The Bootstrapped Founder

Arvid: 00:00

Hey, it's Arvid, and you're listening to the Bootstrap founder. Today, I'm talking to Pierre de Wolfe, the founder of Scrapingbee, and I'm fascinated with this business. It allows people to reliably get information from all kinds of websites, many of which don't really like people grabbing their valuable data. But in this industry, Pierre has built a trustworthy and more importantly trusted brand around this very technically challenging business. And we'll talk all about dealing with abusive customers, the impact of the AI revolution on content and scraping, and how to differentiate your business in an entrenched landscape.

Arvid: 00:44

Whatever Pierre and his team are doing, I think it's working really well and it's working well for me and my own scraping needs. And just like with scraping. Me, I'm a customer of paddle.com, the sponsor of this episode. Pierre will talk about the technical complexity of scraping and getting the right people to pay for your service easily and that's exactly what Paddle is great at. Taking on the nitty gritty details of charging money, so you can keep building your business.

Arvid: 01:08

So go out and check out paddle.com if you're looking for a payment provider that works for you. And now here's Pierre. Pierre, thanks so much for being on the show. You went from being a developer to running a company that processes billions of web requests and what I assume a month, a week, a day, must be a lot of web requests on that side. What was the first moment where you realized this whole scraping thing that you were building could actually be a real business?

Arvid: 01:38

And then maybe what made you bootstrap it despite there being a lot of people already in a pretty crowded space out there in the world of scraping?

Pierre: 01:47

Thank you, Aravid. Thank you so much for having me there. So glad to be here. That's a very good question. I think it all comes down to the previous product with Bootstrap, with scaling my co founder and lifelong friend, where we use actually a web scraping API.

Pierre: 02:04

We used a web scraping tool that we knew was somewhat successful, but it was honestly not that good in term of performance and speed. And so at that point, we knew, okay, there's something to be done in the watch reaping space. We also knew there were lots of very big companies out there. I'm talking 100,000,000 a year, companies. So we thought, okay, even if we manage to build something that captured, you know, a slight a small cramp of the market, there's something there.

Pierre: 02:38

You know? I mean, the market was there, the need was validated. Why did we bootstrap it? Because I think that's the only things we knew to do. We didn't want to raise money for all the reason people don't want to raise money for that were talked, I assume extensively on this podcast already, but then there is also a part of, we weren't sure we would be able to raise anyway.

Pierre: 03:12

We didn't have the connection, we didn't want to make a business plan, we didn't know if our idea would be interesting enough for people willing to invest it. So so, yeah, that would be the the answer to to this question.

Arvid: 03:26

Did you ever consider it as the business was growing? Like, did you ever think, okay, I could use like half a $1,000,000 right now?

Pierre: 03:33

So half a $1,000,000, definitely not, but we ended up raising a small amount with a tiny seed, not a boot striper friendly fund. We ended up not using any of the fund we raised, but it was awesome to first get into some kind of, so it's not really an incubator, but let's call it that way, for the connection, for the community, the crowd, the help, the mentoring out there, and then the money, even if you don't spend it, knowing that you have this kind of money really helps a lot, especially in the early days when you spend you know, when you count every euro, when you create tens of Ahref trial account, at $7 a week. I think they removed it, but I remember we used every email we could at that time. And when you just don't want to take any risk because then if you lose $200, €200, that's 10% of your bankroll vanished. And when you have 100,000 in the bank account, it definitely changed everything.

Arvid: 04:46

That's so cool. Like the bootstrappers mindset I that I hear, that's so palpable. Right? Like, you have a lot of money, but you're not even using it. Yeah.

Arvid: 04:53

Like, you're thinking, oh, that's that is so rare. In in the world of founders where everybody's just talking about, like, spending money on marketing, spending, spending, spending. To hear you saying, okay. I don't need this, That's such a fresh breath of air. It's really cool.

Pierre: 05:06

I think it all comes down to one big thing and it's not necessarily a good thing that good strappers, they don't want to hire, you know? And what's the biggest cost? What's the most expensive? And if you want to grow the team, to hire a developer, a very experienced one, or start building a support team and all. And so if you remove that, then just not that much money you can spend in a month, you know?

Pierre: 05:32

Like once you have changed your MacBook, when you are already spending 5, 10 ks a month on Google Ads, I mean, you know.

Arvid: 05:40

Yeah, that makes sense. Like if you remove a lot of expenses, obviously you're not going to spend as much, but I do wonder, like what's the team like right now? Because you have grown a team, right?

