Cloudflare TV

Building On The Edge with Special Guest Hamlet Batista of RankSense

Presented by Zack Bloom, Hamlet Batista
Originally aired on 

Join Zack Bloom, Head of Developer Marketing at Cloudflare, as he interviews Hamlet Batista, CEO and Founder of RankSense and they discuss his experience building a business on Cloudflare Workers.


Transcript (Beta)

So I'm Zack Bloom. I'm here live with Hamlet Batista. And Hamlet is the CEO and founder of a company called RankSense.

And I'm incredibly excited to have this conversation for a couple reasons.

One is because I don't think anyone on this earth does a better job of telling the story of what I think the future of the Internet is.

I think if people are not familiar with it, a few years ago, Cloudflare came out with a product called Worker.

And Worker has made it possible for anyone to affordably and efficiently write code and run it in 200 places around the world, really close to virtually every user on earth.

So suddenly, people went from writing code and putting it in one place to writing code and making it available to everyone.

But like any technological innovation, it wasn't just the technology that was necessary in order to kind of demonstrate that this is going to be the future in order to make this future reality.

It also required people to build things with it, required people to have really good ideas, and use the technology, use the tools of workers to make those things a reality.

And to me, RankSense is a perfect example of that, of what the future of the Internet can and should and hopefully will look like.

It also is a really good answer to one of the most common questions we get, which is what can people or do people do with workers?

What are some of the interesting things that you can do with this new technology that you really couldn't do otherwise?

And so I'm so happy and thankful that Hamlet agreed to talk with us today.

And I'm personally just very excited to hear his story and the story of RankSense and what the future of the Internet just might look like.

Thank you, Hamlet. I'm really excited. I'm more excited to be here with you, Zach, you know.

And we have, you know, interacted over the couple of years, I think now that we've been cooperating for more than a year and a half.

And yeah, no, it's incredible.

It's amazing. I feel really privileged to be able to be in this position where I feel like we're living amazing times, right?

You know, you see those rockets getting to outer space, all these AI stuff.

I mean, I think we're incredibly lucky to be living these times.

So excited to be here and share our journey so far.

Yeah, if you think about the scope of human history, if you're living right now, you see more change and transformation, even just the creation of the Internet.

You know, it's hard to imagine, given how much of our daily lives interact with and use the Internet, it's hard to imagine how brief of a period it has actually existed in human history and how, I think, early we actually are in the development of the Internet.

This idea that we have maybe not actually figured out yet the right way to deploy software, or we maybe have not actually figured out yet the right way to interact with a world where we can suddenly talk to every single person at any given moment, and that we don't just get to witness it, we get to create it, is really cool and maybe different than the lives that any of my ancestors have had throughout history.

Yeah, yeah. And also the fact that things are improving exponentially, right, which it's not intuitive to see that, but it's happening, right?

And I look, why are things improving exponentially? It's because we're building tools that when we use them to improve, and then it's like an iterative process.

The tools that we use to build the other tools get better, right?

And this is a perfect analogy of what we're doing, right? You guys are building tools for developers like ours.

As you're making them better, you enable us to continue to improve our products, and it's like a virtual cycle, right?

We provided feedback about, oh, this is the stuff we need, it's worth it.

So it's incredible and it's amazing.

So, great. I mean, it's the privilege of building developer tools.

This idea that you cannot just build a product, but you can build something that people can use to solve, oh, not one problem or 10 problems or 100 problems, but basically an unlimited list of problems.

It's challenging sometimes, I think, but it's very, I mean, I remember when I first got an email from you, probably like you said, two years ago, and I essentially woke up in the morning and someone in this world who I never met, sat down and worked with a product that I was working on at the time, because I did work on workers back then, and had built some cool thing that I never imagined and could never have conceived of.

And it was such an amazing moment.

I would love to hear from you what the story of RankSense is, and it might make sense first, actually, for us to tell people what it is, what problem it solves and who it solves that problem for.

Yeah, yeah, absolutely. So yeah, so RankSense is, we call it, in the face of it right now, an Agile SEO platform, right?

And as developers, we're familiar with the Agile methodology, right?

So doing things iteratively, having a feedback loop, contrasting that with a waterfall approach where you, because you're doing things iteratively and learning from them, you can adjust, right, and make progress more predictably and take less risk, you know, and all these different advantages, right?

So when you say, what does that mean, Agile SEO, right?

One of the problems, you know, first, you know, people don't know SEO.

SEO, search engine optimization, is about getting, you know, organic, free visitors from Google, Bing to your website.

And if you have worked in that, it's a very manual, and people's driven industry where 90% of the work is human driven, and you have to do a lot of stuff manually.

And if you're doing so manually, you have to wait typically six months to see the impact of your efforts.

So think about how painful the problem that is that you're waiting six months to learn if something that you did three to six months actually works.

And it's for spending all the time, right?

So, you know, as you go, imagine you're building an app, and you have to learn whether what you did, it works, you know, and you know, that was the case before when the software was built on a waterfall approach, you have these massive projects, and at the end, millions of dollars is spent, and at the end, whether people are going to use it or not, nobody used it, right?

So there is that advantage, right? So I say, wow, this framework, this idea of software development, can solve this problem that we have in SEO, which is, things take too long, you can invest a lot of time and effort and not see anything out of it.

And that's typically a big case. I know the problem with that, that happens with that is that people lose trust, because it takes so long, or it was because it was something else, they don't trust it.

