Meet Clarinova: A New Way to Know Your Audience

[01:00:00:00 - 01:00:05:14]
 (Music)

[01:00:05:14 - 01:00:23:20]
Speaker 1
 It is the first time that the Flynn effect has a guest that is a non-Flynner. So Dan, Becca, welcome.

[01:00:23:20 - 01:00:27:17]
Speaker 23
 All right. Well, thank you for having me. Yes. I didn't realize that was so special.

[01:00:27:17 - 01:00:42:00]
Speaker 1
 Sure is. And we'll get into the backstory a little bit, but you are a former Flynner, and now we are partnering on a very fun endeavor called Clarinova that Flynn is a junior partner in.

[01:00:42:00 - 01:00:52:23]
Speaker 1
 So we'll talk all about that, but I guess maybe you can just kind of tell our friends here a little bit about like your backstory and, you know, your interest in technology and all the fun things.

[01:00:52:23 - 01:01:31:16]
Speaker 23
 Sure. Sure. Well, I come from a varied background of technology, went to school for technology back in the day when it used to be called IT and it was a whole practice. Kind of wandered my way through the marketing world into more data related things and then more marketing strategy, customer experience and research and have just always enjoyed all things technology. So everything that's happening now with AI has been like almost this little bit of a rebirth or a reimagining of like all the joy and all of the crazy things that you can do with technology

[01:01:33:07 - 01:01:35:19]
Speaker 23
 that's come into play. And so it's a little bit of a

[01:01:37:00 - 01:01:59:01]
Speaker 23
 just sort of culmination, I guess, of all of these different things that kind of led to this idea behind Clarinova and kind of getting started with synthetic audiences, which I'm sure we'll talk about a little bit more. Of course. And so I think that's what you can do in terms of research and using some of the power of data and some of the power of AI and blending those two things together to help people solve problems and

[01:02:00:02 - 01:02:03:17]
Speaker 23
 work through different research ideas and topics.

[01:02:03:17 - 01:02:10:06]
Speaker 1
 Yeah. No, absolutely. And I'm so excited for you because so we worked together here at Flynn for seven, eight years.

[01:02:10:06 - 01:02:11:13]
Speaker 23
 I think, yeah, seven years.

[01:02:11:13 - 01:02:44:21]
Speaker 1
 Yeah. I mean, a good period of time and you were on our senior management team. So we would have these conversations often around like, what are we building at Flynn? And there were often like things that were coming up of like, well, what if we built like a product or a platform or something? Like I knew it was always an area of interest for you. So it's really exciting to see this like coming through to fruition. And obviously you've got the technical skills to make it happen. But so, yeah, why don't we just talk about like, what is Clarinova? Obviously you've mentioned these synthetic audiences, but like, what is your vision of what Clarinova offers?

[01:02:44:21 - 01:03:46:11]
Speaker 23
 Yeah. So Clarinova is really a way for people to get more feedback from their audiences without having to go through a lot of the headache and hassle of scheduling and finding and the costs and times. Yeah, the paying. So all of those things that come along with research that make it a little bit tougher or make it harder for clients and people to say, yes, let's do this. How do we eliminate some of those barriers? And, you know, again, it's never going to be perfect as real research, but it can get you a really far along path to that. So being able to take your data, model your customers, and then be able to have conversations and put things in front of them, whether it's messaging or creative or talk about product roadmap and features, what types of things might be important to that audience and be able to get that feedback super quick. Right. And it's better than just getting, you know,

[01:03:48:07 - 01:04:08:11]
Speaker 23
 random feedback from or best practices feedback from an LLM or something like that. But without the, again, hassles of, OK, now I need to schedule, I need to find, I need to do the recruitment for my audience, I need to get them in, we need to have conversations and then you got to get transcripts and things like that.

[01:04:08:11 - 01:04:25:12]
Speaker 1
 And then the recruitment for nothing, even just the scheduling part of that is such a massive time suck. You know, we do a lot of traditional research and there's a time and a place for that always. Right. But it, it, it, that will, you'll get people canceling on you. You got people in different time zones. Like it is, it's a headache.

[01:04:25:12 - 01:04:34:03]
Speaker 23
 Yeah. How many times has it been? Oh, okay. It's been five minutes. It's been 10 minutes. Are they showing up? Are they not showing up? Yeah. Totally frustrating.