Pierre: 05:49

Yeah, so now the team is like 6 full time employee. So it's like, let's say, okay, so Kevin, co founder, he does all the up thing, marketing stuff. I'm doing the whole tech product side. We got Etienne with, yeah, doing development for basically a lots of things, the web app, a bit of ops stuff, a lot of things like, Swiss Army Knife kind of dev. We have 2 support guys, Saeed and Izar, 24 not 247, not what I meant, but full time on this.

Pierre: 06:32

And we have Peter, our head of SEO. So this is a core team, and we're also working with freelance. So for example, Etienne would be working with some DevOps people specialized in one thing. I'm working with a front end designer and developer integrator, and, Peter is probably the one working with the most freelance out there to manage the whole content of the website.

Arvid: 07:04

That sounds very lean, like very, very constricted. That's awesome. Like and it's also surprisingly small. I think you you tweeted about this recently that people of a week ago or so, everybody was asking you, how do you run an operation this big with just 6 people? Yeah.

Pierre: 07:20

Yeah. We went to a web scraping conference in Austin, organized by ZEITH, a big web scraping company based in Dublin. And, yeah, it feels like lots of people knew about scraping B, especially at this conference because it's a very small world And yeah, everyone was surprised how small the team is. I don't think it's I mean, you can be proud about it, of course, because you're like, yeah, people think we should be bigger, we're small, but in the end, it might not be the flex you think it is because it has a big cost to be very lean, especially for giving an eye in terms of energy, time spent and also quality. Like when you do a bit of everything, it's hard to be great at everything.

Pierre: 08:09

I mean, it's impossible. So I'm telling you this because that's something we want to fix this year, to grow the company a bit bigger, to to sacrifice a bit of profitability, to to to make the the company more smooth, you know.

Arvid: 08:27

Yeah. I have to ask, like, how profitable are we talking? Like, when people say, hey. 6 people is way too few people for how much you're making. How much are you making?

Pierre: 08:36

For various reason, we don't share how much we're making a year, but we're making several millions of profit per year.

Arvid: 08:42

Yeah. I I remember back in 2020 when you were posting your first tweet about, hey. We just hit 2.5 k MRR. Right? That's that was a different day, and that's so cool.

Pierre: 08:53

That was a different day. Definitely. And even at this time, I mean, when we were doing 2.5 ks, we were not paying us any money we were scaling. So of course we were profitable, but yeah. So yeah, we're pretty profitable and, and now we're ready to to sacrifice a bit of profitability to, yeah, to improve the operation.

Arvid: 09:14

Yeah. So I I do wonder because what I what I've heard you say is, hey, if you have something that works, why change it? Right? That's that's been, like, also something that you've been very publicly talking about. So what is it that you want to accomplish with a bigger team that you cannot accomplish with the team right now?

Pierre: 09:32

So I think we've come to the end of doing the same thing that works, you know. Like for example, let's talk about SEO. You know, we did a lot of SEO, we're still doing a lot of SEO, and it worked and it still work. But at some point, I don't know if you've, read the the post about the I think it's from Jason Cohen, you know, and it's talking about, like when you have an acquisition strategy that works, like the chart would look like an elephant, you know? Mhmm.

Pierre: 10:08

Yes. And SEO was the same for us, you know, first is slow then you're good, then you hit the plateau for various reasons. Let's call it churn, let's call it algorithm change, let's call it competition. And so the only way to find big growth again is to add a new chart on top of it. And so in this case, it would be a new acquisition channel.

Pierre: 10:32

So we feel like we're in a phase or maybe we did a bit too much of the same thing, and now we need to start experimenting other things.

Arvid: 10:41

That's very interesting. Yeah. I I guess. I mean, that's that's the the plateau that every SaaS founder is afraid of. Right?

Arvid: 10:49

Where where things stop working, but still kind of work. It's like always this balance between, well, what is the next thing? How how do you find the next thing? Like, what is your approach to getting to that next plateau hopefully?

Pierre: 11:02

So this is really exploratory. So for example, we this year we went to 2 conference, we went to the, web surfing 1 in Austin, we learned a lot of things. We also went to the Ahrefs, Ahrefs, one in Singapore about SEO and there we had lots of ID, lots of things to test. We can talk about, you know, sponsoring, short form video content, free ebooks, course, web scraping university, web scraping video long form channel. You know, there's so many things out there to do.

Pierre: 11:44

So how do we choose what we do? It's just a cost reward balance we try to find with Skagen, like, okay, if we want to try this, how long should it take before it works? How long should we give it before we scrap it? Because for example, YouTube, it's well known that YouTube is a very long game. You have people like Pat Walsh, for example, who created their channel this year.

