Co-founder of Flashpoint.ai 

 

620. Christian Derosiers

SPEAKERS

Christian Derosiers, Will Bachman

 

Will Bachman  00:02

Hello and welcome to Unleashed. I’m your host. Will Bachman, if you visit umbrexdotcom/unleashed you can sign up for a newsletter where you’ll get the show notes, the transcripts and some bonus material after I publish every episode, let you know about it. Okay, so today I am happy to be here with Christian de Rosier, who is a co founder of flashpoint.ai. A tool that can help you with different aspects of market research. Christian, welcome to the show. Hi. Thanks for having me will So, Christian, why don’t you start by just giving me an overview of the tool that you’ve built,

 

Christian Derosiers  00:44

sure so flashpoint.ai is an AI native market research agency. What that means is we encompass all of the tools and methods that you’re typically accustomed to getting when you go hire a market research agency to execute research for you, but at a fraction of the time and cost. So it’s kind of fulsome market research across surveys, media monitoring, expert calls and more, including some proprietary tools, but at a speed and cost that you’re more accustomed to getting out of tools like Survey Monkey.

 

Will Bachman  01:24

Okay, great. So walk me through. And as you know, our audience here is mostly independent management consultants, so people who would be, you know, in some leveraging these tools like Survey Monkey for doing surveys and so forth. So walk us through maybe a typical use case. Perhaps you’d like to show me your screen, and we can do a bit of a live demo, so you can walk me through some of the you know, the tools and how they work, sure,

 

Christian Derosiers  01:55

so I’ll share my screen in a moment. But basically, we’ve designed this tool so that for business leaders, so that the business leader can or the consultant, or whatever other synonym of that you would like, so that you can live at the level of business problem. Our view is that it’s a bit of a hyperbole, but no one really wants to do a survey. They have some other business objective that’s upstream of that, and they’ve translated that into a specific set of questions and a selected audience to go get insight against that question. And so we’re building a tool where you can live at the high level business problem. You want to develop a new product, you want to enter a new market, you want to do a customer segmentation, you want to think about pricing, and you can live at that level. We will help you clarify your objective and your research plan. Then we will translate that into a coordinated set of tools against that objective. So we’ll build you a survey, we’ll set up some media monitoring, we’ll set up some expert network calls. Some of them are purely automated and digital. Some of them, we rely on our industry partners, and you can kind of view each of the tools we set up for you. You can edit them, you can add your own and you can use them on an a la carte basis. So we’ll walk you through end to end the research process, from clarifying your objective to recommending and setting up tools for you, which you can then edit if you want, or just accept the recommendations, you can check out and pay on demand. And then we have a whole reporting and presentation generating generation component, where, after you get your results, you can easily analyze them, find the insight and create something for the client,

 

Will Bachman  03:58

and share a lot about your con, your your company, you have raised at least one round of funding. I believe,

 

Christian Derosiers  04:03

correct. We raised our seed round a few months ago. Congratulations.

 

Will Bachman  04:07

Thank you. So, yeah. So why don’t you, let’s, why don’t you share your screen here we’re, we’re going to do an audio only episode, but I’ll narrate what I’m seeing. Cool.

 

Christian Derosiers  04:19

So we’re on. Do you want to narrate? Or do you want me to

 

Will Bachman  04:23

well, I’ll narrate. And then you tell me what I’m so what I’m first seeing the homepage here is really the tool itself. So it’s not like a sales kind of page. It just, you just jump right into it looks a little bit like a, like a chat type of screen. It says, Flashpoint, ai, ai, market research agency. And then there’s a text box for me to type in my request, you know, so and then to hit, to hit enter. On the left, we have some buttons, like a home, a plus, and a folder to file stuff notifications. But really, we’re just launching right into starting with a, you know, a. In a query. Cool.

 

Christian Derosiers  05:02

So I’m just going to write a fairly generic research query help me understand the market trends related to lactose free ice cream in the United States. So you can write as much or as little detail as you want. It’s just taking some initial context to get you started. All right, so I’m going to hit Enter,

 

Will Bachman  05:35

okay, so it did some thinking, and then we have this AI web overview that it’s working to generate. We’re screen now the untitled project, and they have a kind of a the top there’s this, what do they call it? The breadcrumbs to go Launchpad, goes to action plan, goes to checkout, goes to reports. So right now, we’re on LaunchPad. So tell me what we’re seeing here on LaunchPad.