So a big another big problem is that people when they invest on SEO, and I like to use it, that's the gambling money that companies have, it's not it's the money that they're that they're willing to lose, because they don't have the trust that they have when they pay for advertising, they get immediate feedback, it works, right?

So that's kind of like the problem, right?

So you think it's a big problem, right? It's a significant problem, which is for a little bit of context on this, is there was a time in my life when I heard someone say SEO, and I imagined what I guess you would call black SEO or dark SEO.

Yeah, where people would be hiding like hidden little keywords in some way that people couldn't see to try to trick search doing things.

And I didn't really understand that so many businesses entirely rely on SEO to function, because search engines have taken such a powerful position in the Internet, and therefore in all of commerce, that you have to, you know, you have to allow them to exist and support their existence.

And you have to really do whatever you can to do, I guess, what we call SEO, but really, in some ways, it's just building a business online, in order to get this traffic.

And if you don't get the traffic, your business disappears.

There are examples of businesses literally falling off the face of the earth, and drying up and dissolving and all of the work and effort put into them disappearing, because they lost that ability to appear in search results.

There's just so much power there.

And so I talked to many businesses where SEO, or at least being present on search results is a huge part of their basic existence.

And if they're going to continue to be able to hire people and grow and scale, so it isn't really just these dark patterns.

It is really survival and merchandising and trying to figure out how you can create something that people can actually find on the Internet.

Is that fair? Yeah, that's exactly right. Yeah, that's exactly right. Yeah.

And to contrast that, right, because a lot of people, you know, and I know that SEO and a lot of circles get a bad rap because of the buy-in aspect of the industry.

And it's similar to you going to school and you having, okay, you know what you need to do to get good grades, or you cheat, right?

You cheat and you fail, you don't really know, and then you get the grades.

Sounds like you won in the short term.

But when you go to the, when you go and try to apply for a job, you didn't really develop the skills, you're going to fail.

So similar situation, you know, you can fake the signals in the search is used to be more successful.

Now it's harder. It's almost, you know, Google says now they released a report that said they, Google is successful suppressing that black hat stuff 99% of the time.

So, and it's because they're not, they're not experts.

So they made it so difficult that it's almost the investment to try to fool Google is higher than you actually do the right way, right?

But then, you know, talking about that, right? So SEO, right, it's important people are doing it.

How do we solve that problem, right? What is rank sense?

So rank sense, our mission is to compress the time from six months to six weeks.

That's our mission is how do we get this SEO stuff faster? How do we get it faster?

Because if we get it faster, you avoid the scams, right? You get the feedback immediately, you avoid that you take advantage of the opportunity cost by massive savings on times and effort.

So that's the solution, right? And what's happened is that we have been working on this for four or five years or more on this, addressing each one of these, you know, blocks of, you know, how do we get this first?

First thing, and that's where leads to how we ended up, you know, working with you guys, right?

Which is at a higher level is how do we make this stuff faster?

First thing is, okay, the first obstacle that we have is that successful SEO that, as you mentioned, you're building a business, you want to get attention.

The best way to think about search engines is as libraries. So all these web pages, all these books, I like books in the library.

So the goal of the search engine is to catalog them, right?

So they have to inventory the pages, the books, build a catalog because you don't want, if you're looking for a book, to have to search every single shelf to find it.

You want to go to the search index, looking for the title, for the subject, for whatever, and find it.

But somebody has to do the work, you know, in the library.

The librarian has to do the work to catalog every single book.

That's kind of like the simplest explanation about how the search engine works.

So where is the gap? Where's the opportunity happens in SEO?

It's when you just build your site and you only care about the user. Okay, I'm an author.

I created my book. And then you assume that just because you created the book is going to be successful.

That's not how it works, right? So you have to invest the time as an author to make sure that you name your product right, choose the right title, you know, catalog it correctly, do the research on your audience to understand who cares about this book, who cares about the product, so that you, that metadata information, and decisions in terms of the title, the subject, and all that stuff, when it comes to the library, it's going to increase your success chance.

That's the best analogy between SEO, right, and the physical world, so that people can understand, right?

So that essentially, translating into that, what we do at RankSense is, okay, your job is to provide fantastic content, fantastic information, and build credibility, you know, like I'm doing, you know, by speaking with you and doing all this public speaking that I do, building trust by sharing my expertise, right?

Having the website and all that stuff, but now, where is the missing component to make a successful SEO is the metadata tagging.

How do you label things within your content, then, that most of them are primarily visible to search engines, so that the search engine, the librarian of the web, can catalog it correctly, so that it performs where there's demand, and you attract that traffic.

So if I understand how to build a web page on the Internet, I write all the content, but I also include these meta tags, these tags that go into the HTML, literally, open brace, meta, and then information about the page, things like its description.

And it's probably pretty common for people to do that very hastily, or without a lot of care, and then to discover that search engines are using that information in order to determine what gets shown and what gets ranked.

And to have a business that might otherwise be very successful over getting a lot of this organic traffic from Google, not get it.

And so what RankSense is doing is, I guess, and I don't want to skip any steps for you, but sitting in between the web page and the search engine, using Cloudflare Workers, to optimize these meta tags, and change the content on the fly in order to find the content that's most successful with search engine.

Is that right?

Yeah, that's, you know, at a conceptual level, yeah, that's what we are.

That's what we're doing. Now, drilling in a little bit about the how, it's more about the gaps.

