Hans Dekker’s Post

Transcript

Hey everyone, Hans here with yet another quick video. Today we're building something a little bit different. We're building a startup or company database. So this will be a database of companies that have recently raised money very recently. Like ideally we'll have, you know, the funding news in there as quickly as possible and then we'll score and enrich the companies with how likely they are to outsource something and other. pretty exciting data points like business model, et cetera, et cetera. We're using Clay, so the Clay new RSS future, I've been meaning to find an interesting use case for it. And as always, I'm very excited to show you what I'm building. So let's dive in. Here we are. So here's an example of what we'll be building. This is a tool or a website called Seed Table. And with all of these, they're pretty similar. I'm pretty sure all of them are just a ripoff of Cyberleads. Pretty sure that he was the first person to do it, where he built basically a list of companies that just raised money. that are potentially likely to outsource. And then a lot of other people started copying him because he talks about how much revenue he was doing publicly. So by now he's doing, I think around five, 600K a year because he also has services on top of it where he has these leads, but he will also reach out to them for you. Then seed table, from what I can tell just by the... the wording that they're using, I'm pretty confident that at least they got a lot of inspiration from Cyberleads, but it doesn't really matter. This is the tool they've built. So they're focused on European startups only. And just to give an idea of how copycats usually do, these guys are only doing like 10 to 12K a year. And they have some, you know, how likely the companies are to outsource the industries, the business model, et cetera, et cetera, and contact information. We'll be building this entire tool that you see right here, just in Clay. And yeah, I'm pretty excited to show you how we go about doing that. And it also gives me an opportunity to show you the score, so how you can use the scoring feature in Clay. So here. I see table and say, okay, we have your one to 100 score where they have a decade over of experience with investing and they calculated dynamically using a combination of quantitative and qualitative data points, which is pretty interesting. So the size of a fund, we won't be doing exactly this the size of a fund, we will be using the size of the check. the size of the fund. Yeah, that's interesting. You know, the bigger the fund, the more likely they are to make successful investments, I guess, or at least have done their research on the company. So that's what we're building and we're using like I said the clay RSS future for it. I know I'm repeating myself in every single video but with everything there are a lot of different ways you can build it. If you have access to Krunz Base Pro you can set up Epiphy, Krunz Base, Scraper. You can set this up and somewhat ironically I guess Epiphy is in here as well. in this first sample of 10 rows. So you can do it that way, use your Crunchbase, but for now I'm not paying 500 bucks a month, a year, just to get access to there, with the risk of being kicked off the platform if I'm scraping down a little too liberally. So Crunchbase is an option. There's this website called Funds, so funds .net and they have an API, but you can also do CSV exports and that will cost you like 200 bucks a year. They're really, you know, they're really fast with with their data, so that's one way. But again, we're using the RSS future and what we'll be doing is we'll use techfundingnews .com as a source so they have, you know, news articles. then they're publishing every single day. They're publishing, so they're covering the entire world. So they're worldwide coverage. And just like every other company like this, they have a feed. So if you go to just the domain and then you do forward slash feed, then this is their RSS feed. that we'll be using. If you're ever not sure where to find this, how you can find it is you do a right click, view page, source, then you control F or CMD F and you look for RSS. And then right there behind href, that is their feed. So this is their RSS feed and this is what we'll be using. So we basically plug that into clay. In clay you basically you... create a new table, you pick the RSS future, and then as a URL, you plug in that, you know, techfundingnews .com slash feed. And they're publishing 10 items, so we get 10 items in here. Works the same as a webhook. Whenever they publish a new item, it will get pushed into Clay. So then what we're doing, then the downside of going with this approach is that you don't get all the data. you know, ready for you. So you need to get a little bit creative to give you an idea of the type of data that we're getting. We're getting a link to the actual article, the title, then the content in a couple of different formats, the creator, the categories, which, you know, I should be adding as well. And like the snippet of the content, so the preview, then again, the content in different formats. Now, they don't give you a link to the website or... of the company or the LinkedIn page or what have you. That's because we're sort of hacking this method together instead of using an official API or something like Crunchbase. But that's fine because we have AI, right? So basically what I'm doing here, I'm using GPT and I'm going over this pretty quickly. Some of you may have already seen it. I'm launching in community. In the community, I will have the full tutorial, so the entire build out available, as well as the table. So I'll hit share, and you can actually grab the table. You can build this for yourself. Or if you just want the data, I'll make the data dumps available in there as well, so like a streaming table that will update, and then you have access to the data for yourself, if that's something that you want. For, you know, if you want more information on that, check out hansdekker .net, forward slash mastermind. That's how you can get access. And... I've used, I've built a really similar table to this one before and someone sent me a message on LinkedIn saying they send 60 emails to that list and they booked two meetings. So that was pretty cool. Hopefully this one will do the same. The same types of numbers. So we're using AI to