[01:04:34:03 - 01:05:12:11]
Speaker 1
 No, absolutely. And I think like the, the portable nature, but the real time data that you can get back is so awesome. And we'll talk a little bit about like how Flynn has used that this capability already. But I think one of the really interesting things about Clarinova is that there's kind of two paths that you can go in terms of the actual audiences. And so there's this like Clarinova universe of folks that, you know, have certain data points associated with them. And for lack of a better term, we're kind of like the gen pop side of things. Yeah. And then this more custom side of things. So can you kind of talk about like how you view the differences between those two and kind of like what goes into each?

[01:05:12:11 - 01:05:37:10]
Speaker 23
 Yeah. Yeah. So wanted this to be a tool and a resource for anybody that wanted to jump in and do this type of research that wanted to get some feedback and wanted to be able to get it quickly. So the fastest way to that is by having a little bit of an off the shelf audience that you can tap into. So we built this audience that was modeled after the U.S. population that you can really tap into. And if you're looking for

[01:05:38:15 - 01:06:38:10]
Speaker 23
 females age 35 to 54, or you're looking for seniors, you know, whatever you happen to be wanting to find, you can go through and find a population that fits that different behaviors, attitudes that are again modeled and indexed to the U.S. population. And then you can start to tap into those different folks in the form of a focus group and have some discussions with them. So that's kind of the quick start path, I guess. If you don't already have some audience or customers, maybe you're getting newly started or you're just trying to get going as quickly as possible. That's kind of Avenue A. The second path is if you want to design a custom audience for yourself. So that could be built off of past customers. It could be built off of a new target segment that you want to go after. It could be built off of prospects. But if you have some of those data points that exist, we can then use that to model a new universe, a new audience population for you.

[01:06:39:14 - 01:07:12:09]
Speaker 23
 Pulling in those different demographic attributes, financial attributes, behaviors, attitudes, all of those types of data points to build this replica almost, right, of a model of your customer. So again, it's not exactly using your customer data, but it's modeling to get as close as possible to that universe. And then from there, you have the ability to pull in and talk to whoever you want within that group of modeled customers that you've built. So it's flexible to be able to really use either or or a combination of the two audiences, however you want.

[01:07:12:09 - 01:07:33:12]
Speaker 1
 Yeah. No, and it's so awesome. We have used it for really quick feedback. Obviously, as an agency, we have used it. So for example, we have a financial services client who that's kind of where this sort of began is by modeling one of their audiences. And now they ask for it by name all the time. They just call it our synthetic people.

[01:07:34:13 - 01:07:55:04]
Speaker 1
 And so we'll run creative ideas, concepts, specific copy, ask specific questions past the audience. But then also internally at Flynn, our creative folks, our strategy folks will ask questions while we're developing things before they even get to the client. So it's just we have this ability to do more of that,

[01:07:56:11 - 01:08:35:19]
Speaker 1
 kind of put more of that like accountability in place, right? Like be doing more consistent checks throughout the process where it used to be like we would pay out of pockets to do sort of like concept validation. And we still do that where appropriate, of course. But you had to be so far along in that process that you pretty much had like fully baked concepts to be able to show to an audience so that they could actually understand it. And then what happens if they don't like either of them, right? Like there is something nice about being able to have these like checkpoints along the way, especially on the creative side before we get into like the full execution.

[01:08:35:19 - 01:09:19:10]
Speaker 23
 Yeah, absolutely. To be able to test along the way. And that's the beauty of a platform like this is instead of choosing to say, let's do all the research either upfront, like limited budgets, right? So maybe we do a study upfront or we do a study at the tail end to validate, you know, you're kind of maybe picking one or the other because you don't have the funds. Now here's a way to let's start at the beginning. Let's do a check in the middle. Let's tweak and modify. And you can kind of almost like always on feedback that you can get to kind of help you along the way. And again, there's no substitute for real, real research with real people, but this can definitely help accelerate that. It can help you plan for those studies and it can just

[01:09:20:20 - 01:09:24:01]
Speaker 23
 get you closer to the clients that you're trying to serve at the end of the day.

[01:09:24:01 - 01:09:42:04]
Speaker 1
 Well, yeah, I was just going to say like if it allows us as an agency to be that much smarter or it allows the clients who are client side to be that much smarter in terms of the decisions that they're making because a lot of the time and you know, this is the fine balance and we're still going to always have this as

[01:09:43:06 - 01:10:09:02]
Speaker 1
 a marketer, you have to both understand your audience and use a little bit of gut to say like, okay, no, I think this is what's going to happen. But sometimes your personal opinions, beliefs, propensities for different types of media can cloud those things. Like I'm a person who, you know, when we talk about Facebook, I'm like, nobody's on Facebook. Well, that's not true, Katie. Lots and lots of people are on Facebook, you know?

