end the way they do.

What happens after you're flagged

Detection rarely means an instant ban. Most panels route flagged accounts into a review queue first: your surveys get silently invalidated (you complete them, the client rejects the data, you don't get paid), your invite volume dries up, and then the ban lands — often right before a cashout threshold. That timing isn't a coincidence; it's the panel protecting itself from paying out fraudulent balances. Balances on banned accounts are almost never recoverable, which is why prevention beats appeals every time. If you're ever wrongly flagged, a clean technical setup is also your best evidence in an appeal: separate profiles, separate IPs, consistent history.

The isolation setup that respects the rules

The legitimate use case is real: households where three people each do surveys, testers working across panels, agencies running respondent operations. The setup that keeps these clean:

  • One browser profile per human. Separate fingerprint, cookies, storage — the panel sees separate devices. This is the core of the one-profile-one-proxy structure.
  • One sticky residential IP per profile. Same IP every session, matching the profile's timezone and locale.
  • Consistent demographics everywhere. Same age, same city, same household answers on every panel. Write them down once; never improvise.
  • Human pacing. Real surveys take real time. Speeding is the fastest way to burn an account that was technically perfect.

And the line that matters: this setup keeps legitimate separate humans separate. Using it to run five fake personas on one panel is exactly what layers 1–4 exist to catch, and panels are good at it. Don't.

For the honest earning picture — what this actually pays and how the routine works — see the survey + sampling stack guide.

Question: which layer do you think caught you, if you've ever been banned — the IP, the device, or the answers? Be honest; the patterns help everyone.