 

Christian Derosiers  06:06

So the Launchpad is where you clarify your research objective. So we have what the user wrote here at the top under objective. We give you a quick AI web overview. So here, right now, we’re relying on the perplexity API. So the purpose of this platform is to go do totally new research, but we want to tell you what’s going on right now. What can we tell you immediately about what you’re interested in? And so we go pull some information, which you can read for further context, and then below it, we have the actual research plan. So you can see it’s creating these different sections, different kind of threat, research threads, growth drivers for the industry, shifts in consumer preferences, competition, barriers to scale, I can edit any of these. I can add new sections for research. So this is just where I’m kind of framing up what I’m interested in. Does that make sense?

 

Will Bachman  07:10

It does, and maybe scroll up just a little bit. We’ll take the first one here. So there’s four or five of these sections. And then one section, for example, is growth drivers and seasonal demand patterns. And then under that, I’m seeing five sections. So one is test if health trends and lactose sensitivity are driving growth in lactose free ice cream. There’s either a little plus or a garbage can next to that, allowing, I guess, to expand on it or to delete it, add a new line or delete okay. And then you can also evaluate whether demand is seasonal or steady throughout the year. Another research would be, line would be, identify which regions or demographics are increasing adoption the fastest assess the influence of plant based and dairy alternative movements on category growth. So there’s looking at different growth drivers, is just in that first section. So it’s proposing these things, but at this point, it’s not telling us to do a survey or media monitoring or secondary research, just posing is breaking down the question. It’s structuring the question correct.

 

Christian Derosiers  08:17

It’s framing up the research and just identifying the things that are important. And again, you can edit this. So if you go and hire a human market researcher at the beginning, you’re just going to have a conversation. They’re going to ask for a bunch of context about what exactly you mean which aspects are important. And so this is just before you get into like, what tools you’re going to use to get insights? It’s just framing up what you’re actually trying to do.

 

Will Bachman  08:46

Okay, sure. So it’s helping to to frame up the question a bit more specific, specifically, correct. Okay, cool. So then what happens next?

 

Christian Derosiers  08:55

Well, before we go into the action plan, you’ll also see on the right here, we have a little AI assistant where it’ll talk you through it’s explaining what you’re looking at. It tells you some of the assumptions it made. It prompts you if you want to add more detail. So this is just an AI assistant for your Flashpoint project. Okay, so I’m just going to click Next. Let’s say I don’t want to edit this at all, all right. And so now it’s taking me to what we call the action plan. So this is where it’s translating the high level objective into a specific set of tools. And you can see here again, it’s saying, Okay, we we are set up this survey for you. We’ve set up this generative r, d, we’ll talk about that in a moment. And we’ve set up some expert interviews for you. And it’s explaining kind of what each is and how they all fit together. And you can see here it’s they’re being built. It’ll be done in probably two.

 

Will Bachman  10:00

Minutes. Okay, so this first one, let’s talk about that says consumer interviews on lactose free ice cream, and it puts a price tag of $1,813 Exactly. Boom. And it seems to suggest it’s going to gather data from 1000 consumers who purchase lactose free or dairy free products for 1800 bucks, pretty so

 

Christian Derosiers  10:23

you’re actually, it’s, we should fix this. So you’re mixing up. This is, that’s an interview. This is the interview. This is what’s describing the interview. So this is an expert call. The interview is an expert call. That survey you’re looking at, oh, it is actually, yeah, the survey here is you’re correct. I apologize,

 

Will Bachman  10:43

no, it’s updating it Okay, so you’re right. Actually, I so it’s the lactose free ice cream Market Trends Survey Yes, is listed as $1,875 okay, yes, and that is going to gather data from 1000 consumers who purchase lactose free or dairy food products. So it’s a Trends Survey. So it is going to survey 1000 people. Now I’m a look now this one that I imagine does not include the panel

 

Christian Derosiers  11:14

fee, right? That does include the panel fee. I mean, that’s

 

Will Bachman  11:19

only like a buck 80 per respondent, which seems to me quite low, because normally you have to pay the respondent something. So, yeah.

