Improving the market research industry, one bot at a time

 

• • • • • •

Learn how Suzy’s latest innovation is tackling AI bots.

Artificial intelligence offers a whole range of benefits for market researchers, such as improving your day-to-day tasks like survey generation and qualitative summary generation. But with all the benefits that AI offers today’s market research professionals, there are also new challenges on the horizon—especially for online panels. Now that AI is publicly available, AI bot activity is spiking across online market research panels—potentially skewing results and data, and ultimately leading brands in the wrong direction. 

The problem isn’t new, unfortunately—and it’s an industry-wide issue. From panel providers to consumer research platforms, bots and click farms impact data quality. 

What actually are bot farms and click farms? They are organized operations that employ real humans who then create automated bots to respond to surveys for monetary incentives. These operations are often found in markets where the value of a dollar is much greater than in the USA, like Taiwan and Venezuela.


Now with the onset of AI, bot farms are becoming incredibly sophisticated, prompting a new, ever-evolving challenge for the market research industry. And as bots get smarter and harder to detect, the problem gets even more difficult to eradicate.

There’s nothing Suzy hates more than bad-quality data. So to stay at the forefront of the challenge, we’ve been hard at work on developing a solution to the AI bot problem.

When market research is done by tech companies, they usually clean the data before the survey is completed. When this work is done by traditional research agencies, they usually clean the data after survey completion – and therefore after the data has gone into your results. 

Suzy does both, cleaning at the panel and survey levels with our latest patent-pending innovation called BIOTIC. This bot detection technology adds a layer of protection by removing bad actors, bots, and AI-generated responses before they can get in and impact survey data.

Plus, if need be, we also clean the data after it goes into your results. 

Thanks to Biotic, Suzy ensures you get data that you can trust. By getting ahead of the game, we can set a new gold standard for the industry to ensure quality, actionable data for enterprise brands that want to focus on their consumers. Here’s what we’re doing and why:

First, it comes down to the audience

At Suzy, we pride ourselves on our audience, which has been built by Crowdtap. It is the backbone of our platform, and we’ve worked hard to develop it over the years. In our gamified Crowdtap app – accessible via desktop and mobile – screened and verified consumers earn points for answering questions and then redeem the points for gift cards, digital subscriptions and charity. Our engaged, active consumers can provide high-quality responses to your questions right when you need answers. 

Every member remains completely anonymous and undergoes a rigorous verification process. Plus, our consumer network is opt-in and benchmarked against the US census. And we never buy responses from 3rd parties. That means you always get the most authentic responses with the highest quality in the industry.

As a result, we are highly invested in putting in the hard work to manage our Crowdtap panel and ensure it’s a quality tool for our clients. This starts with the way we build our audience. 

With our app- and platform-based consumer-facing product, we get a broad mix of consumers joining our panel. Over the last six years of building our proprietary audience, we’ve created a special relationship with our respondents. They return over and over again to answer questions because we’ve created a fun, engaging user experience for our respondents on our gamified platform—and we’re committed to upholding it.

We build in multiple layers of protection

Our thorough, end-to-end security system includes numerous quality checks that help us ensure the panel is authentically human while weeding out any bots. 

To ensure the quality of our panels, we have a detailed security process during the onboarding of new respondents to the Crowdtap app. This includes:

  • IP Blocking - prevents users with bad IPs from being able to join. 

  • Email Domain Blocking - blocks users with fraudulent or disposable email domains.

  • Double Verification (SMS) - ensures every user is a single user.

  • Onboarding Questions - education and attention script that confirms members are real and reading the questions—with reminders on the importance of feedback. 

  • CAPTCHA and Azure Vision - CAPTCHA tests that determine whether users are human with Azure Vision that asks users to upload photos of themselves as proof of identity.

  • New Accounts Review - reviews all new accounts for overarching patterns in member signup.

To ensure the quality of our surveys, we have implemented features including:

  • Email Domain and IP Monitoring - monitors changes in email and IP addresses for known fraudulent email domains and potential VPN usage.

