Sebastian (00:02):

You are listening to the Insightful Connections podcast. Our guest today is Marin Mrsa. Marin is the founder and CEO of Peekator, founded in 2018. Peekator is a quantitative survey platform designed to help companies conduct efficient primary research, allowing them to focus on extracting insights. Peekator specializes in providing high quality panel data backed by a money back guarantee and offers innovative AI solutions like the recently introduced conversational ai.

Marin and I were having a, a little conversation before we got started and prior to starting Peekator, Marin was an accountant in the hospitality industry, which I think is gonna be a really interesting story to unpack. And uh, you know, I've been following Marin on LinkedIn for a little while. He does a really cool thing that I think more founders should do in his post, sort of timestamping how far out he is from the launch of his venture. I have that clock running in my head as well as a fellow founder. Marin, thanks so much for being on the episode today.

Marin (00:53):

Thank you so much for the kind words and also invitation. It's a great honor to be here.

Sebastian (01:00):

So my first question, and I like to get this sort of personal context from everybody, is how did you get into market research originally and how does that help explain where you are today?

Marin (01:32):

So the story starts, it's actually this Sunday will be seven years when I, uh, decided to quit my first and only job before this. So seven years ago I was working at a hotel in a small coast town in Croatia. It was a a, a really beautiful town and a really nice job. I was an accountant for two and a half years there. I really loved my job, but that was my first and only job working for someone else. Uh, back at that time I didn't, I actually wasn't aware of a concept that you can own a company of your own. You know, like, uh, our city was really small and usually you wouldn't see a lot of examples that someone started a company on their own yet alone, that someone would start a company inside market research space. I wasn't aware that there is this whole industry and thousands of people who work in research and stuff like that, but I really loved two things from that first job.

Marin (02:32):

And the first thing was I really loved software. So that was my big love. I was a software guy in that hotel. When my colleagues had a problem with any type of software that they use, they would come to me. I was a power wizard of every sort software. Really loved to finding the small spots, tips and tricks. So I really love softwares. And the second thing, I really loved the a, the <inaudible>. I of course worked as a accountant and prior to that I worked as an also auditor. So my day-to-day job was around the ata. And then at one point, my girlfriend at that time and my wife, she said to me, you are working so hard for them. Why don't you consider starting a company of yours? And then like the click and the bulb turned on in my head I was like, oh, so I can start a company of my own.

Marin (03:28):

And I started to thinking about the first idea and, and the idea that I wanted to do as I worked in hospitality. Of course my first idea was close to that. So the first idea which I had seven years ago was, if I am a owner of a restaurant or bar in my hometown, how can I check the level of service there I was all above the standards and stuff like that. I wanted to assure that my restaurant would follow the standards when I'm not there. So how can I do that? We used mystery shoppers back at that time in our hotel, but that was a really long process from when they come, when they announce it, when they come, when they send a report and stuff like that. I told that with tech and softwares that we could do a much faster and better job And uh, back at that time in correlation, Uber was a new thing and that was just a press a button and someone would come and pick you up and stuff like that.

Marin (04:27):

So then I combined Uber concept with my idea and I had approached, let's try to build an app so that bar owner can call a mystery shopper, throw out an app at that, a mystery shopper can come inside same day and check the service for that. So that was my idea. I felt that, uh, that's a cool idea. That's what yeah. But to be fair, as I wasn't aware too much of the market and stuff like that, it didn't turn out to be a huge success <laugh>, stuff like that. So that was my first idea. I quit my job and actually moved from shipping to Z Zagreb and Zab is the capital of Croatia. And I started a company. A problem was that even I was a big fan of softwares and data. I didn't have any experience before in building the software or building a research company, you know, like so that I, I had to do a lot of self-learning and stuff like that.

Marin (05:28):

So that was a, a really big leap and I wasn't aware how big it was until couple of years after that. I know that all of my friends said that I'm crazy of leaving a, a great corporate job like that for I think like this, which was some crazy idea that nobody was sure is it going to work. And they were actually right because after six months we didn't have, we actually built the app and the app was there and we had some pilots with bars and restaurants, but it didn't achieve too much. So that was like six, eight months into it. And then I actually was also active on LinkedIn back then and one of my posts triggered a bigger corporation which sold what are we doing? And then they approached me and asked, can you tailor the software to us? And they were a big retail shop and after them one bank and that's where we achieved our big first client.

Marin (06:30):

And that was so just that if I said to myself was a really big success because for a 26 years old guy who doesn't have any experience before, to work with the biggest Croatian bank in under a year was a really big thing because in Croatia there is a mindset that you need to know someone in a bank like that in order to secure a deal. You know, like you cannot come with your own ideas and just land a deal so you need to know someone or either or you have a cousin there or a friend or your dad is a CEO or something like that. So just landing that first client was a really, really big thing. So we tailored the first that app for mystery shoppers and we expanded it and tailored more for a bigger bank and that was our first corporate client. And it all started with that one year after I quit my job.

