In this episode of the Greenbook Podcast, host Lenny Murphy sits down with Alfred Wahlforss, co-founder of Listen Labs and recent winner of Greenbook’s Insight Innovation Competition. Alfred shares how Listen Labs is transforming the world of qualitative research through AI-powered moderators, enabling brands like Google and Nestlé to gain meaningful insights at scale. They discuss Alfred’s unique journey from Sweden to Harvard, where his work in AI led to the creation of Listen Labs. The conversation touches on key industry challenges, including data privacy, the evolving role of researchers, and how Listen Labs stands apart by combining cutting-edge AI with human expertise.

You can reach out to Alfred on LinkedIn.

Many thanks to Alfred for being our guest. Thanks also to our producer, Natalie Pusch; and our editor, Big Bad Audio.

Hello everybody. It’s Lenny Murphy, with another edition of the Greenbook Podcast. Thank you so much for taking time out of your day to spend it with us. We know your time is valuable, and we’re going to try and give you value in return. And by ‘we,’ as always, we have a guest. And today I am joined by Alfred Wahlforss, the co-founder of Listen Labs and most recent winner of Greenbook’s Insight Innovation Competition. Alfred, welcome.

Thank you. Thank you for having me.

It’s good to have you. So, for those who don’t know, this is your time to, kind of, brag a little bit, so why don’t you tell us a little bit about Listen Labs, and then we’ll talk a little bit more about you as an individual.

So, Listen Labs, we have this AI moderator, so you can do qualitative research at scale, and interview customers using AI, and then summarize all those learnings, and get recommendations on what to do next.

Now, for those who—you know, we don’t always release video with this, so I will just say, Alfred, you’re young—compared to me, at least so [laugh]—and I just think that it’s always amazing when there are founders that are young and come from outside of the space. So, tell us a little bit about your background and how that parlayed into the Listen Labs.

Yeah, absolutely. I mean, this all started in a land far, far away, in Sweden, where I’m from. And in my undergrad, I started a company that was focused on healthcare, helping nurses and doctors get better jobs. And I didn’t know very much about healthcare, so I ended up doing a ton of research, both quantitative and qualitative. But then I always had this urge to move to the US. So, I always had kind of entrepreneurial dreams, and I was fortunate enough to go to Harvard for graduate school, where I studied data science.

And at Harvard, there was this research project. It was a little bit secretive, but it was about doing qualitative research using AI. And kind of got started, entered this research project, and when I saw the results, I was really mind-blown when you were comparing the moderator versus the AI, and they were, kind of, virtually indistinguishable. And I talked to my co-founder and said, like, “Hey, we need to move to California. We need to start a company doing this.”

And we raised venture funding from Sequoia, and our partner at Sequoia was the first investor in Qualtrics. And with his help, we were able to really speed up a lot of the work that we were doing, in less than a year, and now we work with Google, Microsoft, Nestle, Procter and Gamble, and hundreds of other companies, and help them get better insights. So, it’s been really exciting.

That is quite the ride. Now, how do you go from—I mean, I’ve done my own share of startups—but to go from idea to Sequoia, that’s a pretty big deal, so tell us a little more about that. Was that just luck? Was it perseverance? Was it a combination of—well, actually, let’s frame it that way. Can you talk a little bit about that process of taking it from idea to execution, and that journey of, you know, getting your early-stage funding, and what that was like, for other folks that are, you know, kind of in the same boat?

Certainly we were very lucky in many ways, seeing this research project and so on. But then I think it’s a lot about the team. And so, you need to have a great co-founder, and you need to have someone who is deeply technical on your team. You know, I’m technical, but my co-founder, he’s from Germany, actually, but we met at Harvard. And he grew up in a small village in Germany where there wasn’t that much to do other than code, and he coded a lot, so he became, like, the national champion in competitive programming in Germany.

