• A decent panel you trust.
• Great screening questions.
• Your favorite trick/trap questions.
• A pretty rigorous data cleaning process.
• Possibly even a basic fraud defense software.
• Maybe IP address and computer browser session info.
  
  

We’ve created the market research industry’s first-ever “No-Fly” list to ensure the permanent exclusion of all known bad actors from entering or re-entering any survey, when run through Rep Data. Rep Data is uniquely able to create and maintain this list thanks to its ability to see and record 4.1 billion survey scans across nearly 200 sample sources, with approximately 15 million more scans coming in each day. Most have no idea how many times these same bad actors are popping up across the same sample sources over and over again, regardless of which source you use.

Even if a respondent is taking your survey through a first-party source, you might not know that it’s the 267th survey of their day, because it’s a bad actor equipped to game the system across many panels. You also might not know that they’ve already answered that same survey in three other sources that happen to share samples behind the scenes due to supply shifts. The industry is messy, and the No-Fly list addition to Rep Data’s Research Defender fraud prevention suite helps researchers clean it all up for their projects, permanently.

What you have now isn't enough, unless you have Research Defender.

Fraud isn’t new, but rapidly expanding and changing AI-enabled fraud is. This means, even after your current cleaning efforts, fraud is hiding inside your results, adding plenty of noise and bias in areas you can’t see. We see new emerging fraud tactics take enter across different panels all the time, and we’re the only ones capable of both seeing it and putting an end to it.

Simply pick 1 of 3 easy ways to start:

1) Plug in with our straightforward API.
2) Buy sample directly in the #1 DIY sampling tool.
3) Buy sample from our top-quality services team.

Fraudsters slip right past most digital fingerprints by intricately rotating IPs, devices, browsers, and so on. We leverage far more signals than others to identify and stop duplicate entries.

130+ Unique Signals: 
Highly specific hardware, software, and browser attributes combine to deeply fingerprint fraudsters and thwart their attempts.

Highest Identification Rate: 
While other fingerprints see many fraudulent new identities as legitimate, our deeper fingerprint catches clever duplication attempts.

Bad actors commonly attempt 50, 100, or even 1000 surveys in a 24 hour period across multiple panels at breakneck speeds. Defender is the only product which catches them.

Cross-Panel Monitoring: 
Connects the dots across 200+ panels, all the major marketplaces, and direct supplier feeds to stop duplication at an industry-wide supply level, not just in single networks.

Scale of Coverage: 
With more than 15M scans typically per day (4B+ annually), we have at least 10x more visibility than the 2nd best fraud detection software.

Makes 1 strike permanent on all sample sources with a dynamic suppression system. As this list continues to grow, fraudsters struggle to gain traction anywhere against Rep Data clients.

Blocked Here, Blocked Everywhere: 
Once a bad actor is found, anywhere, even once, Research Defender adds them to an internal suppression list, and they can’t re-enter.

Daily Updates: 
The list evolves daily as new signals arrive from the other layers. This is a living product, and it only gets stronger as time goes on.

Our Second Shield AI learns from every response later overturned in 2 or more reconciliations of the same respondent by different clients, then feeds signals back.

Pattern Recognition: 

Each reversal becomes a labeled training point using CPI, open-end length, device/OS, geo, timestamps, supplier, and behavioral markers.

Supplier Performance Tracking: 

Flags sources that repeatedly push through bad data to enable smarter sourcing, soon there will be an option to have this fully automated.

Research Defender taps into 9 actively maintained 3rd-party fraud lists from various adjacent industries. These are costly to maintain, but they catch repeat offenders others overlook.

Payment & Account Abuse Lists:
Lists from ecommerce and financial fraud that surface.

Email, IP, & Proxy Blacklists:
Suspicious domains, public proxies, or bot farms tied to fraud.

Known Fraudster Registries:
Industry-shared lists of respondents exposed as cheaters.

Device & Identity Reputation Services:
Cross-industry profiles of devices and IDs flagged for abuse.

Fraudsters will do anything to defraud your research. They are finding new tactics to break through. Defender identifies these threats, new or old, and shuts them down.

Known Fraud Tech: 
Emulators, TOR, VPNs, WebDev Tools, automation, proxy rotation, spoofing, crawlers, subnets, navigator webdrivers, obfuscation tech, etc.

Emerging Fraud Tech: 
We're always finding and catching new fraud tech tactics such as leveraging WebRTC to slip into surveys.

Automation and inattentiveness reveal themselves through specific mouse, keyboard, velocity, and contextual behaviors. As the scripting of bot behaviors gets more complex, so too do our defenses.

Sophisticated Bot Behaviors: 
Teleporting mouse jumps, ruler-straight cursor paths, zero-variance click intervals, machine-precision typing cadence, too-smooth constant-rate scrolling, identical start times across clusters, behaviors that exactly match the behaviors of another, and more.

Inattentive Human Behaviors: 
Erratic mouse zigzags with minimal dwell, rapid fire next-button clicks, bursts of typing with long pauses, inconsistent scroll spurts, unusually fast completions relative to survey length, repetitive answer patterns, and more.

Plus, The Simple Poor Behaviors:
Copy/paste, profanity, mindless open ends, over-similarities, out-of-context answers, a lack of consistency, and more.

AI-generated responses are flooding surveys, from polished open-ends to synthetic personas. Defender flags them.

LLM Text Patterns: 
We have a running list with hundreds of known patterns, with linguistic and signature structures of generative AI.

Coherence and Context Testing: 
We evaluate if open-ends actually answer the question, align with earlier responses, and show human specificity.

Proprietary LLM Detectors: 
Beyond text analysis and interaction signals. We do not disclose the full methods; just know we see when there isn’t a human driving the respondent session.

Fraud evolves constantly. Defender evolves faster by combining human monitoring with system-wide signal sharing. We feed back what we learn into the product, so it stays ahead of the fraudsters.

Internal Monitoring Team:
Analysts scan daily for signals and anomalies and investigate spikes.

Client Feedback Loops:
Issues flagged by clients are investigated and turned into defenses.

Supplier Feedback Loops:
When a partner shows rising risk, we pause and diagnose.

Continuous Innovation:
Billions of signals expose new patterns before they spread widely.