Sebastian (07:25):

Zero to one. Yeah,

Marin (07:27):

Zero to one in one year

Sebastian (07:28):

<laugh>. Yeah. I'm really curious 'cause I actually love the idea of Uber for mystery shoppers. What were the barriers to adoption and how did that end up pivoting into what is now A-A-D-I-Y quant platform?

Marin (07:40):

So the barriers, I, you know, like the, at least in Croatia, what we found out at the end of the day that owners of restaurants weren't that much concerned about the standards of their service. They were much more concerned how to retain the staff because the biggest problem they had was finding new staff and how to achieve that they stay and it's still the same issue today. You have the highest turn of employees in hospitality. So that was the biggest problem they had. How, why am I going to achieve that? My staff is going to stay here for the next year. So they were much more concerned about that, not about the standards and stuff like that. Of course there were a bigger restaurants and bars which cared about that, but that wasn't enough to have a decent economic sense for us.

Sebastian (08:36):

And how did the, that first deal that you guys landed really guide the evolution of your tech from sort of this seemingly more qualitative focused platform to now what is A-A-D-I-Y quant platform?

Marin (08:50):

So our first client was a big bank and they used our tech for mystery shopping. But if you look at it, the tech is similar. Mystery shopping is more qual of course than the quant, which we do now. But if you look at it still basically a form underneath the structure, you know, so you have a question and answer. So that was the baseline that we did for mystery shopping. And then of course we in the next year after that, we landed actually a much bigger client than that one and that was the biggest retail shop in Croatia. And they wanted us to interview their customers on exits of their shop. We expanded the software for that. Uh, so that was the first part where we had real service after mystery shopping. And then of course year after that we had a client who asked, okay, I don't want to go to uh, physical stores I want online.

Marin (09:49):

And then we built the online part for them and then we formed our own online panel as well. So our first two, three years were mainly dream by our clients and basically because I didn't have much experience in this, you know, like, so we really relied on our clients and we really partnered with them and I think that's what they loved about us. We weren't so pushy to them that we would enforce our ideas. We really went to the meetings to them and tried to listen to their problems and what are the next research projects they want to do And then it would go back and build the tech that they need. So those were like the first 2, 3, 4 years of our work back in <inaudible>.

Sebastian (10:31):

Cool. Yeah, that's so interesting. I think early clients can play such a pivotal role in helping you find, you know, your way to product market fit because they know what they need and if you're willing to follow them, they'll take you to to the real problems that they're experiencing and give you an opportunity to solve against them. So you obviously have A-A-D-I-Y platform now that sort of integrates both the survey building functions and online panel into sort of a single environment. What have you learned about who DIY tools are for and maybe who they're not for

Marin (11:02):

After those three, four years, we finally like built that capability of do it yourself where you have a panel inside and you have a survey tech. So you basically have everything what you need for doing online quant. And then it started pushing more of that because you are also aware the whole do it yourself platform tools was trendy three, four years ago. That was a really big trend and everyone was copying on it. And first one two year we figured out a couple of things that the do it yourself was a promised land for many clients it really looked amazing. You know, like you don't need agencies, you don't need, you know anyone, you just have this tool where you have a sample, you build your own survey, you launch it and you get the results. It sounds really simple and many install startups raise funds on it because they're going to simplify research and stuff like that.

Marin (11:58):

And it really sounds that it solves the problem. But in research space in day to day, there are many times where it just wouldn't fit the problem and do it yourself. So first of all, of course I think, and we have clients who used as you know, do it yourself tool. There are cases for which do it yourself tool is really good. So those are the ones that are let's say, simple research projects more quicker. One when you have like a standardized concept test or when you have a quick pool or something like that and those things work really well. But I would also conclude until recently because the data quality, uh, things are really bad lately and Decel tools were, were amazing until I would say last six months when data quality really went down. So yeah, Doell tools are really good for those ones, but durell tools are not the best.

Marin (12:53):

Where you have a bigger corporate clients who at the end of the day almost everyday research project is different than the one before. So they would always need to customize some things and some things are not easily done throughout a do it yourself tool. You need to build capabilities for that you have, but you cannot cover all the cases and stuff like that. So do it yourself tools and from what I see from colleagues and stuff like that, and often corporate clients didn't have the time, so do it yourself tool was brought as this dreamland tool which will cut the time for you as well because you are not going to need to wait for agencies panels and stuff like that. You are going to do it yourself, but it comes with expectation that you have time to do that. At Dolphin these, these people, they don't have time to <laugh> to set up a do self tool script inside of it, think about the survey, get the results, analyze it.