  • Open-end Duplicate and Relevancy Review - reviews open-ended answers in real-time during business hours to look for duplicate and irrelevant responses.

  • Speeding and Straightlining - monitors the pace and answers selected by respondents across live surveys to identify where behaviors deviate from the norm. 

Furthermore, in order to fully address the bot problem, we’ve implemented Biotic, our patent-pending bot detection technology designed to analyze the end-to-end respondent journey and keep bad actors out of your data. Working before, during, and after surveys, our Biotic technology can ensure reliable, human-generated data.

Biotic’s bot detection technology includes:

  • Biotic Actions - we take advantage of both what we might consider human weaknesses and computer strengths to deploy tests for respondents that determine if they are bots or not. Fundamentally, humans and machines don’t answer questions the same way. 

As humans, we’re all susceptible to certain tricks, like optical illusions. Or we’re able to fill in the gaps and read words when letters are jumbled or missing. Computers struggle with activities like this. So Biotic asks challenge questions that are designed to easily distinguish between bot and human. 

If a respondent proves that they are human, they see less of these types of questions. If a respondent seems to be a bot, they get exposed to a lot more of these red herrings. 

Biotic Action types include

  • Reification is based on the ability for humans to fall victim to visual illusions.

  • Memory Moat is based on the fact that bots cannot store memory in multiple formats.

  • Fetch Guard is based on the reality that a bot cannot do real work. 

  • Penrose is based on the human ability to discern if things are consistent with reality.

  • Biotic Air Gap - even if respondents pass Biotic Actions enough to receive your survey questions, the Biotic Air Gap intercepts all responses for live surveys before they appear in your data. Biotic assesses all respondents and their answers for any suspicious patterns in their behaviors during your survey and throughout their entire respondent journey: at sign-up; before, during and after surveys; and when collecting incentives later.

  • Biotic Open-end Relevance Review: Biotic also determines which open-ended responses are relevant to the question, assigning each response a relevancy score, flagging those with low scores, and sending flagged responses to a human quality checker for manual review. Responses that are flagged and declined within the 10-minute Air Gap will never appear in your dashboard. Responses that are flagged and declined after the 10-minute Air Gap will be automatically removed from your dataset, appearing in the Respondent Removal tab of your action. 

  • Biotic Honeypot - respondents who do not pass Biotic’s checks will be sent into the bot honeypot. When other companies develop tools to identify bad actors, they usually  deactivate their accounts or even give them a warning like "you are speeding," alerting the bots or bad actors to what they need to do to not get caught in the future. But with Biotic, bots do not know they’ve been caught. So these accounts remain active and unaware of their mistakes. 

Furthermore, as AI solutions become more sophisticated (ex: ChatGPT 5), bots will also grow more sophisticated, which means Biotic must be ever-evolving to stay ahead of the problem. So with Biotic, we monitor and identify patterns in the behavior of the bots while in the honeypot, and that intelligence is folded back into Biotic. This allows our team to learn from these bad actors as they evolve and helps us build even stronger defenses against them in the future. 

Plus, our dedicated Biotic team builds and deploys our own army of bots against the Biotic Air Gap and Biotic Actions to continually test and refine how we detect them, and more importantly, how we protect your data from them. 

The Suzy Difference: Data quality you can trust

Suzy stands at the forefront of market research by delivering unparalleled data integrity and reliability through its proprietary screening process and advanced technologies. 

In the end, the integration of Biotic technology ensures the authenticity of data by detecting and filtering out bots and spammers, ensuring that insights are genuinely human and actionable.

With Suzy, you gain a significant edge in making data-driven decisions with confidence and clarity. Trust in Suzy to set the standard for research integrity and empower your strategic initiatives with high-quality, reliable data.

To learn more about what we’re doing to fight spammers and bots, book a demo with our team today to see our platform in action.

 
Previous
Previous

Stay Ahead with Trackers: The Key to Understanding Consumer Behavior

Next
Next

Mastering the Iterative Process in Research