It's, OK, this is how many pages you have. This is what people think about SEO.

They think about rankings, right? That's, oh, you know, I'm ranking for this keyword, that keyword.

But before you can rank, you have to index.

Before you're able to find a book in a library, it has to be in the catalog. And what happens is, the more pages, if you have a site with a few pages, it's typically going to be fully indexed.

Google won't have any problem. But if you have a JavaScript-driven website, if you have a large site, the more pages you have, the higher the chance that a big portion of them are not even getting into the index, because the tagging is not correct in a number of pages.

Because if you're doing it manually with people, you can tag a few pages.

You have a large site, and you're relying on an automated system like a CMS, there might be mistakes in the tagging that cause the pages to not even be included, because Google's not even the robots are not able to find them or group them in the right place.

So if I kind of understand it, when we talk about a CMS, which is like a platform like WordPress, it's not necessarily built in the modern era.

It wasn't built in this era when Google search results could be 50 % of your business.

And so it doesn't always do everything perfectly in order to correctly organize content and present it in a way that a search engine can understand.

I would guess that there's also a separate problem, which is when a human being builds a website, they look at it like a human being.

And they can easily miss mistakes or errors that Google crawler is going to see and cause the content to not be ranked in the way that it should be ranked.

Yeah, that's exactly right. And now we're getting we're getting a little bit into the weeds.

And I love that, right. So now let me give you one example that is probably not common known, right, which is, and I like to use this analogy, right.

So my wife and I will go to the malls. And I hate shopping, right. And I go to the mall, I want to buy something, I go to a store.

And I go specifically buy whatever I want.

And I come out, right. My wife is the opposite. She will go to the mall, and she will check every single item because she wants to learn for her.

It's like, it's like a trip. It's like a vacation to go to suspense all the time.

So I get frustrated. So I'm sitting outside just, you know, trying to, when is it going to be done?

She can spend hours, she doesn't even feel it.

Right. So I made that analogy, because the behavior of the user when they go to the web, when he goes to the website is completely different to the behavior of the bot.

The user goes to the website with a purpose, it was going to navigate, let's say, imagine a eCommerce website, thinking about buying a product, do a search, I find my product, if I need something else, you'll be very specifically a few pages, check the pages, get what they need, it's gone, right?

Typical behavior of a user.

The behaviors are changing. That's me, right? Go to the website, find what, no, the behavior of the bot is more like my wife, it will actually go and make an inventory of the website.

So it was going to check every single page of the website on every visit, right?

It's going to try to check every single thing, right?

You say, oh, you know, and why is that different difference important?

Then you're going to see why that causes a problem. On an eCommerce website, we have an element that is called the facet of navigation.

The facet of navigation is very powerful, it's very useful.

You can narrow down what you want really quickly, make a few options, you get to the actual product, right?

So you're familiar with that, right?

Guided navigation, you find it, right? Now, think what happens, imagine now, a search engine bot like Google, trying that facet of navigation, clicking the options.

It's going to try every possible combination, and the number of pages that it's going to try to inventory to access, it's going to grow exponentially.

So a site that might have like a thousand or a couple thousand pages, when the search engine is trying every possible combination in the facet of navigation, might end up with millions of pages crawled.

You say, oh, why is that a problem?

Because if I have a million pages, well, the problem is that Google has a budget per site.

So Google will only crawl a limited number of pages. And because if it's wasting a lot of time, it's not going to have time for all the sites.

It's got to be, you know, it's called the Google crawl budget. So the search engine has a limited amount of time.

If it wastes all that time on just the facet of navigation, it's not going to have enough time to crawl the rest of the site where you have the real content that it needs.

So that is a problem. You think something that you wouldn't think was a problem, because it's valuable for the end user, actually causes a lot of problems for the search engine.

And, you know, we work a lot with e-commerce sites, because that's just one example of the problems that can happen, right, that are invisible, and that, you know, you will need to actually address to be able to fix that.

Right. And that's to give an example of a target there.

So I think we have a really clear understanding of the problem and the thing that RankSense is doing.

It's closing the loop on SEO. So it's integrating where you are in your search results with the content on your web page, the meta tags on your web page.

As it makes changes, it's able to see how the website does better or worse in search results, make better changes in the future, and almost iteratively over time, improve someone's search ranking.

The thing that that is, to me, very, very interesting, because I think businesses might get half of their business from Google in the modern day, but they don't really have the tools to be able to spend half their time optimizing Google.

And so I think it's fascinating thinking how we can all be more successful in a world like that.

It's also really interesting to me how you built this, because the part where you can intercept a web page and modify it on the fly without slowing it down, you can change these meta tags, is something that I'm not sure was possible five years ago, when you started building this.

So I would love to hear what the journey was of trying to do this, I don't want to say obvious, but like very clearly good idea, how you actually went about building it.

Yeah, so that's really cool, right? So the backstory is that my inspiration for this product came from the security industry.

So I have a sysadmin background back in maybe probably 20 years ago.

I used to be at Solaris.

I don't know if people still know that whole operating system, you and I know it from some microsystems.

So as a sysadmin, I had to learn, you know, firewalls, I had to learn security.

So I said, wow, the concept of a firewall is a really interesting, a web application firewall is a very interesting concept.

So if you think about what I built for RankSense, it's like a web application firewall, but reversed.

So how do the web app, the WAF works, right?

Is that you have the HTTP traffic coming in, and you have rules that block malicious traffic, you know, based on the patterns that you have, right?