Pierre: 12:15

I mean, they restarted it this year and they managed to have very high growth from 0. And then you have people who just created 500 videos in the last 3 years and it will still have 100 of you per video. So the goal is, yeah, to find the right balance between vast to extreme and to say, okay, if we don't hit those number at this time, we scrap it. And to to kind of find find the slope, you know, it doesn't matter if you're still if your numbers are still low after a few months, but you need to find a bit of, momentum. Mhmm.

Pierre: 12:56

You know?

Arvid: 12:57

Are are you gonna make this choice about, like, when to reevaluate it on a on a different time scale for every experiment? Or is this gonna be, like, hey. We're gonna give it a month for each of these, and then we're gonna see?

Pierre: 13:10

Usually, I think we're gonna give it at least 3 to 6 months. But then if if after 4 months, for example, you see that nothing's working either because you were not working with the correct person or because your ID was bad or because the execution is bad, we reevaluate things constantly with skin. So beforehand, we'd be like, okay, let's give it 6 months. But if after 3 months we see red flags everywhere, we don't have a hard time saying, okay. We should get things short here.

Pierre: 13:47

It's not working. It won't work.

Arvid: 13:49

Okay. That's good to know. And this is the marketing side of things and the sales side of things, which is interesting. I I wonder, do you have the same approach for feature development for product improvements? Because you recently released this really, really cool AI feature that I've been using, like, since before it was public.

Arvid: 14:05

I don't know. Like, I I I saw it. I I started using it, and you were like, hey. This is new. We're gonna launch it tomorrow.

Arvid: 14:12

But, yeah, it's it's, do you do you approach building new features the same way? Like, you give it some time and you check?

Pierre: 14:18

So for features, it's a bit different because since we have a lot of user, we have a lot of feedbacks, we have thousands of support conversation where people are asking for things, we have also competitors, we have saw it's very different, but in the same way before each feature or big feature, we're like, okay, how much time will it take to build? Which is much easier to anticipate than the marketing side of things. How much will it cost to run, how much people should use it before we call it a success and most importantly, how much could it cost to maintain?

Arvid: 14:59

Yeah.

Pierre: 14:59

And same, if a feature is hard to maintain and no one's using it, we don't have a hard time scrapping it. We were lucky enough that we haven't had to scrape a lot of things from our feature, run map also because our product is quite simple currency, you know, it's just a simple API. So what we call a feature or basically API parameters.

Arvid: 15:23

Yeah. But that's the thing. Right? You like, you you have to expose them to people. You have to educate them about them being present.

Arvid: 15:31

They have to use it. They have to figure them out, and then they have to use it reliably. That must be hard to track. Like, how do you do that?

Pierre: 15:38

So to track so basically to track usage of new features, we use Datadog. So basically all the API requests arrives to Datadog, and then for every new feature we launch, I create a dashboard. When I see percentage of new user using it, percentage of actual user using it, and a growth of this usage. And we also release feature gradually. So basically the first thing we would do is like we develop it, we don't document it anywhere.

Pierre: 16:12

So basically no one knows about it, except if we're going to send a message to a particular customers, hey, by the way, we just added these parameters. Here's a quick doc, it's not documented yet, play with it. Let us know what you think about it, which is exactly what happened with the AR stat. Then I'm going to write the documentation about it. So still only either the new user or the hardcore users heard about it because they like to read our documentation.

Pierre: 16:42

Then, 3rd step, we would update the request builder, so as we have basically an HTTP visual request builder on the dashboard. And here we can emphasize on some new features by saying, hey, by the way, we have this new feature. And 4th step, we send an email to everyone and update the marketing website. So it's like a staircase approach, which allows us to catch the problems beforehand. So for example, when you start using the AI feature we developed, it was a time where I was reading every single request that was sent and the answer, and to try to see if people understood, if the documentation was clear enough, and it was not.

Pierre: 17:34

And so it's an opportunity to learn without having to ask lots of question to to our users.

Arvid: 17:40

That is such a cool approach. I I love the fact that you keep an eye on this stuff. Like, that is just I mean, you that is expected of a founder. Right? Like, founders need to understand what's going on in their business, but doing this on the granularity of an API request level, that's top notch.

Arvid: 17:54

I love this. By the way, I I wanna point out your request builder is probably the reason that I'm a customer. Oh. Like, that thing is is just the the easiest way for me to just see if and how it works was to put in a URL, click all the little things, and see which combination of them actually gave me the result. Right?

Arvid: 18:14

Like, I've been trying to scrape, particularly, like, podcast related information because that's what PodScan needs, a lot of information about ratings and reviews from all over the place. And I need information from bigger platforms. And I wanted to see, can I get this information? Because I tried with other scraping services, and I couldn't. So I went in there, put in my URL, and then just click the kind of proxy that I needed and the kind of data that I wanted and the kind of request it was gonna be until I saw the result.