 

Christian Derosiers  11:29

So this is a, it’s recommending here, what researchers would call a Gen pop survey. Yeah, if you’re doing kind of general population average US consumer, it is actually quite low price. The more specialized the sample, the price can vary wildly. Yeah. So it just really depends on which audience you are targeting.

 

Will Bachman  11:51

Okay, so this is really, I mean, that’s, that’s quite cheap, but general population survey, so like coming up with a survey, putting it out to the field, collecting the responses, validating that the responses were actually not just sort of people going through and just marking AAA, but actually validating the responses are by real humans, and then paying everybody, and also validating that people are, you know, purchasers of lactose free products or dairy free products, all that for $1,800 seems, seems quite, quite a low price, pretty good. Yeah, go ahead. Sorry, yeah. I mean, that’s Wow. I had no idea that 1000 person survey was, was, was that? Was that low in price? And so, like, how many questions would that typically

 

Christian Derosiers  12:44

be? So we can built the survey itself. So I’m going to click into it, yeah, you can see here, we can even just immediately open the survey. So you can see here a little preview of the survey. Oh, yeah. You can see it in different views. I could even it immediately gives you this link, yeah, I could send this link to you to test and it you can see it here in what we call the text view. We have a few different ways to view the surveys. So this one is an 18 question survey,

 

Will Bachman  13:18

okay, and where would it check to see if people have actually purchased dairy free products in the past. Would that be part of the screening question that we’re not seeing?

 

Christian Derosiers  13:27

Or that happens in two places, primarily the panel partners. So we work with a lot of the leading panel suppliers to provide the survey respondents. They have their own pre screening capabilities, and then, so there’s some screening that happens on the front end. And then, depending on your budget and the length of the survey, you could have your own screening section. So this tried to keep the cost pretty low and keep a short survey, but you could, I could manually add my own screening section, or I could ask the AI to add a screener.

 

Will Bachman  14:05

Okay, so, and is that, wow, that’s typical, huh? Like, it’s $1.80 per response for a pop Gen pop survey.

 

Christian Derosiers  14:14

Wow, that is, I mean, that’s as cheap as they get, basically,

 

Will Bachman  14:17

yeah, you’re wondering, like, who’s filling these things out, like, right on Facebook to get, like, a free avatar or something, you know, silly card something or other free, I mean, wow,

 

Christian Derosiers  14:30

yeah, so that’s actually a great kind of thing to hit on. Here are, there’s what you could broadly call selection bias with any particular tool, surveys have some of there’s the issues you raise, which is, there’s basically adverse selection in survey responses, often where people are just responding for the incentive. There’s often even a difference between stated and revealed. Perhaps. Differences. So I might if you ask me, what I pay twice as much for ice cream from an ethical dairy farm, I might say the answer. I might say I would. But when the reality is that when I’m actually going to buy I don’t care, and there’s other kinds of data quality issues and selection bias and surveys, similarly with interviews, and we also, we can get into our generative r, d tool. And part,

 

Will Bachman  15:25

let’s talk about that, yeah, so just to close out the survey piece, so this tool would allow you to say, okay, yeah, you know, maybe adjust the survey questions a little bit, but you could just click a button and without going to a panel provider and programming it in SurveyMonkey, your tool would allow you to design the survey field the survey, get the results within all within Flashpoint correct.

 

Christian Derosiers  15:48

And even further, if you can see the interface here, we’ve built it to look like Google Docs. Yes, you can just type into this, yeah, the way you would type into Google Docs, and it will build the survey, sure to need to learn any special interface, right?

 

Will Bachman  16:03

So you could build a survey. You can, you can send this, you know, send it to a panel. You can get responses back. Let’s talk about your generative R and D, but that’s kind of interesting. Let’s talk

 

Christian Derosiers  16:17

about that, sure. Just to close out my thought or my kind of point from earlier are the reason why you engage in agency is because they’re going to use multiple methods to help you triangulate an insight, because there are kind of selection bias issues with any particular tool. And so rather than try and wave our hands and say, Oh, we we norm for things, or we eliminate certain kind of structural issues with individual tools. We kind of lean into it and say, we’re going to give you the multimodal, triangulated insight, rather than try, and, I think maybe dishonestly, try and argue that we’ve eliminated some structural issues with any individual tool. All right,

 

Will Bachman  17:04

let’s talk about the generator. So

 

Christian Derosiers  17:07

generative R and D we it’s patent pending. It’s a proprietary tool that you can think of like in market conjoint analysis, if you’re familiar with that. So let’s take a given question like, how do lactose free people choose ice cream?