[01:10:10:12 - 01:10:23:15]
Speaker 1
 But it just allows you to get a little bit more of that objective, like step outside of yourself to get a better understanding of folks who, you know, do mirror your target audience, which isn't always easy to do.

[01:10:23:15 - 01:11:02:18]
Speaker 23
 For sure. Especially as marketers, I think you spend so much time trying to understand the industry, trying to understand the products that you're trying to sell and market to people. Sometimes it's almost like getting the blinders on because you think, well, I know everything there is to know about this. I know it inside and outside, but sometimes we forget that the consumers don't necessarily have that same viewpoint. They don't know everything there is to know about the product. They don't feel about it the same way that you feel. They don't have that same understanding. So just trying to get in their shoes more and more every day, how can you do that? Right. And again, it's another way to try to get it closer to them.

[01:11:02:18 - 01:11:25:23]
Speaker 1
 Well, right. And I think if you've worked in the same industry or on the same product for a really long time, you do tend to think like, we tried that. It didn't work. But it's like, look at how much the world changes in a five year period of time. And like people's beliefs and attitudes change as well. It's that piece of it is like tapping into it that, you know, I think is so, so awesome. Now, in terms of other use cases, like what are the ones that you are excited about?

[01:11:25:23 - 01:11:44:10]
Speaker 23
 There's a lot. I think being able to put creative in front of someone and be able to, you know, synthetic a synthetic someone, right? Yeah. And be able to get feedback about the style, the language to see all of those types of things. I think the

[01:11:45:16 - 01:13:10:20]
Speaker 23
 product features and roadmap development is another really interesting area to say, okay, well, this is how we want to build something out based on, you know, you usually build like a product roadmap and a backlog of things that you want to tackle and features and prioritize them based on what you think is going to deliver the most value to an end user. But sometimes that's also countered with, well, what does the business want to do? So this is another way to just, again, get another gut check on that or a validation to say, are we going in the right order? Is this going to, you know, if we had to rank these 10 things from a consumer perspective or from a client perspective, what's that going to look like in their eyes? And it can help give you just a bit of a different mind shift. So I love that there's so many different use cases for this and that it allows you to really be always on just kind of, well, let me just, you know, phone a friend, let me ask real quick and get that quick feedback. And at the same time, if you had something larger that you wanted to go after and you said, well, you know, this is maybe how we've prioritized it, but we want to dig in a little bit more and really understand where the friction points are. You know, that's another place where you can do a deep dive and kind of get lots of different qualitative feedback to come back and sort of tease through. And maybe ideas would surface that you weren't even thinking about that you can then go validate with your real audience or test in the marketplace, but yeah, just different ways to find things.

[01:13:10:20 - 01:13:28:16]
Speaker 1
 100%. And I think like all of that, you know, if we're going to answer the question of like to what end, it's like it's all to hopefully have more predictable outcomes, right? And better performance at the end of the day. Like that's what we're all here to be doing, at least in the marketing context. And I'm sure there are many other, you know, contexts in which clear and over could be used

[01:13:29:17 - 01:13:48:23]
Speaker 1
 now to play the role of the naysayer for a second. Like, what would you say the difference or the benefit is to like, you know, I can go to chat GPT and say, okay, I want to talk to somebody 55 plus when this home bracket, like try to put yourself in their mind. What's the difference there really in terms of

[01:13:50:10 - 01:13:51:07]
Speaker 1
 legitimacy?

[01:13:51:07 - 01:15:20:18]
Speaker 23
 Yeah. Well, yeah. I think in my opinion, it comes down to a lot of the nuance. How do you get to a very specific audience member? I think when LLMs are trained, they're trained on massive troves of data. How much can you feed into it to try to understand everything? And then what is getting spit back out of it? I think there's a little bit of, it's only as good as what you are, what you are kind of putting into it and what you're kind of asking about. So if you're just asking for something about a 55 plus person, and you've given just a few details, you're going to get, I think, a blend of all of the potential things there that relate to this. You might get some things of best practices. You might get some things of something that's been done in the past. It might be loosely related to that category, but maybe it's not an individual. It's almost like if you took a blended result and said, "Well, what's the average of this?" Right. And I think that's sometimes where you might get that kind of averagey sort of feedback. Where I think this is different is that you're asking unique people with unique backgrounds, specific information, and then that's going to get crafted. You're going to get a different output based on the personality of who you're talking to, based on the specific experiences they've had or the attitudes that they hold specifically versus here's what an average of what the 55-year-olds think about this area or where they lean or a percentage of that.