Marin (13:54):

So it actually requires a lot of time and skills to run a do self tool and nobody thought about that. They actually need time to think about the questionnaire, approve it with their stakeholders, stakeholders running back scripted, get the data. It takes a lot of time and many corporate clients don't have time for that. They want outputs, they want actions and that's why they value agencies and that's where do yourself tools come short because it, the promised land was we are going to give you a tool and you're going to run it but you don't have time for it. And then it just made you hate it because the agencies would usually script for you and send you the link and now you need to learn the tool. Every tool, every <inaudible> tool has its own rules. Education you need to learn how to do conditions by piping variables.

Marin (14:51):

All this stuff now you need to do that you weren't supposed to do before, before the digital cell tools. You know like, so it's not for everyone and especially lately when data quality really became a big issue we derive of AI and stuff like that. So these days for example, we reject more than 30 or 40% of the data that comes our way in a duty yourself setup. If you don't have someone who is monitoring the data, you need to do those checks on your own as well. So of course tools are going to do some data quality checks of course, but those are not enough. They do, let's say we call those respondent checks but they don't do specific survey checks for every survey. You need to do certain steps that you ensure the data quality is good if you use a do it yourself tool, in most of the cases nobody is doing those data quality checks.

Marin (15:49):

So now also you have to do the data quality checks, report the data back, wait until the new data comes in and then you can start with your job. So data quality now adds a lot of new things for you to do because if you don't clear the data quality, if you don't do the checks, your whole set is going to be really bad. You know like and the data quality is the essence and the first and the last thing our researchers should care about. You know like data quality is super important and everything starts and ends with it and the due to itself set up it's hard and it's only going to be harder.

Sebastian (16:27):

So you touched on data quality and I think that's a really hot topic right now and I definitely wanna get back into that. But I also wanted to ask you, so you know you've kind of drawn out a bit of a gap between sort of the promise of do it yourself tools and then what they're actually delivering for brands that may want to say we're done with agencies wanna just you know, plug in this tool and be done with it and get insights quickly and pay less and all of these things, right? I'm curious what PEEKATOR is doing to sort of bridge that gap. 'cause I kind of feel like you brought all this out and, and I'm assuming there's something there that <laugh>, you know, that's

Marin (17:00):

Nothing <laugh>. Yes, of course, but you know like I would say that most of the tools these days do this as well, but some of them don't. But I assume that they will start soon. So you have a do it yourself set up but you have do it with us set up as well. So do it with us set up where essentially you have insights team working with you to help you with the tool and in lot of cases it also means we are going to use our own tool in behalf of you. So there are a lot of uh, benefits to that approach. First of, first of all you get all the good stuff out of the tool and you get the speed as of course our in-house team is going to be like 10 times more faster than you. It now is the tool and we're going to set up the tool much more faster.

Marin (17:48):

So it's a do it with us approach where we work with you inside our own tool and that can be really cocreation as well because on certain projects you're going to do it yourself on certain projects we are going to do half of it, you do have. So it's a more called collaborative approach where we work together inside our own tool and that really from our own experience, it's the best approach. We call it more like extended support, you know like it's not support because in support you are only answered the questions but here you work with them inside the same tool and then you have the best of both worlds. You have a research agency support with the tool. So we help you with questionnaire advise you, but we also script with you, you script some part of it for one survey you are going to script.

Marin (18:37):

Then when we test you're going to go inside the tool and change some things. You know like, so it's really collaborative approach and most of the clients really love that because they feel they have extension of their own team inside us and they have a tool and you still get the speed, you get the tool and you get the support. So it's like best of both words what we call and we call it a hybrid approach or do it with us approach. And I would say a lot of doit yourself tools are actually doing that these days as well because they saw the friction from handing the tool like here you go, use the tool and leave us alone. They saw the friction in that and they know they can help with that. So a lot of them starting doing that as well. Here's

Sebastian (19:21):

The documentation you have time to read, you know?

Marin (19:23):

Yeah <laugh>. Yeah like here you go, like a hundred page like you read it and script it and send us the link.

Sebastian (19:29):

Yeah. Cool, cool. Okay, so you talked a little bit about data quality and I'm curious how do you think we got here with data quality and what do you think the implications are? And you mentioned a little earlier that data quality has always been something that you've been pretty passionate about because of your background in accounting. So I think, you know, it's an interesting question to ask with, you know with your accounting hat on, right? How do you think we got here and, and you know, what do you think the implications are and what are you guys doing about it as well?