 

Will Bachman  17:30

Well, can I just read what it says there? Yeah, because that spells it out to me. It says, conduct generative R and D testing by running ads to evaluate the interest in demand for lactose free ice cream in various markets. So it sounds like what this does is it’s it actually would create, like a Facebook ad, run it on Facebook, in real life, and then see how many people click on it, right? Is that kind of the idea behind this?

 

Christian Derosiers  17:54

Yeah, that’s basically the idea. So it’s right now it runs on Google ads, so in conjoint analysis, where you’re trying to identify the decision drivers. Similarly, here, it generates an ad for each potential decision driver. We can walk through it, and then it deploys them as real ads on Google ads, you can target them like Google ads, and each ad takes you to an AI generated landing page about the topic of the ad, and so you can observe kind of which ads get response what people do on the pages. And rather than framing it as like a fictional we’re not tricking the consumer and saying, like advertising a fake thing, but rather, we’re building and advertising guides to making the decision. So you’ll see an ad for like, here are guides to finding great, great tasting, lactose free ice cream, and it’ll take you to a page about that. And we can observe what real people really do, rather than what kind of survey respondents say they

 

Will Bachman  18:54

would do. Okay, can you show me, maybe an example? Either? Yeah, I’m

 

Christian Derosiers  18:58

going to click into this. We can see this. So you can see here, it built it. We’re in a, what we call a Content Review. So it has a bunch of value props, so like different decision drivers. Is this making sense so far? Yeah, I get it. Yeah. And then you can add more if you want. You can delete any of these. And then for let’s take this one, dairy free delight with real vanilla beans. I can click here and see what the actual ad would look like, and I can edit any of this, and I can also see what the landing page will look like, and I can edit any of this, including links to real things. You can see here it has a link. This is a real lactose free vanilla ice cream. And we can also track, like, what links people ultimately go to.

 

Will Bachman  19:51

Okay, yeah, that’s wild. So it just like created a whole web page, just instantly right there for. Correct, that’s wild, which would actually be on a real like, your real URL would be, like a real correct, be a real website that

 

Christian Derosiers  20:09

you could correct. We have a web property that kind of acts as a kind of got consumer guide or purchasing guide platform, and so it is. It’s aI generated, but it’s real information. So it’s will really generate a page to help you find, in this case, lactose free, ice cream that kind of tastes like real vanilla. And again, it includes real it’ll give you real information and links to real things you can actually buy related to that thing, that topic.

 

Will Bachman  20:40

So this is quite interesting. I’ve heard of Oh e commerce people testing products. They’ll do like, I think they call it dry checkout. They’ll have a real web page with real products, and you can click to add it to your shopping part. You get all the way to the shopping cart, you click Buy, and then says, Oh yeah, sorry. You know it’s not available or something, but because they’re just testing it, that product doesn’t really exist. But this is pretty neat. This instantly, just creating this page, what have what have clients of yours learned from doing this process? Can you give us some sanitized examples of the types of things people check and what sorts of insights you get from actually running ads?

 

Christian Derosiers  21:21

So it’s pretty good for a few things, some of which you’ll become increasingly clear as we proceed through this flow here. For generative R and D, it’s really good for concept testing, new product development, kind of in disruptive innovation, because it allows you you’re not attaching your real brand to it, and it allows you to test concepts and test things very widely that you might not want to attach a real brand to. So we did an engagement with a small liberal arts school. And they all kind of, as someone who went to a small liberal arts school, they all kind of say the same thing. Their pitch is all the same. We have small class sizes, and it’s all the advantages of a big university, but it’s special because it’s small. Again, I say that with love as someone a happy graduate of a liberal arts school, and they wanted to test different potential avenues they could take to market the school, but you can’t go really. You’re kind of stuck with incrementality if you’re attaching things to your real brand. And so we did a really kind of wide ranging test, and you’re going to laugh, and it is extreme, but also the point of this tool is to allow you to be extreme. So we tested all sorts of things from different permutations of special study abroad offers, oh, study abroad with financial aid study abroad or esoteric industries like disaster management plus liberal arts, some like very practical things with liberal arts degrees all the way to we tested ads for anti woke liberal arts schools. Now that’s of course, it gets a laugh every time, and it is funny, but it’s also kind of interesting and powerful to be able to, like, allow you to get sick, like, find the Signal and the Noise. And in fact, in that example, that ad did get the highest response, which allowed them, we did it very quickly and cheaply, which is some of the advantages of generative R and D, but it also gave them some insight into some latent demand that they then have taken on to Think about future planning for the college. Huh.

