Fixing Sample Quality: Adrien Vermeirsch on Fraud, Profiling, and the Future of Human Data

As concerns about data quality, respondent fraud, and AI-generated insights continue to grow, the market research industry is facing a critical inflection point. In this episode of the CEO Series, Leonard Murphy speaks with Adrien Vermeirsch, Founder and CEO of Enlighten, about why traditional profiling and targeting methods are failing researchers, how fraudsters are exploiting the ecosystem, and what it will take to build a more trustworthy future for human data.

From semantic matching and respondent experience to AI-moderated interviews and synthetic data, Adrien shares a candid perspective on the challenges and opportunities reshaping the sample industry.

Transcript

Leonard Murphy: Hello everybody. It's Lenny Murphy with another edition of the CEO series. Thank you for taking time out of your day to spend it with myself and my guest. And today I am joined by Adrien Vermeirsch. He is the CEO and founder of Enlighten. Adrian and I uh reconnected few months back um because you were making some really interesting LinkedIn posts around issues around sample quality and I thought I want to talk to (Leonard Murphy) this guy and it's takes just a few months to get this set up but glad to have you.

Adrien Vermeirsch: Yeah, thank you so much for having (Leonard Murphy) Welcome. It's uh it's a pleasure. Um so why don't you uh give...

Leonard Murphy: Folks a little bit about your background and uh what you're trying to do with (Leonard Murphy) Enlighten and then we can go from there.

Adrien Vermeirsch: Me. Sounds good. Uh so yeah as you mentioned I'm the the founder of enlighten I worked for five years at a sample company named potluck I started as a researcher then I moved in the product and innovation team and then in the supply team and I always had a big focus on data quality from implementing the right internal tools and developing the right internal processes finding the right external providers and I did a lot of research on research as well analyzing the impact of source then uh survey frequency on data quality survey results and that kind of things and obviously working on data quality day-to-day I identified many challenges and challenges that I wanted to solve by launching my own company so 10 months ago I left potlock to launch enlighten uh which is solving the profiling and the targeting issues we are facing in our industry h so I don't know if you want me to to tell a bit more about...

Leonard Murphy: No, go ahead. I think it's a logical -

Adrien Vermeirsch: Enlighten now or later during

Leonard Murphy: Springboard into the broader conversations around, you know, sample quality, etc., (Leonard Murphy) etc. Obviously, you were at potluck, you were inside the uh under the hood, so to speak, right? So, uh, and you what you saw must have been, um, (Leonard Murphy) uh, it forced you to act. It, it dropped. Yeah. Like, wait...

Adrien Vermeirsch: The

Leonard Murphy: we got to fix this and I think I know how. So yes, tell - 

Adrien Vermeirsch: Something is wrong here. Yeah, exactly. So, Enlighten is a research recruitment engine. Today, the industry still relies on a lot of close-ended profiling and pre-basic targeting methods. So, the matching is often inefficient. And at Enlighten, we do things differently. First, we build richer participant profiles from open-ended points. And second, we use semantic matching to figure out who is really a good fit for a given study. So instead of sending lot of people into screeners and hoping some will qualify, we identify strong matches first and only recont those participants through suppliers own system. So thanks to this approach uh we have higher incidence rate, fewer screenots. Overall, we offer a better respondent experience and...

Leonard Murphy: Okay. 

Adrien Vermeirsch: We do offer as well more transparency for our clients.

Leonard Murphy: Okay. 

Adrien Vermeirsch: So, I I really built Enlighten to fix this profiling and targeting issues because to me currently many suppliers are just sending blindly traffic and people will misrepresent themselves. They will say obviously I'm an IT decision maker. uh obviously I'm the CEO of the company because like two times out of three they are terminated uh and they see a screen sorry you don't qualify so it's really a poor experience for them and many respondents understood that they need to fit some criteria in order to pass the screeners to take the full survey and to get the incentives at the end. So if we have a more controlled environment and if (Adrien Vermeirsch) people only see the SE opportunities that really match their profiles, we would have better quality. Uh so this is exactly what I way 

Leonard Murphy: So is enlightened...

Adrien Vermeirsch: I build

Leonard Murphy: Uh than a sort of think about the (Leonard Murphy) supply uh you would be on the front end of either a panel or programmatic 

Adrien Vermeirsch: Enlighten. 

Leonard Murphy: Supply. Uh so you're building a profile of the respondent before they ever go (Leonard Murphy) into the survey itself. Is that about right? 

Adrien Vermeirsch: Yeah, it's really a two-step process. Uh, first, the suppliers I'm...

Leonard Murphy: Okay.