So it's okay, this is going to be an attack, you know, and you patch the traffic on the fly, right?

Which allows the sysadmin to save time on the backend, so that while you're patching on the fly the traffic, you're not making changes on the origin, you're making changes on the traffic, right?

So you think about that's a very powerful concept that exists for a long time, right?

And save the time.

I'm sorry? Basically Cloudflare. Exactly, basically Cloudflare, exactly.

And, right, I said, wow, this is a really interesting concept. And when I came, I was trying to figure out this problem about fixing the issues on the fly.

Well, I can use the same idea of the firewall, because I understand it from my time working it.

But instead of patching the requests coming in, right, I'm going to patch the responses getting back, right?

And in the responses, I'm going to be patching also the HTML inside the code, because, you know, the firewall, you can do a lot of the work and the headers, right?

So I said, wow, and with this concept, not only I'm able to patch the content, I'm also able to rewrite URLs and do a lot of really cool stuff and redirects and stuff like that.

I said, wow, this is really cool.

And I said, wow, you know, this is something I can do with a reverse proxy.

So I have a friend that he was doing something similar with a reverse proxy back in 2002.

And, but he was having like a, he mapped this, he created like a duplicate site, he created a duplicate site.

And he blocked everything to have the two sides with the reverse proxy mapping to a folder.

2010, I joined a startup when I was trying to do something similar.

I said, look, we can do this in real time.

2010, with a reverse proxy, we were adding 500 milliseconds. Just think about that.

When we're doing the proxy in 2010, I joined that company, right?

So you can imagine with that latency, it wouldn't be a successful product, right?

But we got really big brands, deep, really big brands doing that.

So just to understand, you were able to build this whole system that would intercept web pages, modify these meta tags, make them work better inside search results.

And you built it with kind of a traditional, somewhat centralized reverse proxy system.

Reverse proxy is kind of what we're talking about. It's really a layer in between whatever you normally would have put on the Internet.

And it's visiting you, but instead of having this beautiful capability of being deployed all around the world, it's generally only running in one place.

So for a lot of visitors, when they would go to this e -commerce company's website, it would have to go all the way to wherever this reverse proxy was running, and then back to where the e-commerce company was running.

And that extra, that extra leg, with adding, it sounds like an average of half a second to page length.

Half a second to human beings doesn't sound like a ton of time, but e-commerce is all kind of a statistical game.

And the slower a website is, the fewer people end up buying stuff. And so slowing down a website by 500 milliseconds probably translated to a lot of lost money for these, potentially more money lost than they ever gained doing the SEO.

That's not right? Well, in that case, some of the deployments were really successful because things were so bad back then in terms of SEO that the upsides were massive.

I mean, we were doing performance deals with a large retailer, and just to give you some numbers, that retailer on a 3% lift, they were paying over $1.4 million a year.

It's a crazy amount of money back then. But then the problem we faced with that is that they copied the technology.

They said, oh, we figured out, and they didn't build it internally.

So it was a terrible situation. I left that company, and I started Ranked since 2015.

Fast forward 2016, I signed up IBM as an alpha customer.

So I tried the whole thing again with real time, right? And I said, look, wow, I don't have an enterprise sales background.

And the pain was so big that in their case, they were trying to standardize.

There is something in SEO called hreflang tags.

Hreflang tags are submitted tags that support multi -language, multi-region websites so that when people are searching in France, they see French results, stuff like that.

IBM has a number of CMS systems, a combination of CMS systems.

And that made it very difficult for them to try to have a standardized format, the URLs, because some will have the language first, the country later, one reverse, one with slash, four slash, one parameter.

It's completely different.

They have some parts of the website with static HTML. So they were trying to fix us with Akamai.

They were trying to rewrite the URLs with Akamai.

But they faced the problem that, yeah, Akamai could rewrite URLs, but the content would break because there were relative resources inside the content.

And there was a friend of mine that I used to consult with them from an SEO center.

They have this problem.

And I hear about the prototype you're building. And I think it would be a good fit.

I said, oh, you're going to give me a meeting with them? I was going to get a meeting with them.

I created a demo where I proxied their site and showed them my solution doing what they wanted.

Look, you're rewriting the URLs and keeping the content the same.

And I used a super clever trick to do that because it was a SSL site.

So I used a man -in-the-middle proxy with some hacks and stuff like that to be able to proxy their site and show them the changes that it would work.

It only took five minutes. And they saw that I could do what they wanted. They said, no, absolutely.

The prototype. And we got them as the first alpha client. Imagine that.

I said, wow. But it's in a directory. So they got Akamai, delegate a couple of directories for a couple of groups that agreed to the pilot.

And we set up our proxies inside the IBM network, inside the network behind their corporate infrastructure, set them up.

I had a dashboard with Kibana, doing the logs in real time.

So that was kind of like the genesis of the product that we have right now.

So we have the real-time dashboard, doing the analysis, doing the proxying and all that stuff with Akamai in the right.

Then what happened? Then disaster stroke. So it was an alpha version.

And I was really, you know, wow, you know, going with the alpha version of our product to the largest customer that we could possibly imagine, you know, and then, and we had all these plans and all this, you know, you know, backup plans, you know, and there was some mistake on their end and our end where I think we were having some issues with the caching of the pages and Akamai where we're making the changes and they were not getting updated quickly.