And it’s kind of this—these large competitions where you compete in making the fastest algorithms. And then he also worked at Tesla Autopilot, so he really had the technical chops to make this a reality. And then I had a lot of, like, ideas and things we needed to do. And so, you need to find someone like that, that can, kind of, really complement your skills. And then I think it’s really helpful to pitch to—to have a community of other founders. So, I would really suggest, you know, make stuff, put it out there, and then other people will come to you, other kinds of founders. So, we were building a little bit of things before as well, and through that founder network, we were able to get introduced to Sequoia. So, that was kind of some advice on how we got there.

Good advice. Good advice. So, you and I first met about a year or so ago, when you were just kind of entering into the fray, if you will, and then you entered the competition in Austin and won. And I think one of the things that I heard so consistently—because I was not there, unfortunately—but was how buttoned up your presentation was. You guys even had lab coats, right, you were color coordinated when you came on stage, and you dropped the questionnaire from IKEA of [laugh] IKEA—

[laugh].

Here’s how things work. Talk about that a little bit, about the importance of how you’ve learned to—and we’re going to get into the business in a minute—but learning how to present and communicate effectively as part of this early-stage process? Because I think it’s something a lot of folks don’t learn how to do very well, and you guys, man, you buttoned that up. You were just spot on.

Yeah well, you know, consumer insights usually reports to marketing, right? We’re all kind of marketers in some way, most of us anyway, and it’s really important to think about how can people remember you when you go on stage, right, number one rule of advertising. And so, it doesn’t take that much to, I think, if you really prepare a little bit before, and you try to think, okay, how can I really stand out? And I would really recommend a physical prop. So, for those who weren’t there, I printed out a large survey I’d gotten from IKEA and showed just how big these things are on the presentation by dropping, like, hundreds of pages of paper.

And doing those small things, it might feel like a gimmick, but that will allow you to be a little bit different. I think that’s something that Sequoia always talks about, is different, not merely better, so that’s also another framework that might be helpful when making presentations. And I think this also goes to when presenting research insights, right? Because, as researchers, we collect all this data, and it’s—we love, you know, to dig deep into the data, but often the stakeholders will come away with one or two things that they can then implement, and they need to remember those things. So, it needs to be somehow, like, entertaining as well, or kind of memorable for them.

So, you make an interesting point about the memorable versus better. So, when generative AI kind of exploded, right, roughly, what, are we almost three years in now, the low-hanging fruit within the research space was qualitative, particularly on the back end, right, the ability now to easily analyze unstructured data in a very different way than we were able to do with just text analytics. So, we saw that kind of massive shift to oh, wow. This unlocks many of the bottlenecks from a process and efficiency standpoint on the back end of the qualitative process, which was enabling more scale. And we started to see the shift rapidly towards okay, if we are not bound with these inefficiencies in just data processing and analysis now, what does that do? Can we start replacing quantitative with qualitative from a form-factor methodological standpoint?

Yeah. No, it’s a competitive environment, for sure, and with startups, it’s a little bit counterintuitive, but that’s usually a good thing because that means that you’re onto something that is really valuable because otherwise, if it’s not valuable, then no one else is going to try to do it. So, we’re really excited about all the other companies that are educating the market and helping customers understand how game-changing this technology is. And if I was a customer, what I would do is look at multiple solutions—because there’s often a hundred different little details that separate things, separate software products—and try maybe one or two of those solutions and do an A/B test. Then I would say, the way we look at the market is, you have the incumbents, the very large companies, that are all, kind of, tacking on some kind of AI because they want to bump their stock price, but they don’t really reinvent the entire research methodology using AI.

Which I think is—you know, because they can’t, right? They have all this legacy technology, all these legacy customers as well, that are kind of stuck in workflows, they can’t just change those workflows. And it’s a huge opportunity to rethink that because you can speed up the research 10x, reduce costs, 10x. And so, there’s a bunch of startups, which is kind of the second category, that are doing that. And, you know, we’re one of them, but I would say there, it’s important to look at the companies who are taking this seriously.

