The Onus of Data-based Research: Combatting Modern Fraud

About a 2 min. read

Authors
Riley Boeth
Analytics Director

In the research & insights industry, there is one particular threat that consistently demands true vigilance: fraud—especially with regards to survey data. Perpetrators’ approaches have become slipperier in the face of more advanced counteraction and prevention methods.

This summer, an article was published detailing a newly discovered method of large-scale survey fraud: the false survey platform. The company spurred to check their data due to a sudden increase in completion rate:

… [the data] seemed legitimate – the users’ devices had different IP’s, they didn’t seem to use a proxy, they had different user agents and the time it took to complete a survey differed (meaning, the users’ behavior didn’t seem as scripted like bots usually do).” But not all was as it seemed. Had they accepted the increase in rates and left it alone, patting themselves on the back for picking a fast survey provider without clear issues, this article would not exist. They went to the site that produced the increase in traffic and completions, called ‘Sandybucks’ – they “quickly concluded that there’s no “Sandybucks” beyond this signup page – it’s just a front.[1]

A front that makes its money teaching others how to build click farms and answer surveys rapidly, while hiding from those of us who try to catch them.

In 2021, GreenBook highlighted in Market Research Fraud is on the Rise – Let’s Conquer it Together that  “15 to 30% of market research data is fraudulent. In some studies, the figures are even higher. “Given the many means of survey fraud, from humans and trained bots alike, this is likely a low estimate. It is a problem that exists along a spectrum, from lazy respondents to extreme examples such as the false survey platforms mentioned above.

So how should we as an industry conquer this ever-expanding threat? How should you, as a client or vendor, pick research partners who won’t let this sit on the back burner while your data quality continues to plummet?

A good vendor and partner operates as your bodyguard. Constantly looking for threats both known and unknown. Regularly adding to their repertoire of methods for combatting the damage and preventing future attacks. No bad actor is going to give us a grace period before pivoting to a new way of evading capture, and thus our industry cannot either.

Below is a multi-pronged approach CMB recommends to remain vigilant about survey fraud:

  1. Incorporate tried-and-true methods discussed in our previous post: Overcoming Data Quality Challenges. A regular ‘well-visit’ to evaluate sample and vendor partners’ processes as well as internal processes for quality checks can have great benefit as fraud continues to evolve.
  2. Implement modernized approaches where possible, including but not limited to: Natural Language Processing and AI methods to tease out information in open ends that might be missed by classic response data checks; and segmentation methods to detect inhuman response patterns that classic quality flags would likely miss.
  3. Stay current on broader industry trends and research by attending conferences, and reading industry and academic publications; then use these learnings to push fraud detection and prevention methods ever-forward. Some conferences include Sawtooth Software, Insights Association, and Quirk’s.

With the right partners, methods, and an attitude of constant improvement and exploration,healthy research can be achieved for many years to come. Want to get started? Let’s talk.

[1] Market Research Fraud: Distributed Survey Farms Exposed | GreenBook