And we wanted to release a patch to overwrite the last modified header so that we could trick Akamai to refresh the cache faster when we made the update, right?

And my DevOps guy logged in on a Monday, deployed the change and didn't verify that the deployment actually loaded and worked.

So we took down the IBM website on a Sunday.

Imagine, right? We took it down, you know, and yeah, no, they got incredibly nervous after that, right?

So they, but they were so, you know, so interested in the solution that they said, look, you know, the problem was that the backup configuration and Akamai didn't trigger the bypass.

So they made a configuration mistake.

We made a mistake on our end. And sadly, they said, okay, no, we can't, we're still risky.

And they gave us, they introduced us, we got an Akamai trial to try to build, automate that part so that there was no all these money, manual monkey patching on the software, right?

So then we work with, you know, trying to make it work with Akamai, you understand, it was a bit of a pain because of, you know, and we're talking about 2016, right?

So we said, okay, we learned their lesson, but the fact that IBM and the fact that, you know, we sold all the interest, I said, wow, and IBM paid us for the pilot.

We said, there's something real, something valuable here, which, you know, you know, a lot of us startups want to give it for free just to get, you know, and I said, wow, you know, they gave it to us.

So we, so I, then I had, at that point, I had a reverse proxy network in AWS.

But Akamai, they don't have the service internally. We had a number of clients in AWS, and then we're having problems with the deployments will take two weeks.

And then I said, you know, I started playing with Kubernetes was getting, you know, popular.

Wow, Kubernetes is our solution for this. So we go, okay, move everything.

Oh, this is awesome. Google was better in terms of Kubernetes.

And also, we needed the CDN, we needed, you know, because just asking people to proxy without getting the speed advantage, it was a big hard sell.

So we said, well, if we can have the CDN, so that it speeds up as well, in addition to the SEO, it made it a lot easier.

But then we tried the Google Cloud CDN. Wow, this is even better.

And when we moved to Kubernetes, we could get deployments in five minutes.

I say, wow, this is night and day from AWS. So we shifted from AWS to Google Cloud, where we saw everything happening faster, you know, we offer people the CDN, so we will rewrite the headers so that every static stuff will get cached on the CDN.

So they will see a benefit. So we had that. Now, the next problem that I had after this, do you have any questions on this so far?

No, I mean, I love hearing problems, and hopefully, we'll solve them.

But yeah, it's great, because it lets me vicariously live through your experiences and make a different set of mistakes and not the same sets.

I love hearing this. Yeah, no, no. So then what happened is that when we started working with e-commerce websites, security, as you can imagine, is a big problem.

You have to get this PCI. No, we're not storing credit cards.

No, no, no. It doesn't matter. You need to have PCI as a certified service provider.

So we went through that massive project to get PCI compliance.

We know in the Native West, we did all that stuff, and it was a pain, a really difficult thing.

And then I had another client that we signed them up with like five, six, seven brands, and they paid already.

And they were like, oh, you know, we didn't renew the PCI stuff because we don't want to go through the whole pain stuff again.

And they're like, what? And then their tech didn't want to approve this stuff.

And I would say, what? And I heard about around that time, I knew about the launch of the workers platform.

Now, when I tried the platform, you know, I think it was 2017 or 18, right?

I do have this first release. So we tried it. You didn't have a similar rewriter, right?

So you have, we were trying to do the HTML modifications from the platform.

I tried to give it a try. And they said, well, it's not working.

It wouldn't work. Try it. So I gave up with it. But then I had this client, I knew that you had the PCI compliance.

Okay. So the reason you were looking at Cloudflare wasn't even the performance stuff that ends up coming later.

It really was the idea that you could have PCI compliance out of the box, be built on top of Cloudflare.

Exactly. Yeah. So exactly. So I have that client and I said, look, you know, what if I, uh, what if we use Cloudflare?

They have PCI compliance.

I didn't have anything with you yet. And they said, oh, the CDR. Oh, yeah.

They're PCI. Yeah. And now that I say, oh, I have to actually, and I tried and it didn't work.

And I said, wow, guys, I promised, I offered them Cloudflare. We don't have it.

And we haven't signed up already. So we have to build it. So I said, I sat with my dev guy and you saw, we posted on the forums.

I think you replied to some of the people on the team.

And I said, oh, you're probably going to use a JavaScript parser.

And we tried to say that fail. I said, okay, let's find all of them and try everyone.

So we tried like 10, we benchmarked all the parsers. And we had like a few days to do that.

And we found one, only one of them that was able to do the changes and not hit the limits, the CPU limits and the workers that we have.

So we, it was the HTML parser too. So we got that parser. And you had already had the contract signed.

Yeah. Told them that you could do this. And then now you're presumably up at 3am or something, trying all these different HTML parsers, desperately trying to find one that can actually do the thing that you need with the work that are available at the time.

Exactly. Right. And one thing is the parser and the other thing is patching the content and not breaking it.

So we also have to go through hacks and stuff to be able to do that as well.

And when I talked to your developer and he was like, why don't you just use Radix?

I said, why are you talking about, we can't, how are we going to use it?

No, we can't do that with the type of changes that we have.

We know to know for sure that we're not making a change in sign a comment or some other place.

We have to know for sure. It has to be with the parser.

So, so we got that to work and this is the best part. So, so when I got it to work for that client and I, and I saw the, not, not initially it was because of the, of the PCI compliance, but then I saw the benefits of the speed improved 30%, right?