So, we have raised millions of dollars in funding. Most companies are kind of bootstrapped or might be agencies that don’t have the technology understanding to do this. Well, you know, as an example of this, we have invested hundreds of thousands in hiring expert moderators to create discussion guides, follow-up questions, and then we have used that data set to fine-tune our AI. Most companies are just using prompting for this, which is far less superior in terms of guiding the AI.

That’s a great answer. Let’s talk about the data security aspect because in my conversations with brands, that has been the number one barrier, for large organizations, is the concerns around not feeding the beast from a competitor standpoint, right? That my—

Yeah.

Our data is our data, and you know, our key competitor is not going to benefit from learning [laugh] off of our data as well. And that’s been a real challenge when folks are just, oh yeah, we just are using ChatGPT, you know, 4.0, right, just kind of a standard API piggyback development model versus something that really is enterprise-grade. I think that we’re still a little ways off for wide scale adoption from, let’s call it, the Fortune 100 global brands because of those concerns. It’s coming, but they are being somewhat cautious. Is that your take as well, through—you’re out there knocking on doors, and you’re talking to folks every day, kind of hustling. Is that your read as well?

Yeah. I would say it’s, you know, A Tale of Two Cities. Like you have some companies who are very AI-first, and often they get from the board of directors that they need to implement AI everywhere. And then there are some companies who are, you know, very, very afraid of AI and worried. And sometimes those two can be kind of all-in-one, and they have this pressure to implement AI, but they have to do it safely.

And I think there are these open platforms like ChatGPT where they do actually train on the data, unless you toggle it not to do that, and that’s why I think it’s important to work with someone who is enterprise-grade, who don’t do that, who have all the necessary paperwork to prove that they don’t. And there’s a lot of caution, understandably so, because there’s been all this PR as well, and you don’t want to end up in such a situation, with negative press on you that you have trained some AI and leaked data. So, I think it will take some time before people realize that most of the companies are not training on the data if you’re buying enterprise licenses, and to build that trust.

I agree. All right, so let’s get into Listen Labs because I’m certainly one of those folks that, as we first encountered you and others with the fully moderated approach, a little like, “Well, wait a minute, you’re going to take my job,” type of thing, right? And the—[laugh] and I think that that is a visceral, emotional reaction that many people are still struggling with when you look at very, very strategic applications of AI, from the standpoint of where it’s not just an augmentation to a human, but arguably, is a replacement in some capacity, in some, you know, some limited use cases. So, what’s that been like? You’re selling into research organizations, and there’s somebody that’s like, “Well, moderate a bunch of groups, and no, I don’t want to do this.” [laugh]. This is scary. Talk about that emotional component as you’re really disrupting a process—and fundamentally, that’s what it is; this is a process, and technology always disrupts processes—but with a human component as well, that can be a little scary. Have you dealt with that?

Yeah, I think there is a worry from the community, and especially with all the films and, kind of, sci-fi about AIs taking over, et cetera, and there’s a caution there. But I would say jobs are generally a bundle of tasks, and those tasks, over time, tends to change a little bit. And if you look at old school surveys, you know, it used to be someone walking around and asking people on the street, “What do you think about our candy bars, one to five?” And that kind of field work has been largely reduced by online panels, but it’s not like there are any fewer researchers. I would argue, in fact, it’s way more researchers because they can kind of do research cheaper and faster, which means they can do much more of it.

So, I think that’s really where we’re heading, that when research is faster and more, kind of, accurate as well, you can do it at a larger-scale, and it’s cheaper, you kind of have infinite demand of research in general, so you’re just going to do more and more of it, and it’s not going to reduce anyone’s job, I think. But maybe some of the tasks will change. It will be more of working together with AI. I think it’s a bit of a cliché, but I really think it’s true: you won’t be replaced by AI; you will be replaced by someone using AI. So, that’s kind of how we’ve been talking about it.