Arvid: 18:42

And once I had the result, I just copied that URL and put it right into my application and just kept churning out information. That is a really cool engineering as marketing, I guess, or as a, engineering as customer retention or acquisition tool. It's it's such a smart idea to make it easier for for me to get to my moment of value almost immediately just with a little bit of clicking.

Pierre: 19:06

Thank you so much. Yeah, I spend, I say I because I developed it, and we developed it very quickly after we launched, like probably 6 months after the launch. And I think we were one of the first to do it in the web scraping space. Having a query builder was not a very brand new idea, you had this elsewhere, but in the web scraping space, definitely not. And what we wanted to achieve is like to get the smallest time to copy paste possible.

Pierre: 19:37

Yes. For a developer because it was pre AI and all, and we felt like when I'm using an API, I don't want to read out to instantiate the SDK, where to put my secrets, what kind of configuration variable name I need to put, if this parameter accept a tick or a boolean and all. I just want, you know, to request this resource and get a response. And then if I need to tweak the request, I'm smart enough to read the code. So, yeah, this is why also on the documentation for every single parameter, we generate a snippet in 6 language.

Pierre: 20:18

And it seems useless because, you know, you're going to generate 1 if you want to add a wave parameter, then another one if you want to add a render. Js parameter, but people love it. Just copy paste in their script or whatever, it works, and then they can start working on it and iterate. So so yeah. And the good news is like a brand new version of it is coming out, I think, probably Friday.

Arvid: 20:44

Wow. I'm excited. Like, I mean, the one you have is already good enough to get me to sign up, but, hey, it's it's always room for improvement.

Pierre: 20:51

Yeah. The current version is good, but it's not great for discoverability. You know, we have so much feature here and the hierarchy of feature is very bad currently because I just get added feature over features. And so we've That's what I noticed. Yeah.

Arvid: 21:05

I remember this. I I I remember looking at it and was like, hey, this item is here

Pierre: 21:09

Yeah. Yeah.

Arvid: 21:09

Yeah. And this related item is down there. They probably were built in this order. Exactly. Exactly.

Pierre: 21:15

It's a bit of a clutter, and so we've worked with a UI designer. Here, you can be great at everything. So we've worked with someone very good at UI, UX, and he rebuilds the whole thing and it's being, developed and integrated. And that's, yeah, really matter of days.

Arvid: 21:35

Yeah. I I love it. And I love the AI feature. Like, let's let's talk a little bit about this because to me, in in the world of let let's just say the overlapping world of SEO and scraping and content, AI plays a big role in every single part there. Right?

Arvid: 21:49

Like, I I posted something yesterday on Twitter. I think I was talking about the the dead in Internet theory, where it's all about bots are talking to bots and bots are writing articles for bots to read, and then those bots make a summary that another bot reads and then does something. It's it's just computers talking to computers all the time. And I think the prevalence of AI has helped this become a problem even more. Right?

Arvid: 22:11

Because now when I when I look about what I'm actually doing is, like, I am using an AI to scrape a website that has probably some parts that were created by AI. So now it's really just a computer talking to a computer and extracting information. That it it feels like AI is a is a problem in this space. How do you how do you deal with this? Because you probably also know that the the big AI, the LLM companies, the big big model companies that you can, like OpenAI and Anthropic and all these places, they are using scraping quite significantly.

Arvid: 22:44

And that has if it's if scraping had a problem before, it has exacerbated the the public opinion of of scraping in the first place. So how do you deal with the wave of AI, both tools and content in the world of scraping?

Pierre: 22:57

So the way I see it is like a signal via noise problem. So and and what's interesting with AI is like, first, I've only thought about AI about allowing people to extract signal from noise, to extract information from data. So basically you have ecommerce page, you want to extract structured information. But then with LLM, the other thing happens, like people add noise to signal. So let's say, you know, the most basic thing is like craft a very nice email to this person asking for a refund, you know, you're just adding noise to signal.

Pierre: 23:36

And so you're right. It's like so we're currently in an ever ending loop of this, you know, some people add clutter, others filter it. In terms of scraping, I mean, what's cool is that people will, I think, always need scraping because everyone want to own their own data. And so Transfis, the only way to do it at scale it's to either pay some website a lots of money the way Google paid Reddit, for example, or you scrape publicly available data. And so that way you can work on the same data as everyone else, but own the signal and your added value will be there.

Pierre: 24:24

You clean this data, treat it, train it and all. And of course, LMM has made this easier in a way to extract signal from data and also harder because there are much more noise out there. But, yeah, that's how I see it currently.