Marin (19:54):

So you know, like I was an auditor before and I was that guy who would irritate everyone on the team because I would find small errors and then I would send it to correct it and stuff like that. So I always had an eye to spot mistakes, errors and stuff like that. I really love that and I carried out that approach to our uh, company. But data quality, the issue from my point of view starts way before and the problem starts with our end client and in most of the cases the research is done for a brand and client which uses research to have a decision and stuff like that. And we all know that end clients, they only care about the outputs, right? So they don't care about the sample too much, they don't care too much about methods sample, all this stuff that we researchers are passionate about, they like to know, but they don't ask too much.

Marin (20:51):

In most of the cases they assume that you're doing a really good job about it and they are only concerned about the outputs conclusions. Those 60 slides or 10 or a hundred, they are only concerned about those and how are they going to present those to their own shareholders. So I think the underlying problem was they always assumed that data that comes is legit, right? The data is really good. If they bought a sample of 600 people, they always assumed that data quality is really good. You know, like they shouldn't ask anything about it. They assume professionals are on the on the other side and they can do it And that of course and clients who care about this stuff, uh, but I would say that most of them just don't have time for it and they need to assume and they trust you on your brand.

Marin (21:45):

Of course they're not going to uh, start a project with a panic approach where they need to question you about every single thing. So they would assume the data quality is great and everyone will sending a sample. And then when you come to the middle part to the panels providers or agencies or tools like us or someone else, whoever provides a sample you usually would receive or request a quote, right? And in that quote, you have a set of things and in 99% of those requests, they're all the same. They have sample size, they have country, they have profile, they have objective, they have visibility, in some cases they have incident rates, all sorts of stuff. But there is not a single question or requirement about data quality. Nobody ever asks for it or from supplier side, nobody educates what are they doing about it.

Marin (22:44):

And then you have the famous CPI or cost or cost, you know, like the famous CPI and that can rage from 0.5 to 1 5, 6, $10 and often you don't know what's the difference between the ranges, you know, like if they all sell the same, what's the difference between those? So I think the problem which we had that just to get back to it, and clients assume that data quality is really good, suppliers never educated too much about it and then the bad suppliers came and they see that they have a lot of space for not the best practices and of course that you can save a lot of money if you don't care about data quality. And then they would compete on CPI or price and the data quality would just go down, down, down and down. And of course that you can't have a really good data quality if you, you charge the lowest CPI, of course that you can't, right?

Marin (23:49):

So just the whole environment, how data quality is shut aside, it's a public secret. So I really believe that data quality is a public secret. Everyone now speaks about it as well, but you're also going to see when they speak about it, they always speak like really care about data quality, our latest stack. Uh, you know, like they always speak in really generic broad terms that you cannot really understand what those mean. They just say like, oh we take care of it, but how? How do you take care of it? Can you please explain? And of course today with AI you have all sorts of problems with it. Bots generated answers and ghost farms of course. So you on one side you have click farms bots and all the fraudulent ways how these click farms are trying to get inside the surveys. And usually those are taken care on the panel side, but then even when people qualify for panels, there are people that just want to earn a quick buck and they don't care about your survey.

Marin (24:53):

So you still need to take care of the survey part, not only about the respondent part, you need to take care of every single survey you do and every single survey is different. So you need to have all sorts of data quality check for every single survey. So what we did there, we saw that there is a really problem about, you know, education about uh, this stuff. So first of all, we educate every our client what exactly we do in data quality. So we have a 13 step approach which we share with them. It's transparent what exactly we do. We have a 13 steps, some of them are before, during, and after the survey is done, we share those with them and we say, if you see any of these traits in our data, we are going to give you money back. Data quality is a corner store.

Marin (25:43):

It's really important. So because our tech is not going to be valid if the data is not highest quality, you know, like who cares about the tech if the data is wrong? So we are going to put money where our mouth is, you know, like we are going to bet on it. If you don't, at the end of the survey, if you feel that it's not very good quality there <inaudible>, we're going to give you money, money back. And so first of all, education of end clients, even those who don't have time, just speak with them, explain them how panel world works because many people and clients are not aware how panels work. You need to explain them some stuff. So I think as industry clients need to ask more because their outputs are going to be wrong if the data is bad and suppliers need to educate more because if you just once gamble your reputation on data, your clients are not going to trust you when they start to question your data. That's a black hole that it's really hard to get outta it.

Sebastian (26:49):

Marin, where can people go to find more about Peekator?

Marin (26:52):

Peekator.com is the best place to go and find more about us. But yeah, also I am uh, as you said, a really active on LinkedIn, so feel free to connect there. And yeah, those are the two base I would say. Awesome.

Sebastian (27:05):

Thank you so much for taking the time to chat with us

Marin (27:06):

Today. No worries. Thank you.