Online surveys are a cornerstone of modern market research—fast, cost-efficient, and a crucial tool for capturing customer opinions at scale. Yet this indispensable method is under attack. Increasingly sophisticated fraud methods, from Click Farms and AI-powered bots to manipulative registration practices, are challenging the integrity of survey data. In a webinar hosted by ReDem and moderated by Sabine Hedewig Mohr of Planung & Analyse, Julia Mittermayr and Florian Kögl shed light on professional survey fraud methods and their far-reaching effects on market research results.
The keytakeaways summarized
1) What? - What Is Survey Fraud and How Does It Work?
Survey fraud has rapidly evolved in recent years. What once stemmed from inattentive participants has grown into a systematic issue driven by advanced methods and technologies. One major example is the rise of Click Farms, networks of virtual devices used by fraudsters to complete massive volumes of surveys. A striking case discussed in the webinar involved a fraudster in Venezuela completing 3,000 surveys per month and earning $2,500 monthly—a stark contrast to the country’s average wage. While Click Farms were initially operated manually, with individuals managing multiple devices, they are now largely automated. Bots have taken over most of the work, simulating human behavior to fill out surveys at scale.
Manipulating Verification Processes
Setting up a Click Farm is only the first step. Fraudsters must also register with survey panels, often bypassing verification processes. Tools like TextVerified enable them to create unlimited fake email addresses and phone numbers. Even seemingly secure measures like SMS verification or phone calls are easily circumvented. Alarmingly, even video verifications can be manipulated. A webinar example showed fraudsters using wigs and changing backgrounds to repeatedly pass the same verification process.
Location Spoofing
Another crucial tactic is IP manipulation to mask the fraudster’s true location. Tools like TunnelBear and proxy servers allow them to spoof IP addresses and appear to be located in the target region. A video demonstration in the webinar illustrated how effortlessly a fraudster in Canada could change their IP address in seconds to pretend they were in Germany. Advanced IP rotation software even enables them to generate new IP addresses continuously, simulating multiple unique participants.
AI-Powered Survey Fraud
AI has brought survey fraud to a new level. Generative AI tools like Anthropic’s Claude and ChatGPT can craft responses that appear convincingly human. During the webinar, an experiment showed ChatGPT answering the same question twice. While the first response was overly detailed and verbose, a simple prompt adjustment produced a shorter, conversational reply—virtually indistinguishable from a human response. This advancement makes detecting AI-generated answers increasingly difficult without advanced tools.
Even more concerning is Claude’s “Computer Use” functionality , which allows the AI to autonomously complete surveys. In a webinar demonstration, Claude was prompted to open a survey, navigate through questions, and generate realistic responses—all without human intervention or programming expertise. This lowers the barrier to large-scale survey fraud and poses a significant challenge to data quality.
2) So What? - What Are the Consequences of Survey Fraud?
How widespread is survey fraud, and what are its consequences? The webinar revealed several stark examples. A study by Grey Matter Research, analyzing data from five of the ten largest U.S. panels, found that 46% of panel data had to be discarded due to severe quality issues. An even more extreme case involved a survey on commercial beekeeping contracts, where over 90% of responses were deemed invalid. While ReDem typically reports lower removal rates—between 10% and 30%, as discussed in the webinar—even these figures illustrate the systemic nature of bad data.
What happens when fraudulent data is left unchecked? The consequences are profound. Fraudulent participants and bots tend to give overly positive responses: Yes, they recognize the brand. Yes, they would buy the product. Yes, they are satisfied. This upward bias skews results and leads to dangerous misinterpretations.
One example from the webinar involved a study on the brand awareness of a charitable organization. 58% of bad-quality participants claimed to recognize the brand. After cleaning the data, this figure dropped to 15%, revealing the actual level of awareness. Without this correction, the organization might have wasted resources on ineffective marketing campaigns.
A more dramatic case involved Procter & Gamble, which conducted a survey to gauge purchase intent for a new whitening toothpaste. Initial results indicated that 54 % of participants intended to buy the product. However, after thorough data cleaning, the actual purchase intent fell to just 20–25 %. The product launch failed disastrously: negative reviews dominated platforms like Amazon, causing significant reputational damage and contributing to financial losses amounting to millions.
These examples illustrate how serious the consequences are when companies act on the basis of distorted data. Bad data not only undermines the credibility of market research, but also has a direct impact on companies and their economic success.
3) What Now? - How Can Survey Fraud Be Combated?
As fraudsters deploy increasingly sophisticated tools, traditional methods of detection fall short. Florian Kögel illustrated this disparity with a striking comparison during the webinar: "Without the use of modern technologies such as AI, we are trying to win the race with a rowing boat, while the fraudsters have long been using motorboats". To bridge this gap, the market research industry must adopt innovative solutions to keep pace with evolving fraud tactics.
With tools such as the ReDem Score and 360-degree data quality assessment approaches , it is possible to respond to the challenges of today and tomorrow. The combination of several simultaneous quality checks results in the ReDem Score, a transparent and comprehensive assessment of data quality. The core elements of this system are the Time Score, which identifies unusual response times, the Grid Question Score, which analyzes response patterns in matrix questions, and the Open Ended Score, which checks open responses for plagiarism, duplicates, context/topic errors and linguistic anomalies, among other things.
The standout quality check is the Coherence Score, which identifies contradictions within a respondent’s answers. Fraudsters—whether human or AI—often fail to provide consistent responses across a survey, as they do not carefully read the questions. This innovative check sets a new standard for precision in detecting fraudulent data.
With these advanced tools, ReDem provides the market research industry with the opportunity to trade its rowing boat for a motorboat—and keep up with modern survey fraud.
Conclusion: A Game of Cat and Mouse
The webinar highlighted the growing and multifaceted threat posed by survey fraud. From Click Farms and IP manipulation to AI-generated responses, it is clear that the market research industry faces a continual game of cat and mouse. However, with the adoption of advanced technology, stricter standards, and transparent processes, it is possible to regain control and ensure the integrity of survey data.