Arvid: 24:42

I I like this perspective on on scraping in the 1st place. And it's I don't think it's surprising. Like, you operate a scraping company. You have a nuanced perspective on this, but I think you you're absolutely right. You you're you're dealing with publicly available data that is just hard to to aggregate.

Arvid: 24:58

Like, if you don't have the direct connection, like, a a direct API access or something like this. And everybody wants to protect it because they I don't I don't know why really. I mean but I have the same thing with PodScan. I also wanna protect all my information because that has been, like, accumulated and, like you said, cleaned and examined and analyzed, which was why there is an API in the first place. But scraping is something that is just the thing that's just normal, and everybody kinda uses it to a certain extent.

Arvid: 25:26

We certainly use it when we use tools like Ahrefs or anything like it. Like, there's there's a lot of scraping behind the tools that we use as well.

Pierre: 25:34

I think Ahref is I think they say they are the 2nd biggest web scraping company behind Google and it's probably true. And the way I told you that, so you get the data, you extract information, so this is your gold, your treasure and you can get in multidimensional because you can have history there. This is exactly what Ahref has. Ahref, they have a snapshot of the internet for the last, I don't know, 15 years, probably. And this is awesome.

Pierre: 26:08

And then, so you can have a time dimension, get history, and then you can have other, you can add other dimension, try to cross reference stuff in there. You can cross reference for you. Let's say, yeah, you could probably build the Ahref of podcast, you know, like those kind of words are trending a lot and get lots of user It's

Arvid: 26:30

quite literally what I'm building, Dan. Guest.

Pierre: 26:32

Yeah. So, yeah, I think this is a, yeah, a very interesting way and probably is the most not efficient, but sure way to value the data you're extracting Yeah. From the web.

Arvid: 26:46

I think that's right. I think you you you kinda compared it to the the the modern gold. Right? It's like the data is the value. Data contains the value.

Arvid: 26:55

You still have to combine it to Exactly. To extract it. You have to kinda boil it down. You have to, you know, melt it into its shape and then reform it into something maybe more pure. But the the data the potential of information is in the data.

Arvid: 27:09

And I I think you have to understand that almost every single business has something like this. Right? Every single business has some kind of data that can be analyzed for even more value in some way for somebody else. Like, even if you're just a b to b SaaS, the information is contained in what your customers do. Like, how they aggregate their own projects, how they structure their data, what they pull in, what they get out of the product.

Arvid: 27:34

There's there's always something in it. I wonder one thing because you were saying, like, Ahrefs gigantic company, Google gigantic company. They have, like, 1,000,000,000, billions of requests probably per hour. How do you deal with your own clients with just a sheer amount of data that flows through the system? What is your infrastructure like?

Arvid: 27:52

I really wonder. Like, I have no idea how big, like the scraping bee back end, the distributed proxies and all of this is. How much can you divulge? Like what does it look like?

Pierre: 28:04

Honestly, it's not as complex as people would think of in terms of, if I had to do a diagram of the scraping via infrastructure, it will be quite small. What's hard, definitely it's to work on the pipe, like, because we have a lots of network bandwidth, we need to ensure that connectivity and speed between each bricks of our infrastructure works well. We probably have custom conversation with engineer at all the cloud provider we use to ensure that first, they are okay with us doing lots of web scraping, and second, that they allow us to use as many bandwidth as we need, let it be paid or free, because once you start scraping at scale, it can start triggering lots of internal alarm, you know?

Arvid: 29:01

Yes. Yeah.

Pierre: 29:02

So How does that work? You you need to be careful about that.

Arvid: 29:05

How do you get into these conversations with these hosting providers, with these cloud providers?

Pierre: 29:11

So, I mean, at first, we didn't get into those because creeping me was too small. And then we would start receiving some emails about, we started noticing some weird metrics about your application. Can you please let us know about it before we completely shut down your account? And then you start explaining, you know, we're scraping beer, we're incorporated in France, we're a regular web scraping company, no worries. We only scrape legal website and all.

Pierre: 29:40

And now we're at the size where when we want to work with another provider, we go as a big sales team route and all. And so the door I kind of opened for us and they understand our problems. So so imagine it's much easier right now than it was, let's say, 1 year or 18 months ago.

Arvid: 30:02

What did you do to gain trust? Like, because to me, like, the the world of scraping, there's always bad actors, there's always good actors. So how can you convince the people you wanna work with that you try to only find good actors and try to actively combat the bad ones? Like, how do you build that trust?