Adrien Vermeirsch: I'm working with send panelists on my profiling survey so that I know exactly who they are, what their job and that kind of things. And then I compare in a way these profiles with the targeting requirements of my clients. And when it's a strong match between the two, I'll tell to the suppliers, I want to recont this specific respondents because I know it will match with the targeting requirements of my clients. Uh so I'm really

Leonard Murphy: Okay. Okay. 

Adrien Vermeirsch: Flipping the dynamics I would say just to offer a (Adrien Vermeirsch) better respondent experience because when a respondent is reconted is because I'm really confident in the fact that you will qualify to the to the survey and uh to me it offers a healthier ecosystem both for the suppliers the buyers and

Leonard Murphy: Yeah. Yeah. So.

Adrien Vermeirsch: The

Leonard Murphy: The you know when we spoke before right that uh -

Adrien Vermeirsch: Respondents.

Leonard Murphy: Was trying to Yes. trying to go through it's similar, -

Adrien Vermeirsch: If I

Leonard Murphy: You know, similar problems, right? And (Leonard Murphy) what what we found um was resistance uh from the sample suppliers despite their recognition of like, yep, that this would be good. Uh this would be helpful. But two pieces. One...

Adrien Vermeirsch: Did

Leonard Murphy: Uh the resistance to the (Leonard Murphy) integration required. That was one problem. Um uh just from a you know we have a they have a product roadmap. (Leonard Murphy) They're busy. They're doing you know other stuff. So that was one lift of getting them to do something extra from an external integration standpoint. (Leonard Murphy) to the then to get the respondent to do...

Adrien Vermeirsch: Mhm.

Leonard Murphy: Something. Um that they were not necessarily being paid (Leonard Murphy) for um uh those are two challenges and and it sounds like you may be facing similar (Leonard Murphy) challenges. So what you know what are you hearing? What are you seeing in the market? And I'm not trying I'm sharing experience of like that's what seven years ago that's where (Leonard Murphy) we were and then finally threw up our hands and said ah forget the whole damn thing right you people are a pain in the butt. Um uh -

Adrien Vermeirsch: I saw, I saw the exact same dynamics. Um, as you mentioned, initially I wanted to sell this approach to suppliers. Like I wanted to sell them a product that enable them to do a better...

Leonard Murphy: So...

Adrien Vermeirsch: Profiling and a better targeting h so that they have higher completion rates and basically make more money or make more matches. Um, all the suppliers I've been talk talking to uh confirmed that I was working on the right..

Leonard Murphy: Yep. 

Adrien Vermeirsch: Challenges. uh like uh uh they know their completion rates are too low currently and they know that it causes a lot of churn but they also know that they will never implement such a solution because it's really the core of their business driving respondents on silver opportunities it's the core of their business the rooting system has been built five years ago they won't update it and they won't trust in a way blindly a startup that just...

Leonard Murphy: Yep.

Adrien Vermeirsch: That just started so that's why I pivoted at the beginning of

Leonard Murphy: Yep. 

Adrien Vermeirsch: The year to still keep this philosophy of with a better profiling and a better targeting we will have a healthier ecosystem but just selling this approach to the buyers and for the suppliers it's uh it's it's nice I would say h because they make more money both at the profiling stage because I reward them uh whenever I profile a respondents and then whenever I recontact a specific respondents they do have a CPI related to that. But my studies offer a much higher incidence rate. So, basically I could offer higher incentives. I mean the suppliers could offer higher incentives or the suppliers could have a higher margin thanks to my solution because I do offer higher incidence rates. So that's why I pivoted. I faced the exact same dynamics as you did seven years ago and I hope that I found the the a way to still monetize uh this uh this targeting and and profiling approach the way I I imagined..

Leonard Murphy: Okay. So, you've taken that's really that's interesting. So, you've gone to the buyer which in this case could be it could be PNG or it could be Apos, right? So...

Adrien Vermeirsch: Initially.

Leonard Murphy: The and similar to what the data quality (Leonard Murphy) platforms um have done rather than being at the beginning of the supply your (Leonard Murphy) uh on the demand side and uh putting that in as a filter basically uh or as (Leonard Murphy) an additional component but then you're sharing the data back with the -

Adrien Vermeirsch: Mhm.

Leonard Murphy: Supplier. So you're building the network.

Adrien Vermeirsch: Mhm. 

Leonard Murphy: Over time through natural demand. Instead of asking them to do something special, you're just opt helping them optimize and create a better experience for the buyer and respondent. Is that right?

Adrien Vermeirsch: Yeah, exactly. And..

Leonard Murphy: That makes...

Adrien Vermeirsch: Uh the the the best option I I found basically uh and this is one of the main takeaway of of building a startup. I think...

Leonard Murphy: Sense.