basically say, okay, here is an article, get me the company that's being featured in the article. So then it gets us the company. Then we have the company name, then... For free using Clearbit, we can then do the company name to company domain or to website feature, which is pretty darn accurate. Obviously here, this is not really relevant and we didn't get a result. That's perfectly fine because it would have been a bad result anyway. We're using AI again to get the total amount raised. We're getting that from a title. So here in the title, always we'll say the exact amount. We want to format that as a number because we're going to score everything later. And the more they raise, the higher that score. Then, okay, this is the company. We get their domain. Then with their domain, we get the LinkedIn page. And then from the LinkedIn page, we get description. What we obviously get. A lot of data points that we're using the description that the year they were founded, the industry that they're in, the country, their follower account, I thought that might be a pretty interesting data point just to get a scope of the maturity of a company, how recently they launched, maybe they just came out of stealth, that type of thing. I bet that for some people that's an interesting data point. Their employee account, then we want to get open jobs. So if they just raised and they're hiring, That's great, right? Then chances are they need a bunch of stuff and you can sell it to them. So we get the total job count. Then I'm using a formula. So here we have all the jobs and basically using a formula that says, so like open jobs, get me a comma separated list of all the job titles. So that is what we see right here. So those are all the job titles that they currently have open. then we use AI. And I've shown this one before where basically we say these are all the open job titles. With those job titles, pick the departments that they belong to, meaning that, okay, these are the departments that they're most likely to outsource. And this is a list of all the departments that LinkedIn uses as their filter. So basically here in this case, I think this is Appify actually. They're looking for technical customer success manager, product manager, then like more product managers, engineers, and a business development rep. So then AI tells us, okay, they're likely to outsource product management, engineering, business development, and support, which is great. And then obviously if you sell anything, like any of those services or what have you, or you're just gonna use that for personalization, great. Then you can use that. And then getting the company type B2B2C, that's pretty straightforward. Then this one right here is the scoring future. I put in five here, that should be four. And so then we have the four for tier that we're using for scoring. I put together a little bit of a random scoring model just because I wanted to create a tutorial. And I think it works. We're basically looking to generate a score between zero, and 100 just like seed table does. And then basically, so you have these values here, right? So you can say, first you tell it which value to score. So in this case, the funding amount, I want to attach a score to the funding amount, higher funding amount being a higher score. So then we tell funding amount and we formatted that as a number. So then we're telling it, it's a number and you know, between, so we're giving it a range. And if the number is between zero to one, so just in case, you know, it's an off -topic news article, then give it a score of zero. If it's between one and a million, give it a score of 10. Between a million and 10 million, give it a score of 20. Anything more than 10 million, give it a score of 40. A lot of different ways you can score this, but it's just, you know, a basic model you can use that way, and that's how you use ranges. They explain it more here. but this is how it's set up. So, you know, your first value here is your first score here. Your second value here is your second score here, your third, your third, et cetera. And so on, actually. And then the second value that we're scoring is country, which is a bit of a random one, but you know, if they're in the US, like if they're in North America or English speaking country, then they get a higher score. I haven't had to, I even put the full list in here. but there's just like a random small list of countries that are more likely to be open to receiving communication in English and they get like a low, like, so this is more important so they can, like, so an account can earn more points here pretty much. And then this is not as important. So that is, there they earn fewer points. That's the idea behind that. Same here for follower count. Again, I thought it was a pretty interesting metric and I wanted to try out some interesting scoring methods here just for the sake of doing it. And here in this case, if they have fewer followers, they actually get a higher score. Just like I said earlier, imagine they have a lot of open jobs. They just raised a really big round. but they have really, they barely have any followers on LinkedIn and chances are they just came out of stealth and they're looking to accelerate and they're looking to outsource. That's the logic behind that. You can probably argue against that and I'll be very open to hearing your opinion. And then total job count, that's a pretty obvious one. If they have zero to one, between zero to one open jobs, then they get a score of zero, one to 10, five. 10, 120, 100, and up 30. And that's that only run it if we have a domain. And then that gives us the total score. So then these companies have a score between zero to 100, a higher score, meaning they're more interesting accounts for you to potentially retail to, especially if you're marketing, IT, or support services, or product management, engineering, business dev, or support services. et cetera. That's how you build a tool similar to or a data set. I should probably say similar to C table or cyber leads. Hope you found it helpful. Again, if you want access to the full tutorial, upload the full build later this week, but you can already get access to the actual table as well as the data. You can get that at HansDekker .net forward slash mastermind. Hopefully I'll see you there. If not, I'll hopefully see you in the next video for now. Thanks for watching and bye bye.