But sometimes there can be a bit of a pushback, for sure. But you know, these researchers are already using a lot of unmoderated tools like surveys, and sometimes you still need to do human-moderated interviews, and especially if you don’t know anything about the space at all, it might be helpful to do a few interviews yourself first, and then let the AI do the rest, and you can kind of go back and forth like that.

So, here’s my take when these conversations have come up, and you tell me whether you agree or disagree. A hundred percent, technology will replace process, so we are not—maybe when AGI arrives, we’ll worry about things like intuition, creativity, empathy, things that, especially in qualitative research, are part of that process, understanding the business—well, so let me step back here. Here’s what I think qualitative research is evolving into, and research as a whole: we are going to be the keepers of understanding the business issue, and then making sense of the recommendations, and embedded in that—those two key components—are things like creativity, inspiration, intuition that are inherently human characteristics, depth of experience, you know, some intangibles, you know, that are hard to get to, plus the ability to actually form relationships that are important in building trust and influence, things of that nature. So, the tools like Listen Labs and others are building will increase the productivity and the efficiency, the throughput, the volume of research. I agree with that a hundred percent, and I like your point of AI will not replace you; someone utilizing AI will replace you. I think that’s a good way to put it. So, you’re close to the technology. Do you think that positioning that I just laid out is accurate as you’re having these conversations with folks on the adoption of the technology, that that’s where we are?

Yeah, I think that’s definitely true. And… I mean, another thing as well, is context, right? So, it’s really hard to give the AIs the perfect context on your business, what the problems are, and those kind of things, and presenting the insights in a context of your business is very important. So, that’s something that researchers do, and kind of also judgment of what is really, really interesting, what’s surprising, what’s new, it’s still something that AI in the analysis is not as good at, so it tends to have, like, a large volume of insights, but then you kind of have to filter out, okay, we learned this and this and this. Okay, great, ten things, but as I mentioned before, maybe you only need to take away one or two things.

But it’s much easier to, kind of, filter out from a large volume of insights to the one, two, three that matter, versus having to come up with all of those first yourself. So, I think that’s also very important: the judgment and the context. But creativity and intuition is something that AI is still kind of lacking. Though creativity, I think, is—yeah, in some ways it’s able to come up with a lot of different ideas, and sometimes it can be a little bit generic, but because of the volume, it can kind of make up on some of it as well.

Okay. All right, so I think that takes us into kind of product-market fit then. So, I look at the low-hanging fruit for Listen Labs. And you know, full disclosure, we had had a conversation after you won the competition with, kind of, my take on things, as part of the prize there. And, you know, my thought was, look, for things that are fairly standardized, let’s say customer experience, user experience, we can argue really, really well that having a qualitative exploration is going to add depth and nuance that you do not get within a survey, however, there’s not a hell of a lot of room for deep, strategic explorations, right? The topic is fairly straightforward.

And in my view, that is the easiest point of transformation from the traditional models that exist today into the AI-driven models of fairly structured, fairly straightforward, fairly limited in terms of scope of the business question. The more strategic, and you know, early stage it is, when you—you know, I think that’s where there’s a harder sell there. Whether it’s capable or not, I think it’s a harder sell. You know, a human still wants to be involved. As things become more structured, that becomes a lot easier to make that transition. Is that what you’re finding? And if you found one particular use case your clients are like, “Yeah, let’s do this all day long,” right, for our tracking survey, for instance, something of that nature? What’s that been like for you?

Yeah, you know, I think we have been doing a ton of foundational research, actually, so it’s able to do that relatively well. For example, with GLP-1, the Ozempic of the world, we have been doing some research there to like, how are people using it? How is it changing their behavior? And we’ve been able to have comparable results to a, kind of, a traditional qualitative research study. But in terms of the bread and butter of our use cases, it’s really things like concept testing, message testing, creative testing, where you do want to have, kind of, the numbers of what is the purchase intent, or how are people liking this ad, but you might also want to understand the why.