[01:15:20:18 - 01:15:31:17]
Speaker 1
 Because the Clarinova folks, I don't want to misquote here, but they have something like 150 attributes per individual or something. There's a lot that gets fed into each of those.

[01:15:31:17 - 01:15:35:04]
Speaker 23
 There's a ton. Yeah. There's over 700 different attributes.

[01:15:35:04 - 01:15:38:17]
Speaker 1
 Okay. I spoke apparently. It's very fun. You're like, "Go four times higher." It's a lot.

[01:15:38:17 - 01:16:27:15]
Speaker 23
 And it's something that's ever-changing to the way that we're comprising these audiences and building these synthetic folks out, finding more data points and finding the right data points to pull in in different areas. At any given point, it may be a little bit more, it may be a little bit less. But I do think that it's that diversity of data points that really helps drive different opinions coming out of these synthetic audiences, these synthetic people you're talking to, which is why this is also more qualitative in nature, is that you're really able to get different pieces of feedback where you get these nuanced opinions, these variations where you're getting the little tiny bits that you maybe wouldn't get out of something where you were doing a large

[01:16:28:21 - 01:16:30:23]
Speaker 23
 study or something where it's an

[01:16:32:02 - 01:16:52:22]
Speaker 23
 average of feedback. You can get these little tiny bits that when you put all those things together, that's what makes qualitative research interesting and gives you those things you can latch onto that drive strategy, that can drive those different insights that you'll get from people as individuals versus an average of the larger group.

[01:16:52:22 - 01:16:56:21]
Speaker 1
 I think that makes complete sense. Absolutely. It's a very good point.

[01:16:58:08 - 01:17:14:12]
Speaker 1
 Even just talking about the LLMs in general, I think one of the very cool things that you built into the platform itself, which also because of your experience is very easy to use and has a great UI and all of those things, it's like what you would come to expect from LLMs.

[01:17:15:16 - 01:17:22:19]
Speaker 1
 You've also built it so that you can select. If you have a favorite, if you're a cloud person, you can use that, for example.

[01:17:23:19 - 01:17:26:08]
Speaker 1
 Are you feeling like that

[01:17:27:08 - 01:17:34:23]
Speaker 1
 is going to help with that nuance as well or is that more so just the personal preference of an individual who's using the tool?

[01:17:34:23 - 01:17:45:23]
Speaker 23
 I think it's a little bit more personal preference in that area of as we explore LLMs in different AI tools. People are going to gravitate towards and be more comfortable with different

[01:17:47:00 - 01:17:55:02]
Speaker 23
 things that they've used in the past. Some folks are team open AI, some are team anthropic and every flavor in between.

[01:17:56:04 - 01:18:16:10]
Speaker 23
 Having that flexibility to be able to try different models on the back end that are powering this, again, at the end of the day, I think we're getting to a place where they're all getting so good and they're all getting so rich that they're not that different necessarily in nature, but maybe you have more confidence or

[01:18:17:11 - 01:18:22:17]
Speaker 23
 you've felt the results are better from one versus another. You can choose that flavor.

[01:18:22:17 - 01:18:42:19]
Speaker 1
 Even formatting, the way that Claude feeds you something is different than chat. I think people are starting to recognize that style of formatting is clearly from Claude and things like that. If you've got preferences that way as well. I think it's nice that you can switch it up and then you could even compare across LLMs or across the different audiences that you're doing.

[01:18:42:19 - 01:19:22:14]
Speaker 23
 As tools evolve, we wanted to be flexible to say, if a new model comes out that's really so much leaps and bounds ahead of another one, maybe we want to swap it in. Or, "Hey, our company policy is we can only use this provider, so let's pull that in." Or, "Hey, we don't trust that what's really going to some of these LLMs through an API is really going to stay private. Can we do something more on device?" We have options there too, where it's really a localized model that doesn't go to one of the big frontier organizations to give you a little bit more peace of mind that this isn't going to be used for training.