Because we're not double proxying. When we were with Google cloud, we had the proxy of the NGINX proxies or across the Google CDN, right?

Now we have single place and the onboarding, the onboarding is incredible.

So I said, wow, I can onboard people like this before it was all this planning and all this stuff and more people.

Now it's important. It's crazy. And also I don't have to spend money on DevOps.

I can save all that money that I'm paying and then I'm paying in Google.

And I said, wow, the money that I'm going to save on what I'm paying Google, I can hire the developers because at that point we didn't have a, a self-serving app and I can actually build a self-serving app.

Oh wow.

Right. So I use the savings from the Google cloud, from the move through Google cloud, from Google cloud to Cloudflare to build the self-serving app.

You know, the first prototype of the self-serving app that we have right now.

Right. How much was this infrastructure costing at Google back then?

Um, I think it was about 5,000 a month or something around that.

Yeah. Yeah. Yeah. We have, you know, and we have like a, you know, a couple of thousand, a couple of those inclined.

And then what I did is, right.

So I, I saw that, right. I said, wow, I was doing the workers manual. I didn't, I didn't have the app store.

Right. And then I say, oh, they're going to launch an app store.

And then I, I email apply for this stuff. I get an email from Jonathan Bruce, right.

He was managing the, the, the, the program. Oh yeah.

And he was trying to vet and understand what I, what I was doing. I was like, um, yeah, we're, no, we're interested.

And then I saw that the same week you had, uh, a summit in the, in the office in San Francisco on the East coast.

And I flew in there, right.

I bought a ticket, flew in there and I'd say, Hey, Jonathan, I'm here.

I didn't ask for planning ahead of time. And I said, you know, now that I'm here, he's going to come and talk to me.

He came and I told him my story about, you know, AWS and all this.

So look, I already have 20 clients on the workers.

He said, no, he fast approve your app. You're going to be in the first batch of people to use the app store.

And that was when he announced it and he took, took it about two, three months to actually made it available in the process.

We started adding the capabilities that we needed on the app to, to do that.

So that was, that was really because as I said, look, you know, with the app store, not only we get all these other benefits of the savings and the provision and all this stuff, we get the visibility, the exposure to your customers, which has been a game changer for us since then.

Yeah. What is the performance difference?

I mean, what, what is the experience for one of your customers on workers versus on Google cloud in this case?

Is it just a cost? No, it was over 30% improvement on speed.

Yeah. I mean, because now the speed becomes a, a non-issue, right.

But obviously there's some engineering behind it to make sure that is, you know, that we can hit those performance because we set it to add up more than 30 milliseconds.

And there's this multi-layer caching, you know, architecture that we build for that.

Right. But one of the requirements that Jonathan said, look, if you are the workers and they build per user, their requirement that I have is that it has to be useful for anybody on the, on the cloud.

So that required, when I met with Jonathan, I said, okay, I will do that. When I was flying back on the plane, I designed something that I didn't have that said, okay, this is how it was based on what I built for IBM, right.

When I built, when I was doing that stuff, I said, okay, I'm going to turn this into something that will be useful for anybody in the platform.

That's where we build the real -time monitoring that we have that is passive.

You know, we're not crawling the sites or letting the crows come in and we're doing the analysis of the issues on the fly and with a logging mechanism and then the patching, we fit it in into the, into the worker.

So as you were adding more capabilities, as you know, because when we started, I don't think the KB store was there yet.

Right. The minute that you had it, when you launched it, there were limits in how much we could fit in.

Right. So think about what we did to make it work. We will use some compression.

When you added the, the WASM capability. Okay. Now the compression, we can write it in C code, put it there.

So we're compressing in the capabilities that you have so that we can actually overcome the limits.

And it's, it's been there, you know, with all the journey, as we work, try to work with what we have, the minute that you release the, the limitations that you added HTML, we write it.

We said, okay, that's something we can replace this stuff that we have is now it's going to be better.

Right. Wow. I mean, it's, it's an amazing story to me, not just because it made it 30% faster or because you don't need DevOps, which are always great things to hear, but because it was worth you doing all of that, meaning it was delivering enough value to, you know, it's very similar to what you're saying with IBM coming to you and being willing to sign a proof of concept contract, or they would pay you just to do a demo.

The fact that you were willing to go through all of those hurdles early in the days of workers, before we had a lot of the technology that we have now means it must have really been delivering value to you.

I mean, yeah, there's nothing more powerful to me than someone building a business on top of a platform and really betting all of the work that you do every single day on the idea that this is a good idea going to work.

Yeah. And part of that is because the way that I bet, you know, this is my fourth business, the way that I bet ideas is by customer's pain, right?

You can run surveys, you can talk, you know, all that stuff.

If you are willing, before I even have the product, before I even have a finished thing like you have, right, to invest time, invest money on it, it is a good idea, right?

Because, and if I know that more people like you have it, you understand, the risk is minimal from putting best in the time.

And if you think about it, this is what I said, right?

Just, I said, look, I'm quantifying this effort and the investment I'm going to put on Cloudflare, I'm already saving the money that I'm investing here, and I'm going to take that money and invest it in adding more capabilities to my product that I don't have, like the self-serving capability, right?

And that's the way that you do these things. So that's why for me, it's very important to start from a problem, right?

A lot of, and I think these mistakes in all entrepreneurs, they're too competitor-focused.

It's just looking, oh, what are their competitors doing?

What is this stuff, you know, and trying to make incremental improvements from competitors.