Pierre: 30:20

So honestly, a lot comes down from the fact that since days 1, we put our faces on our website, and it seems very, I say, anecdoctical right now, but 10 years ago, 7 years ago, you couldn't find any name on any web scraping company. They were all incorporated in Eastern Europe, Russia, Cayman Island, things like that. You would have to pay by Bitcoin or stuff like that. You know, it was not bad per se, but it was they add all the red flags. And so we're like, yeah.

Pierre: 31:00

And for us, it seems like it was not logical, so we're like, you know what, we're going to use Stripe like any regular business, we're going to put our face on it, we're going to publicly state that we're incorporated in France, we're going to put our company address on the website. And people were like, okay, first it's real humans. So that's a start. And then for people who are doing regular web scraping, they don't really care that you're a bad actor to do it. The only way to know that we're trustworthy would be to do bad thing using Sweep.

Pierre: 31:38

Nd and getting banned. And so not a lot of factor do this, you know, we haven't had any regular good company doing bad stuff to check if we were doing good stuff, but, we still we're still highly incentivized to do it. First, regarding our relation to Stripe, of course, regarding legality. And also what's cool with scraping bee that scraping bee is too expensive for very bad actor to use it at scale. So for example, if you wanted to launch a DDoS attack using scraping v, it would be 2 order of magnitude, more expensive than by renting a botnet on the dark web.

Pierre: 32:24

So basically, there's 2 very bad actors in the web flipping space, DDoS and credit card tester. And so for credit card tester, what you do is like you're highly cautious about any scraping regarding endpoints that looks weird. So for example, you can scrape any Stripe endpoint using scraping B with post request, and many other things we do that I obviously won't share here, but, you know, you were surprised that we monitor things at a request level, and it's also one of the reason why we do this, you know? Gotcha.

Arvid: 33:05

Yeah. Makes sense. Makes absolute sense. I I wonder, like, it reminds me a bit of, Stripe's radar system, right, where they have this kind of preemptive, checking for bad actors. Are you implementing similar, like, AI based heuristics in your system to figure out if people might do something bad?

Pierre: 33:23

So it's not as advanced as Stripe, but we have a lots of heuristics out there and lots of things that are blocked by default if we don't know you.

Arvid: 33:32

Okay. Oh, yeah. What does that mean? What does it mean if that that you don't know me? Like, when do you know me?

Arvid: 33:37

What's your know your customer strategy like?

Pierre: 33:40

So first, we're never going to know you if you don't want to know us. So basically the way it happens is you have specific waste scraping needs, you see you can do it by default, you see you have to talk to us. So for example, if you need very big concurrency, and then we'll start asking, okay, who are you, who do you work for, what's your use case? And then we're going to check if what you're telling us is true with what you're doing, because we knew everything you're doing. We have lots of things in place that we developed at scale that we can add some web scraping rules for every domain that are scraped.

Pierre: 34:21

So basically we can block every domain in 20 seconds, no one will be able to scrape them, and on a per account basis too. So we can say, okay, no one will be able to scrape this domain. Okay, this user, let's say it's a security researcher. We trust him, or her. They are working in a university and all.

Pierre: 34:44

They have the university email stuff. We had a call. Okay. We're gonna white list this domain for this user.

Arvid: 34:51

That that sounds so hard to do with just 6 people on a team.

Pierre: 34:55

It's hard, definitely. And this is also why we want to to to to expand the team a bit, you know, like you asked us, how do you know your customer? I feel we feel we don't know them well enough and we should proactively engage more with them. So that's definitely something we're gonna invest in next year.

Arvid: 35:18

So when you mentioned the names of your the people you work with, that kinda sounded like a lot of French people. So is is that a local company? Are you are you globally distributed, or are you mostly based in France? How does it work?

Pierre: 35:31

So basically so, yeah, I cofounded the company with Kevin, with French. We're actually in the same high school.

Arvid: 35:39

Oh, wow.

Pierre: 35:40

Our first employee was Etienne, who is also French, who was also a high school friend. That's awesome. But we're the only French guy in the company.

Arvid: 35:51

Okay.

Pierre: 35:52

We also have Nizar who is in Morocco, Sahil in India and Peter in England, in London. And then lots of, freelance in, yeah, Spain, Germany, Canada, probably US.

Arvid: 36:11

That's cool. So I assume scraping beer is a French company though. Right?

Pierre: 36:14

Yeah. It's incorporated in France.

Arvid: 36:16

Did you ever feel you might wanna have, like, a a US company for a business like this? Or but, you know, I'm I'm German. I know Europe and the the VAT and the, like, mini one stop shop. Yeah. Especially if you're German.

Arvid: 36:29

Yeah. I I know the weird tax stuff. I know the the privacy stuff, the legal stuff. There's a lot of red tape and a lot of complexity to running a company like this. So has that ever, have you ever like thought about maybe taking it somewhere else?

