Survey fraud on GPT sites: speeders, bots and duplicate respondents
Surveys pay more per complete than most offers, and survey panels are strict about quality. A few bad respondents can get your traffic scored down across every survey wall you use. This guide covers what the panels look for and what you can check before they do.
Short answer
Survey panels reverse speeders, duplicate respondents, location lies, inconsistent profiles, straight-liners and bots, and they score your whole site on it. Check the network and link accounts before a user enters a survey wall, compare completion time with the survey length, and hold large or first survey earnings.
- Speeders
- Under a third of the expected time is a common threshold for review.
- Duplicate respondents
- Accounts that share a device, payout address or email.
- Location lies
- VPN or hosting IPs, or a browser timezone and language that do not fit the IP country.
- Best predictor
- Two or three reversals on one account in a week.
Why survey fraud hurts more than offer fraud
On a survey wall, the buyer of your traffic is a market research company paying for real opinions. Every response goes through quality checks after the fact. When a response fails, the complete is reversed, often days later, and the user has usually been paid already.
The bigger cost is reputation. Survey walls rate each publisher by the quality of its respondents. Too many reversed completes and your users get fewer surveys, lower payouts per survey or are screened out earlier. Honest users on your site then earn less, and some leave.
Survey walls also keep a profile of each respondent: age, gender, country, job, household. The profile decides which surveys a user is offered. Fraudsters learn which answers unlock the best-paying surveys and fill their profiles to match, then answer inconsistently inside the surveys themselves. Panels compare the two, and a respondent whose profile and answers disagree is reversed and often removed. That is why the same accounts tend to fail again and again: once a farm finds a profile that pays, it copies it across every account it runs, and every copy fails the same checks.
The fraud patterns panels catch
| Pattern | What it looks like | What you can check |
|---|---|---|
| Speeders | A 15-minute survey finished in 3 minutes | Completion time against the survey length and your own users' usual pace |
| Duplicate respondents | The same person answers as several panelists | Accounts sharing a device, payout address or email |
| Location lies | A US-only survey answered from abroad through a VPN | IP country vs browser timezone and language, VPN and hosting IPs |
| Profile inconsistency | Age, gender or job change between surveys | Profiler answers that jump around on the same account |
| Straight-lining | The same answer to every question in a grid | Usually caught by the panel; frequent reversals on one account are your signal |
| Bots and automation | Scripts filling forms, often from datacenters | Automation traces in the browser, datacenter IPs, impossible speed |
Several of these you cannot see directly, because the survey itself runs on the panel's side. What you can see is who sends the respondent: the account, its device, its network and its history on your site.
Checks to run before a user enters a survey wall
- Check the network on the survey click. Send the click to your fraud checks before redirecting. A VPN, datacenter IP or Tor exit on a country-targeted survey is almost always a reversal waiting to happen.
- Compare the browser with the IP. A US IP with a browser set to Manila time and Filipino as its first language is a strong location-lie signal, and residential proxies do not hide it.
- Link accounts before they answer. Two accounts on one device will look like two panelists to the survey wall. Catch them on your side first.
- Look at the account age and history. New accounts that go straight to the highest-paying surveys and cash out quickly deserve a hold.
- Respect the survey wall's own terms. Most require one account per person and forbid VPNs. Enforcing that on your side protects your publisher score.
What to do with completes and reversals
Survey completes arrive as postbacks like any other conversion. Treat them the same way: verify the signature, store the transaction ID once, and check the account behind the complete.
- Hold large or first survey earnings until the reversal window of that survey wall has passed, or until the account has a clean history.
- Handle reversal postbacks. Take the balance back when a complete is reversed, and flag the account for review.
- Watch completion speed per survey. Survey walls often send the survey length. Anything under about a third of it is worth a look.
- Watch earnings spikes. A user who normally earns a dollar a day and suddenly earns twenty from surveys is either very lucky or not one person.
See the postback security guide for signatures and reversals, and the cashout checklist for what to check before paying.
Keeping honest respondents happy
Survey users are often your most engaged users. They spend time, not just clicks. Aggressive rules that screen them out cost you more than the fraud does.
- Do not block on mobile IPs alone. Many real respondents answer on mobile data, where one IP is shared by thousands of people.
- Explain holds. A short message that the payout is on hold for a routine check keeps real users patient.
- Let them fix mistakes. A real user on a work VPN can switch it off. Tell them why a survey was not available instead of hiding it silently.
- Review before you ban. Put borderline accounts in a queue and look at the evidence. Farms rarely answer a support message; real people do.
How CashoutGuard helps
CashoutGuard checks the survey click and the complete like any offer. Send an offer_click when the user opens a survey and a conversion when the complete arrives, with the survey ID as the offer.
- VPN, hosting and Tor IPs, plus browser timezone and language against the IP country, flag location lies before the survey starts.
- Accounts that share a device, payout address or email are linked, so duplicate respondents show up as one group.
- Completion speed is compared with each survey's own history, and conversions without a click or with a duplicate transaction ID are flagged.
- Cashouts go to a review queue with the evidence, and you can tune any signal to your traffic in the rules.
Start in shadow mode to see how much of your survey traffic would be flagged, without blocking anyone. The free plan covers 1,000 monthly active users. See the docs to integrate.