And I think, you know, this broadly comes to the big question of, you know, what is the ultimate goal of research? Or the ultimate dream of a researcher is to have this device that can, kind of, read the consumer’s mind and then provide recommendations. And we kind of have to rely on these proxies. And ideally it’s observed behavior, but most of the time, we’re kind of stuck with either quantitative or qualitative research. But I would argue that qualitative research, you know, I’m a little bit biased, but I think it actually has a ton of advantages over quant in many use cases, despite the fact that it’s not being used at all as much as quant.

And there’s a couple of reasons why I think that. You know, I think, one, it’s higher accuracy with qualitative research because you can analyze non-verbal cues. And so, if someone says, “Yeah, I love this product,” and then you see in their eyes that they’re, you know, they’re completely dead in their eyes, they don’t care, maybe you can discount that answer. And then there’s also this huge problem of fraud in quant that you don’t really have in qual, where it’s a real person, you see them over video, and it’s much harder to be fraudulent when you can ask follow-up questions. And then on the final part is really understanding the why, getting much deeper. And not just kind of why they don’t like something or not, but also using that why as a qualifier for their previous answers.

So, what I mean by that is, in surveys, there’s been a ton of, sometimes, bad results where, with the invention of the television, for example, 90% of people said they would never buy a television in a survey. But if you had done that with qualitative research, you probably could have understood why they said no, and maybe they said no for the wrong reasons because you can ask those follow-up questions and dig much deeper. And so, why are people then using quant? Well, it’s mostly because, one, it’s much lower cost, it’s much easier to work with, it’s faster because you have online panels, then ultimately, it’s also about a preference for numbers, getting statistical answers, and being able to track those numbers. And that’s fantastic.

And you can’t really do that with qual because you can’t do qual at scale. And even if you could do 50 interviews, you wouldn’t be able to analyze all of them. So, those two problems are things that AI can actually solve, and now you can actually do qual at scale. And so, you get all the benefits of qual and all the benefits of quant, and you kind of get very close to this almost mind-reading device. So, that’s a long-winded answer of saying that I think a ton of the jobs to be done for quant research are now going to be done by a tool like Listen Labs.

Agreed, wholeheartedly. I mean, the form-factor will absolutely shift, and it may look like some… I still can’t imagine doing a conversational conjoint study, right? That just seems like that would be a pretty onerous process experientially, so I personally envision that there will be something that looks like a survey around very specific question types or use cases. But for the bulk of what we do, it will be more conversational in nature, and driven primarily via AI. And I do think we get, on the back end, that same—we get the best of both worlds. We get the quantitative data that will be easily analyzed, as well as the depth that comes through the nuance that comes through the open-ended answers.

You brought up an interesting point that I don’t think we’ve really talked about with anybody yet. So, comparability from a cost standpoint to a survey in terms of sample. So, what does that look like for you? If the 15-minute survey, do you say, okay, we’re going to keep this to a 15-minute AI session, right, moderated session as well, so that length of interview stays the same, since that’s the primary driver of cost in terms of sample. Or has that been a challenge? Have you found that it’s longer because you’re doing, you know, more open-ended answers? What’s that been like for you? What’s your experience so far in dealing with the price issue?

Yeah, so it is—because qualitative research tended to be synchronous and over video or, you know, even in person, it tended to be much more expensive. But now with Listen Labs, you can do it asynchronously, and so that really reduces the cost of the sample. And so, for 15 minutes, we do—you know, it is a little bit more expensive than a quant sample generally, but then I think quant is probably a little bit too cheap because there might be [laugh] some bots and survey farms in there as well. And you really reduce that with qual sample. And in terms of the time it takes, it highly depends on the study, but what I found is sometimes you can, kind of, also read between the lines and avoid a lot of the quant questions that you might not… you might not need to ask all of those questions when you’re doing qual.

For example, we took one of those 15 personality tests and converted it into Listen Labs. And Listen Labs can kind of figure out my personality type in three questions that are open-ended, versus fifty questions in the personality test because it’s able to, like, get much more nuance in how I, kind of, react to the questions and those kind of things as well. So, it’s a little bit hard to compare, but it depends on the use cases. But generally it’s around the same time.