Pierre: 36:43

Definitely. So there is 2 things here. First, when we started the company with Sketin, our goal was okay, we need to reach 10 ks MRR to make a living so we don't have to go back to work for someone else. This was the number one goal. And so we were young, inexperienced, and so you don't think about all that.

Pierre: 37:05

You're like, Scripting Video will never be big enough for us to have those kinds of issues. And even if it's at some point big enough to have those kind of issues, then good for us, you know?

Arvid: 37:19

Yes.

Pierre: 37:19

We were wrong. Should we do it again? Of course, we would have done the guns, probably the Stripe Atlas way much easier. For example, I mentioned TinySeed. We were the first French company they invested in.

Pierre: 37:36

And so what was funny is like, I think it took only 3 15 minutes call and a couple of lawyer email to sign the agreements. So all in all it took, I don't know, 2 days maybe, but then all the paperwork related to us being French took I think 5 weeks. Like we started the program, we hadn't received the money yet. So definitely it was very complex. To be perfectly honest, there is lots of issue with doing business in France, but one good thing is like, if you start a company, you get lots of incentive and help in a way.

Pierre: 38:21

So it's like a bit of water in this entrepreneurial hell that is France, but when you start out, honestly, it's good, honestly, it's good. And so this is also one reason we incorporated in France.

Arvid: 38:36

Yeah. And you made it work. Right? Like, the company Yeah. Is there.

Arvid: 38:40

It is profitable. You have employees. It's alright. But I I just wonder sometimes. Right?

Pierre: 38:45

It is costing a lot, like Yeah. But, but, yeah, we we made it work.

Arvid: 38:52

Europe. I I remember when we when we sold FeedbackPanda back in the day, and I looked at the the amount of taxes I had to pay just for selling the company. I was like,

Pierre: 39:00

And then you learn that in the US, like I think the first 10,000,000, like are tax free and you're like, oh my God. And you start making the running the number in your head, but for PodScan, PodScan is in the US. Right?

Arvid: 39:18

That's that's exactly why. Because I also got funding from the Calm Company Fund, which, like, you know, TinySeed and Calm are, it's very similar in that regard. And to be even to be able to do this quickly, I was just like, I'm not gonna turn this into a Canadian company. I'm just gonna go through the the Stripe Atlas route or, was it first base in my case, which is also a com company fund company. But it was so easy to set up this company, and the money was there, like, the next morning.

Arvid: 39:42

I was like, how easy can it be? It's crazy. Yeah,

Pierre: 39:45

that's what

Arvid: 39:47

it is.

Pierre: 39:47

It's crazy. I mean, there is some talk here and there for European Union to create kind of European LLC with a standardized way for foreign investor and European investor to invest in a That

Arvid: 40:02

would be great.

Pierre: 40:02

European company in one way, it would be great. But but, yeah, definitely not the the wisest choice we've made, but, you know, you made the decision based on what you know at that time.

Arvid: 40:15

Yeah.

Pierre: 40:15

And our world at that time was very different from what it is now. So I

Arvid: 40:21

think that's that's the most important thing to highlight here. Like, you were at a stage of your entrepreneurial journey, and you did the best you could then. And if you were to do it again now, obviously, you learned a lot along the way, but you had to learn it first. Right? You had to first go through the hell that is French entrepreneurial life to understand.

Arvid: 40:40

It's it's fine anyway. And and that's the thing. When when I think about European companies like yours, you you kinda see that reflected in the prices too. Like, things are just a bit more expensive because people have to deal with a lot of more things. Right?

Arvid: 40:54

So if it has the benefit for you to kinda shoo away all these bad actors because it's too expensive for them, that's a win win in my book. Yeah. Right? Exactly. How did you get to the the pricing scheme?

Arvid: 41:06

Like, I'm I'm always interested when when it comes to a system, like a pricing model like what you have, which is credit based. I I wonder, like because it has has multiple layers where you have the price per credit and the credit cost for each action. Did you just kind of balance them out over time? Like how much did that change over time?

Pierre: 41:26

So I think we never change the credit per request, you know, like, so if you scrape using a browser, real browser instances, so we call this JS rendering option, it costs 5 credits instead of 1. I think it does always in that way. We did iterate a lot on the pricing tier, especially in the early days, I think we changed our pricing 6 to 7 times in 2 years. Yeah, so to be honest, when we launched, we had one competitor, and so we didn't want to be more expensive than them, because our infrastructure was much more lean, we're actually able to divide prices by 4 on some option. And so we want it to be, you know, you run the number, we want it to be profitable enough for it to make sense.