 

Will Bachman  23:40

Okay, so this lets you test ideas very quickly. And I think the price on this piece, for this particular project, was around $4,500 what like, how many, how many ad impressions would you get for that? And like, how many clicks would showing 5000 impressions, and 500 clicks is roughly what you’d be looking at, right?

 

Christian Derosiers  24:08

And you can alter these so if you say, oh, I want more clicks, and the price will just update dynamically. So you can see here, I increased it the clicks, and you can see the price. It’s taking a minute to update, okay?

 

Will Bachman  24:25

And then I guess there’s like, eyeball track, not eyeball, but there’s user tracking, user experience tracking that you can do on a website where you actually track where someone’s cursor is and where they click and stuff. Are you doing that to see how people explore these web pages, what they

 

Christian Derosiers  24:43

not yet, but we that’s coming currently. We just see which ads get responses which sections of the landing page you go to, and then we can see what links you ultimately we put real links at the bottom, and we can see which links you go to. Yeah. Another big advantage of this, and we’re just going to proceed to the next step here is because they’re ads, they’re highly targetable. You have all of the tools that you get with digital ads, gender, age, income, what device they’re using, we’re going to have interest interest areas and other demographic characteristics, and crucially, you can do geographic targeting. So this, in this case, it set it up for the US. But if I wanted to, I could choose, like zip codes, for example, and I quickly do highly geo targeted research on people and a specific zip code.

 

Will Bachman  25:47

Oh, lactose for ice cream in the East Village in Manhattan, all right.

 

Christian Derosiers  25:51

And so Christian, sorry, another point go ahead.

 

Will Bachman  25:55

So just, just, I see that we’re kind of getting towards the top of our hour here, just tell us just about where people can find the tool, and also, what’s the what’s the cost per user?

 

Christian Derosiers  26:13

Sure, just super quickly on Gen R and D, because we’re using ad targeting, it also opens up emerging markets, so I could go find out how people choose car insurance and Kabul, for example, super quickly. So you can find us on flashpoint.ai. The base fee for a license is $40 a month per user billed annually. So it’s $480 a year. And then the incremental costs come from primarily the sample, how many ads you’re running, how many expert calls you’re doing, and those are taken a la carte with each project. And you can, of course, modify the parameters, so you’re going to pay $480 a year. And then if you want to run kind of simple Gen pop surveys. It’s going to be very cheap each project you do. If you’re doing more specialized things, well, price will go up. And we, rather than lock you into big annual commitments, you can just kind of purchase the resources you need, a la carte for each project.

 

Will Bachman  27:19

We didn’t have a chance to get into it. But do you have any sample deliverables that you could send me that then we could make available to listeners like, sort of, here’s an example survey output. Here’s an example output from the Gen R and D. Is that something that you’d have on hand?

 

Christian Derosiers  27:39

Yeah, so I can provide some resources, either as a PDF, or I can make a quick video that we could attach to this where I kind of walk through some of the results, both for the individual tools, and then also the report and the data analysis and reporting features.

 

Will Bachman  27:55

Yeah, just a PDF would be great. And then listeners, if you go to umbrexdotcom/unleashed you can find this episode, and we’ll make it available so you can download those resources from that page. And for those of you who are subscribers to the newsletter, we’ll also make it a link in the newsletter. When we talk about this episode, we’ll make it so you can grab those, those PDFs. All right. Christian, well, thanks very much for joining. We will include those links in the show notes and appreciate your time.

 

Christian Derosiers  28:27

Cool, likewise, thank you so much. You.