Adrien Vermeirsch: Uh even if you might have good ideas at the beginning uh even if it solves a real issue sometimes it's not a fit with the economic situation of the of the suppliers of the buyers h sometimes a good idea solving a real issue is not enough and you need to pivot or to find other ways to monetize your...

Leonard Murphy: Yep. 

Adrien Vermeirsch: IDs I had to to learn it uh by doing it basically I didn't read any books about it. But trying to sell something, you realize dynamics that you couldn't foresee in the first step. And it's that's why I love being a founder and and building my own thing. Like I learned so much during this process that uh it's a (

Leonard Murphy: Yeah. Well, okay. That's great...

Adrien Vermeirsch: A super interesting

Leonard Murphy: Great, great segue. Thank you for teing it up. So the uh assume this (Leonard Murphy) this is your first company uh as a founder I assume right? - (

Adrien Vermeirsch: Journey.

Leonard Murphy: Um uh that lesson on pivoting that's an important one (Leonard Murphy) 100%. The uh what else have you learned uh what's what's been like you know big aha (

Adrien Vermeirsch: I learn so much every day both in terms of hard skills and soft skills. Uh (

Leonard Murphy: Moments.

Adrien Vermeirsch: Like I'm currently alone with my company working with a with a dev. So it's a part-time dev but we're only two working on it. And obviously I need to handle everything from the sales to the marketing to the product development. Uh the only thing I'm not doing is the the actual development development of of the product. Uh so in terms of hard skills I I I learn a lot and in terms of soft skills uh obviously I need to learn how to talk to people how to to make market my solution and to understand these dynamics by myself of you need to pivot at this moment you need to work on this first and then on this other problematic second so it's really a super interesting journey and I learned a lot about the industry as well during that process because there is nobody to tell me you should work on that today, you should work on this tomorrow and I can really spend my time as I want and I think to offer a good product to my clients I need to understand perfectly the dynamics within our industry and for example I want to solve in a way the the data quality challenges we have in our industry. I want to reduce the fraud that we have in our ecosystem and the best way to do so is actually to be a fraudster myself in a way or at least to understand what the fers are doing uh to infiltrate I would say our industry h so I'm part of many...

Leonard Murphy: All.

Adrien Vermeirsch: Different telegram channels of fraudulent respondents uh just to better understand exactly what tools they use uh what suppliers are being targeted And I (Adrien Vermeirsch) also do take a lot of surveys myself to to better understand the dynamics and what sources redirect to what router and what router redirects to what end clients. H so that I can on my side take these takeaways and build a better product and work with the right actors that I know are doing things in the right way. 

Leonard Murphy: So the uh so you're a spy too. You got a little bit of espionage in uh in there. I I bet you see some oh my god stuff and I know and you share regularly what you what you are seeing on the telegram channels on on LinkedIn which I think um more people should be following you and paying attention because you're you often point out of  like look at this you know look this is crazy stuff what's what's the uh looking at all that uh what was the most kind of oh my god moment that you've seen that you were like Really? I didn't expect (Leonard Murphy) that uh that tactic. Just out of curiosity. So there's like whoa that people are really doing (Leonard Murphy) that. 

Adrien Vermeirsch: Yeah. First thing I would say is that talking publicly about data quality is pretty tough because it's never black or...

Leonard Murphy: Yeah. 

Adrien Vermeirsch: White. Uh like the the reality is always in between.

Leonard Murphy: Right. 

Adrien Vermeirsch: Uh suppliers do things in a not so good way but buyers should do something as well.

Leonard Murphy: What?

Adrien Vermeirsch: To improve uh things. Uh so when we talk about fraud when you talk about data quality challenges uh we often talk mainly about the suppliers that are doing a poor job. Uh I think the reality is in between uh and through my

Leonard Murphy: What? 

Adrien Vermeirsch: Post LinkedIn post like I always try to find this balance between this happens but it doesn't picture The whole picture basically. So it was my first uh first warning. Um but yeah, when it comes to to the the thing that struck me the the most um the click farms uh is the the not

Leonard Murphy: Yeah. 

Adrien Vermeirsch: The not the thing you will see the most in your data sets. Uh but I actually saw real videos of real click farms and you see them operating like I knew it exist but I never saw how it looked like. So seeing videos of people in the same room having multiple tabs open and answering the same surveys over and over like it (

Leonard Murphy: Yeah. Yeah. 

Adrien Vermeirsch: Really shocked me in a way. Uh but it's not the the the fraud you see the most often in in our industry. It's really happening under the hood and it's really hard to to. What is the uh the etect it actually. 

Leonard Murphy: Largest? They've said that there's a type of fraud. Uh what what are you seeing the most of? What is the the the largest fraudulent tactic? I -

Adrien Vermeirsch: I think the the thing we we forget often is (Adrien Vermeirsch) the fraud that is not technical but it's just people misrepresenting themselves. H I think we have a lot of people

Leonard Murphy: Guess it -

Adrien Vermeirsch: In our surveys that pretend to be something or

Leonard Murphy: Comes. 