Pierre: 42:23

And so that's how we came up with a number.

Arvid: 42:27

Yeah. That's how it works. That's perfectly fine. Yeah.

Pierre: 42:30

Yeah. I mean, no very complicated things here with Kevin. We're always like, okay, let's launch it this way. If too many people tell us we're too expensive, okay, we're gonna work again about this.

Arvid: 42:44

Do do you expect to go upmarket like most companies do? Like, that's kind of the story of SaaS. Right? The the moment they get to this this elephant and then there there's another elephant on top, all of a sudden, stuff gets more expensive, and the product gets more expensive and more elaborate, more enterprise y. Is is that where you wanna go?

Arvid: 43:00

Do you wanna kinda always try to stick in the niche, like stay with the developers that you already serve?

Pierre: 43:05

So I think we can do both. I mean, we will never be able to attract to the lower end of the market. So for example, for web scraping for me, anything, no code web scraping tool, this is a very low end in the market. When I say low end in terms of our RPU, like because it's only user who are non technical, who probably needs only to scrape 10,000 pages months max, and is scraping the real money is when people are scraping lots of pages per month. So we still want to be able to talk to people scraping around 100,000 a month, and definitely those people, iChurn, medium RPU, and so if you want to gain growth and growth and growth, you need to move up market, but you can move up market while keeping the same user base.

Pierre: 44:05

So this is why 2 years ago we scraped the $9 a month plan and the $29 a month plan, which makes scraping the cheapest chair is probably one of the most expensive in the wave flipping world, But this is what allows us to, this is a good balance for us because otherwise we would have needed way more support people or no support at all, which I don't believe in at all. And you can also build a team to work, you know, on the custom contract, Stefano, and working with the with the big money Yeah. Up there.

Arvid: 44:42

Yep. That makes sense to me. And I think there's a strong signal value in this too. Like, when I look at the pricing page and I see, like, a $100, that's like, okay. This is a serious business.

Arvid: 44:52

Right? This is not, like, bottom of the barrel, like, cheap as you can be. This is like, okay. The the the quality is supposedly here,

Kitze: 45:00

and then you

Arvid: 45:01

sign up and you you check out, and then you actually find the quality. So it it I I think it's very well reflecting what what you offer. It's really cool. I I really like the product.

Pierre: 45:08

Thank you.

Arvid: 45:09

Thank you, Ollie. Big fan. Well, okay. Now we we talked a lot about, you know, where you wanna go, how you're gonna get there. If anybody wants to join you on this journey and figure out more about you and your business and you personally and your socials, because you share a lot of good stuff, where do you want people to go?

Arvid: 45:27

Where do you want people to follow you and your journey?

Pierre: 45:29

So I'm currently only active on Twitter, basically. So Pierre De Vilve on Twitter, and for scraping me, just go on scraping me.com, and you'll learn everything you need to know about it. Yeah. That's where I share all the stuff I wanna share online currently.

Arvid: 45:50

Yeah. I really appreciate it. I think, I mean, I'm a paying customer. This is not a paid advertisement. Should be, but isn't.

Arvid: 45:56

But I I love the product. I I really enjoy how easy you make it to to scrape, to get information. I think the website is well done. Like, shout out to your SEO people that you they know what they're doing. Like, they I think

Pierre: 46:07

you saw it.

Arvid: 46:07

Get the the content right. And shout out to you for building a a really, really cool product. I I really like it. And, again, thank you for being so active on Twitter in the entrepreneurial community. I think that's a very, very big thing.

Arvid: 46:19

Like, you from a position of having had the success already and and building continuously building a company that's just getting better and bigger and more reliable and more stable. I think you're giving a lot back to the community. I really, really appreciate that. So thank you for all you do, and I hope you get to do it for as long as you wanna do.

Pierre: 46:36

Thank you, Ovid. Thank you for all those nice words. It was awesome being here, finally, talking to you. And Yeah. That's right.

Pierre: 46:46

It took a while. After after reading your content for for so much time on Twitter, your books and all. So, yeah. And I hope we can help Boscan grow as much as possible. And, I I really wish you the best with this new product.

Arvid: 47:02

Thank you so much, Per. It was a pleasure. And that's it for today. Thank you so much for listening to the Bootstrap Founder. You can find me on Twitter at avitkahl, ervid, k h l.

Arvid: 47:11

Find my books on my Twitter, of course, tattoo. And if you wanna support me in this show, please tell everybody you know about PodScan.fm, Leave a rating and a review. It makes a massive difference if you show up there because then the podcast will show up in other people's feeds. Any of this will help the show. Thank you so much for listening.

Arvid: 47:28

Have a wonderful day and bye bye.