So what are really big problems that people are not even considering when you even talk to the clients is non-paying, right?

When we were in talking with IBM, I said, oh, yeah, we have a problem that says, oh, yeah, what can you do about it, right?

Oh, no, you can actually do something.

Look, what, right? That's really valuable. And we have, I love talking to customers, I get into the calls because they tell me that, oh, yeah, I didn't even think you could do this.

And they say, wow, and they get excited, right?

Yeah, I mean, it is, I mean, that's kind of what my experience right now is like, which is kind of, I guess the irony is I'm talking to a customer at this very moment, and getting excited about the fact that the things that workers did were able to, I mean, I guess it is, you turned off your reverse proxy network, right?

Your entire product became built on top of workers. What has happened since then?

What does adoption look like? Are people paying for it? Did it turn out to be a good product?

What's the current story? Yeah, so the way we're working on the product is on phases, right?

So we have right now about 900 active user domains in the platform, over 400 and something users, right?

So we started really aggressively in terms of letting everybody so that we can test, stress test the scalability of the system, because, you know, we need a system that anybody of any size of site plug it in, and it will not, you know, always, you know, slow down or cause problems, right?

So, and also make sure, right, that, you know, what if I can go down the glass, right?

Yeah, doesn't matter, because everything that powers the website is running off of cloud, right?

All that stuff took a lot of engineering effort and testing and all that stuff, right?

So, and, you know, and the way that I'm seeing the workers, we're talking about the future, right, is making websites, you think about it, self-healing, right?

That's what I think is, I like to see the parallels between what we're doing, you know, with workers and with rank sense, with what happens with the hosting industry at AWS and DevOps, right?

So what happened when before you have DevOps, before you have AWS, before you have virtualization, you have to do set up the service manually, right?

When you added this virtualization, you can spin up, you know, instances, stuff like that, then in the logical thing was the DevOps, where you're automating a lot of things, it's systematizing things.

So now, I've seen that before, because that's one of my work, I would manage 12 unit servers, right, you know, maybe 20 max.

How many servers of DevOps can manage now? Hundreds, you know, thousands, and describe things in an answerable, you know, YAML, boom, that's what I need.

Oh, something went down, this Kubernetes is going to heal it, I'll heal it, all this stuff.

I see this happening in SEO, I see this happening, you know, because that's what we're building, happening in all the industry, marketing, security, a lot of stuff is going to get smarter, because this capabilities that you're building, they are enabling developers open the doors for these new amazing things that are going to save a lot of time and effort, right?

And this is not about replacing worse jobs, right? It's not about replacing jobs.

This is about enabling you to do a lot more with less time and effort, because we don't have less these admins, we have more DevOps than ever needed.

But now we can do a lot more than before, right? So that put us back into rank sense, right?

So when I think about it, I see that, right? I said, look, SEO, you know, the next layer that we're adding is the automation.

I said, okay, now how all this stuff that we can do with workers, because we don't need people manually touching things, how do we automate these things?

Instead of writing manually title text, meta text stuff like that, or putting rules, how do we use the latest advances on NLP AI to have the machines generate them at scale?

And that's something that we're already doing, I write about it, and I show that I just get excited about it.

But now the machines are doing it, how do you know it's good? That's what comes back to what we do with rank.

Now that you have the ability to generate those stuff at scale, and this is a performance channel, you know, SEO, how do you know that that's better than the work that you do manually?

What can you learn from it, right?

And that's the next phase that we're working on, that we're hoping to have it in a month or two, right?

So that you can do all this stuff, you know, adding these automation capabilities.

Right now, always with the human supervision, because, you know, it's like 50% of your business is SEO, you're not going to trust it to a computer to be making sessions, you need to have somebody overseeing and making sure it's successful, right?

And it's perfect, because the timing, if you look at, you know, I'm not sure how familiar you are with the progress on the MLP community, it's incredible.

What is it? It's incredible.

Yeah, you know, so I, I'm always writing and getting on top of this, I love, I love to know what's possible, similar to what I'm doing here.

Wow, I can use this for this, I can use this for that other problem.

And also always talking to problem to customers, to learn what are the next problems that they're not even considering, right?

So that things get even better, right? And yeah, I mean, you know, yeah.

I was just gonna say, I think we have a few different audiences, you know, there's the audience of someone who might write a worker, and it's probably excited to hear that they can do things like rewrite HTML pages, or they don't have to deploy their own reverse proxy networks around the world in order to make websites better.

And then there's the audience of someone who might build a Cloudflare app.

And they're probably excited to hear that they can build something that will use that 900 different websites can install.

And finally, we have the audience of someone who should install rank sense, they have a SEO driven business, and they have this opportunity to grow it faster than they otherwise would be able to, because they can make their website appear better in search results with what seems like a pretty minimal effort.

And so I would really love to hear kind of what your message would be to maybe the second two of those audiences, like what you would say to someone who's thinking about building a Cloudflare app, and what you would say to someone who's maybe a potential customer for ranks?

Yeah, so talking about the developers building apps, you know, and I wrote an article for Entrepreneur, if you can Google it, you know.

And the way I've explained it is, it's like the distributed, you know, think about the app store when the iPhone was launching, right?

So the first mover advantage of being one of the first apps builds your caching momentum, right?

And for me, it solved me a lot of problems, because I didn't have to worry about user adoption, right?