Adrien Vermeirsch: Someone. uh in reality they are not so I took the example earlier of IT decision makers um I think a lot of IT decision makers in our surveys are just unemployed people who pretend to be IT decision maker h and you won't spot this through the fraud detection software that we have currently on the market the only way to solve this challenge is to better profile these people and to only send them to the right of opportunities so that's I enum for the more technical aspect of things uh what I see a lot is people overseas that use residential proxies and anti-detect brother to...

Leonard Murphy: All...

Adrien Vermeirsch: Appear as a genuine American respondents and it allows them to take the surveys that are only for American people but from countries that are not the USA and I the the fraud detection softares are working hard to identify this this type of fraud but it's always hard to to spot 100% of it. H it's always the

Leonard Murphy: Right.

Adrien Vermeirsch: The a game of you need to find as many fraudulent as pos fraudulent respondents as possible but we will never catch them all. 

Leonard Murphy: Yeah. 

Adrien Vermeirsch: Uh, all we can do is do our best.

Leonard Murphy: Yeah. So uh want to be conscious of your time as well as the uh the (Leonard Murphy) the listeners the um as we're progressing more now into the world of uh not necessarily sample providers but as human data providers how I've...

Adrien Vermeirsch: Basically.

Leonard Murphy: Been thinking about it right uh because the use cases can be a variety of things you know (Leonard Murphy) it's not just answering a survey or participate in a group it's you know share your data for synthetic or whatever the case may be um (Leonard Murphy) uh what I'm hearing is actually a greater intolerance from brands for the quality issues because the risks of contamination are now higher. Um when it's outside (Leonard Murphy) of a single study, right? The it's feeding into an LLM...

Adrien Vermeirsch: Mhm.

Leonard Murphy: You know, garbage in, garbage out. Um you're much more deeply engaged with the the that spe, you know, the sample side of the industry. Are you hearing first that same shift towards let's move away from thinking about surveys to thinking about connecting humans to provide data and are you picking up the same uh push back now from brands of saying this is the current state of play is no longer something that we can tolerate or deal with the risks are too high. What's your...

Adrien Vermeirsch: Yeah, first I I think uh there (Adrien Vermeirsch) are surveys but now there are also the AI moderated interviews for

Leonard Murphy: Take? 

Adrien Vermeirsch: Example uh and I think it's a great shift uh to to have better quality in a way because surveys might be a bit boring uh AISD interviews are more engaging. So to engage real humans, I think that AI moderated interviews is a better tool. I would  say uh to uh to engage people because surveys are not always great. Um so I agree that there is a shift from we need people to answer surveys to we need people to answer AI moderated interviews performing very specific tasks to train an AI model or to complete surveys. So and and in that world where we need people to give their opinion not only for surveys but for other types of opportunities uh verifying that we are talking to a genuine human and a real human is more important than ever. Uh because in a survey it's not that you don't care but if you have a few fraudulent respondents uh you won't necessarily...

Leonard Murphy: All.

Adrien Vermeirsch: spot it when analyzing the end results while in an AI moderated interviews you will spot it so you need to work with sources that certify in a way that it's a it's a real human and I do see some brands that don't trust uh the the supply anymore some of them even stopped performing surveys h because they they've been disappointed by what they...

Leonard Murphy: Right. 

Adrien Vermeirsch: Got in the past. They realized that half of their data set were fraudulent and they just do qual now because at least they know that they're talking to to real people. So I think that the 2026 will be a a pivotal moment for for the sampling industry. We didn't discuss about synthetic data, but I think it will al obviously change a lot of things for our industry as well. So yeah, it's a pivotal moment and I think that high quality human data will be key uh for...

Leonard Murphy: Yep. 

Adrien Vermeirsch: For the sustainability of of the sampling industry as a

Leonard Murphy: Yeah, I agree. Adrian, uh, thank you so much for taking the...

Adrien Vermeirsch: Whole. 

Leonard Murphy: Time. Uh, I am sure that we will, uh, reconnect and come back as things are progressing. Um, big fan of what you're doing and best of luck. (Leonard Murphy) Um, uh, audience, you know, look up, uh, Adrian and Enlighten. Um, he's doing, uh, doing the Lord's work here. Uh, where can people find you.

Adrien Vermeirsch: uh on

Leonard Murphy: Adrian? 

Adrien Vermeirsch: LinkedIn uh I try to be active and (Adrien Vermeirsch) to share uh insightful data or or thoughts and always uh having (Adrien Vermeirsch) interesting conversations in the in the comments as well.