Any feedback from respondents with almost the Turing test type of issue of, am I talking to a person or am I talking to a robot? What have you heard from respondents about the experience?

Yeah. You know, most people are really excited about it. And it’s a way of—when we’ve done these very long studies where you ask them to kind of talk about themselves, and think about the future, those kind of foundational research studies, people feel that they can really open up, and it’s almost like a psychology session for themselves, where they can just share how they think about the world, and kind of crystallize some of the thoughts they had. And so, we actually get messages all the time from respondents saying, like, “Wow, this was the best study I’ve ever taken.” Now, this is comparing it to a survey.

And if you’ve ever done a study, a survey that’s more than ten minutes, you quickly realize, like, it’s a terrible experience. And so, compared to that, it’s obviously much better. I think comparing it to a human interview, it’s probably not there yet, but there are some other benefits, that you don’t get judged by the person, you can kind of just sit and do it in your own time, and it’s kind of fun just reflecting on your own views.

All right. I want to be conscious of your time as well as our listeners. There’s a question I should have asked early on that I didn’t, and that was a little self-promotional here. How was competition beneficial to Listen Labs? What did you guys get out of that specific experience that helped accelerate the business?

Yeah. I mean, in this roller coaster, the IIEX Austin was the highlight so far. It was really exciting, you know, meeting everybody in person and so on. And a lot of the big enterprise customers also came from that, and it’s clear that IIEX is kind of a beacon for all of the insights professionals, and they really care about what you guys think, so thank you so much. I think it was you who recommended that we even apply, so I will be forever in gratitude to you [laugh].

Okay, very cool [unintelligible 00:30:42]. I did not coach him on that answer in any way, shape, or form, but that was a pretty darn good one, Alfred. Thank you. And sincerely, thank you. We started the competition early on because I wanted to be able to help early-stage companies have that experience, and I think our track record has been pretty good in doing that, so it’s very gratifying to hear that. We may quote you on that in [laugh]—

[laugh]. Please do. Please do.

Okay. Very cool. Anything that you wanted me to ask or that you wanted to bring up that we have not touched on?

Yeah, I guess I haven’t really explained in detail what we do, but I think most people kind of understand it, but maybe I can walk through quickly, like, how the platform works [laugh].

Yes, go—here’s your pitch opportunity. Yes, please do explain.

Yeah. No, so just the basics. You know, you can co-create a discussion guide with AI, or you can upload your own, like, Word file, and then essentially Listen will create a semi-structured interview based on that discussion guide, and can upload stimuli and those kind of things. And we have a million respondents that you can recruit from both B2B and B2C. And at the back end of this, it will then summarize all of those learnings and give you a slide deck automatically, and also highlight reels, and those things. So, that’s kind of how it works. And we’d love to, if anyone wants to do a demo and compare us to maybe the competition or so, feel free to reach out.

Very cool. Now, how can people reach you? Where do they find you?

Yeah, it’s listenlabs.ai, or alfred@listenlabs.ai or follow me on LinkedIn: Alfred Wahlforss. I have a complicated name from Finland. I’m half Finnish. But yeah, alfred@listenlabs.ai.

Okay. This has been really great, Alfred, and best of luck to you. And obviously it’s not about being first; it’s about making the most impact, and it certainly seems like you guys are doing a damn good job of that. So, look forward to you joining the ranks of other companies through the competition that have become industry standards and major players. So, hats off.

Thank you so much, Lenny.

You’re welcome. All right, that is it for this edition of the Greenbook Podcast. I want to give a big shout-out to our producer, Natalie, to our editor, Big Bad Audio, to our sponsor, and most of all to you, our listeners. We appreciate you so very much in taking the time to spend it with us as we have these really interesting conversations. That’s it for this edition. We’ll be back real soon with another one. Take care. Bye-bye.