I got over 2000 and something people try it out, you know, we kept, you know, about half, we're doing work to improve the onboarding, so we can recover a lot of those users, right?

But while we were focused on building and improving all that time and effort, and just the capability, making the product better, we were getting users, you know, without we having to do any marketing or anything, which is for me, it's just incredible.

It's amazing to be able to get that, right?

So no brainer, because it's still early stages, right? We've been in the app store for about a year and a half, right?

Incredible. We know marketing, we're not spending, so save me the money for the marketing side.

This is something that we didn't talk about either, right?

So I have to spend any money in marketing, because they're getting all these users, you know, automatically.

I've been over a year doing public speaking and writing and more to get people comfortable with the idea of the automation and the credibility for us, and we know what we're talking about when it comes to automation and SEO, right?

And the message for people, you know, you know, to try RankSense is, it's more about, look, SEO is important, you know, it is.

Do you want to wait six months to see if it works, or do you want to see it faster?

And the main reason why you're going to be seeing it faster is because of the agile, iterative work that you could have, instead of having one marketer, SEO agency or internal doing the work and a developer implementing the changes, you know, one finishes, you know, and then this one works on all the batches, you can have the marketer doing both things iteratively, right?

And for people on that angle, the easiest way that I explain it for marketers is imagine a Google Tag Manager, but for SEO, where the SEO tags load before the JavaScript, they load, and that gives you a lot of advantages in terms of redirects, you know, and a lot of different cool things you wouldn't get.

And in addition to that, we get the validation.

So RankSense is a validation tool, you make the changes, because you'll say, why would I make changes in the cloud instead of my CMS?

Because you are going to make them quickly and easy, and with RankSense, it will tell you if they're effective, right?

Now that you, out of 100 things you're going to do, maybe 10, 20 are effective, you're going to learn those 10 things, now you take those things to your CMS, that's a significant savings in terms of the effort implementation for the app.

That makes a lot of sense to me. And honestly, I can't imagine why if I had a business that was driven in any way by SEO, I wouldn't close that loop.

And yeah, there's the intelligence on what I actually needed.

One other thing that I wanted to ask you about, which is kind of more self-serving, is you are one of the hardest working people that I know in terms of creating content, which is a terrible way of expressing, writing things, producing things, having conversations like this, speaking.

And you mentioned to me at one point that you don't particularly enjoy doing most of those things, that you do it, the right thing to do in order to grow your business.

And that's a fascinating thing, to have discipline, to run, have started four businesses, and to show up every day to do the things that you think are most important for your business, not necessarily the things that you think you might be good at or the things that you want to do.

And so I would just love to hear in the last few minutes, where that discipline came from, and whether it's a practice that you do, or whether it's just a natural part of who you are.

No, I will tell you that that's, I found a hack. So I'm an introvert, you wouldn't think that I'm an introvert, given all the speaking, and I'm a geek, like a lot of us technical people.

So I have a chip in my folder, in my shoulder, and I have, and I found a hack.

So what is a chip? So quick story, I applied for a merchant account with PayPal, they declined it.

I said, why do you decline my merchant account?

Oh, you guys are SEOs, you're really bad. And they said, where is your list of, you know, they put it in with mail order brides and some other stuff.

So I got really pissed, spend two, three weeks building that. And they said, No, you're terrible.

I don't want to do anything. And I said, Wow, why is that happening?

A lot of bad players in the industry created that reputation, scamming, you know, small businesses, stuff like that.

So I have to do something about it.

And I and the other thing is that I hate this writing stuff. To the other day, I had the opportunity to give a talk to SEOs about Python.

I love writing code.

You know, Python, I find it enjoyable. So I said, maybe I can with this, I can find a way to connect the two things, right?

I Google, I search in Google, how to start a movement.

And I found this video. And I saw this crazy dancing guy say, you know that he started doing this stuff.

And she said, Oh, you just have to put time on the next person.

And I said, That's what I'm going to do. You know, I, I can write code or speak if I don't write code, I can write or give talks or stuff like that.

If I don't have to write the code first, because the code as you can imagine, gets me gets me excited, I'm going to discover something that I want to share, it's going to be interesting.

So every talk, if you look at it, every right, you know, it has code behind it.

And I typically share it. So I do the code first.

That's what my motivation and I do it because I'm going to learn something new.

And when I see that I learned something new, I get excited.

I said, Wow, now I'm going to share it. I'm going to write do a write up, share it.

And when I do the write up, I am trying to inspire the new generation of SEOs to follow my lead and change the perception of the industry because now they're data driven.

So that's simple as that. That's maybe my approach. So I said, and I have to do it consistently.

So that people know that I'm really, I really believe in this and I want to do it.

And that has been, you know, the approach I write code for every talks, every speech.

And I'm inspired because of the people that I'm trying to follow my lead, given this example, at least, but so difficult to actually do every single day.

And so I just have so much admiration for you. And I so deeply appreciate you taking the time today.

How much? Thank you so much. Thank you for having me.

So So that was great. Thank you. A hybrid cloud is a cloud deployment model that leverages two or more types of cloud environments.

A hybrid cloud usually combines a public cloud with either a private cloud, on-premises infrastructure, or both.

An example of a hybrid cloud deployment would be combining the GCP public cloud with the Microsoft Azure private cloud so that they essentially function as one combined infrastructure.

Similar to a hybrid car that combines electric and gas power, hybrid clouds combine the benefits of multiple types of technology for better efficiency, functionality, and price.