[01:19:22:14 - 01:19:51:13]
Speaker 1
 No, absolutely. We just had that conversation this morning. I think on the client level that often happens, they'll pick one path and say, "We know that we're comfortable with this, and so we're going to use co-pilot or we're going to use Claude," whatever. In a situation with an agency, we have those same concerns and we work in highly regulated industries primarily. That's a big thing for us. If we were to ever be asking specific questions, we want to make sure that that's very secure. I love that that's an option too.

[01:19:52:17 - 01:20:02:18]
Speaker 1
 Everybody says in the regular models, if you will, that they're not going to train on it, but it's nice to have the additional peace of mind that you're in a closed loop system as well.

[01:20:02:18 - 01:20:16:10]
Speaker 23
 Check all the boxes you can with any of the organizations that we are working with. Again, if you're using OpenAI, you should be able to trust that it is, but if you have that little bit of skepticism and you want to just double secret verify, make sure that it's not, there's an option there too.

[01:20:16:10 - 01:20:18:06]
Speaker 1
 Then you're going to use your burner laptop.

[01:20:20:16 - 01:20:27:20]
Speaker 1
 Awesome. Well, in terms of what's on the roadmap, because we're just at the very beginning of this thing, what are you most excited about?

[01:20:29:23 - 01:20:31:01]
Speaker 23
 That's a great question.

[01:20:32:23 - 01:20:43:15]
Speaker 23
 I'm excited to see how people are using it and get some of that feedback on where they want to take it. I think I have a lot of thoughts and ideas in my head of where this can go to getting into

[01:20:44:19 - 01:21:01:16]
Speaker 23
 building off of the qualitative and getting into more of the quantitative larger scale to be able to do things in that nature. I think different mediums. We've been making sure that you can do things like test creative, test messaging, test imagery,

[01:21:02:18 - 01:21:12:03]
Speaker 23
 want to be doing some video, maybe audio as well. There's some different nuances to getting into some of those different types of mediums and being able to get feedback on those.

[01:21:14:01 - 01:21:15:20]
Speaker 23
 I think it's so

[01:21:17:12 - 01:21:32:05]
Speaker 23
 early in the development, I guess, that it's fun and exciting that there's lots of different places that this could go. I like the idea of, again, being able to be involved or in every part of the process to help in all of those different areas.

[01:21:33:18 - 01:21:34:16]
Speaker 23
 Speed.

[01:21:35:18 - 01:22:30:15]
Speaker 23
 How do we make it faster to get to insights, to get to output? How do we tailor this for specific audiences or users? This isn't just a platform for researchers. It can be used by folks that are doing creative development, people that are doing product development, people that are executives trying to figure out direction that they want to go. There's lots of different opportunities, but how do we continue to tailor to those different groups and to their needs specifically? How do we do things that are speeding that process along and catering better to what they're trying to accomplish so that it just continues to be more and more of a useful tool? How do we integrate maybe to other existing AI tools that they're using? Could we expose this as maybe an MCP server? If you want to stick in the cloud interface, but you want to tap into the Clarinova audiences, maybe that's an option down the road. Totally.

[01:22:32:00 - 01:22:33:01]
Speaker 1
 No, I'm fine.

[01:22:33:01 - 01:22:33:19]
Speaker 23
 Yeah, but there's

[01:22:35:04 - 01:22:38:08]
Speaker 23
 lots of different ways that it's going to continue to move forward.

[01:22:38:08 - 01:22:51:00]
Speaker 1
 No, it's very, very exciting. I think the sky's the limit. I think you're right. Some feedback will probably drive the thing. The best products have that involved. We know that as customer experience professionals.

[01:22:52:06 - 01:22:56:21]
Speaker 1
 Yeah, any last thoughts you want to share or info that you would like people to know about?

[01:22:56:21 - 01:23:13:16]
Speaker 23
 Website, you can find it at clarinova.ai. Great place to go and visit. Check out some how to use it, what kind of things we offer, different packages. There's plans for everybody if you want to get started right away or if you want to do something a little bit more custom.

[01:23:14:20 - 01:23:27:17]
Speaker 23
 Again, we're always looking for different features and ideas. If somebody's got something in mind that they're like, "Oh, it'd be great if it could be something that would help with this specific area." Love to hear that feedback too. Awesome.

[01:23:27:17 - 01:23:33:13]
Speaker 1
 Yeah, and for any Flynn clients that are interested, they can come have Flynn sort of shepherd you through that process or

[01:23:33:13 - 01:23:39:03]
 (Music)