Sign up for our newsletter
Receive our monthly round up of what's new with WordPress.
In this video Atarim (Formerly WP Feedback) founder Vito Peleg discusses with Tom Fanelli, CEO of Convesio on how they managed Scaling WordPress to meet the demand of the WP Agency Summit 2021 Virtual Conference.
The platform handled 47TB of traffic and 3,000 queries per second, driven by 500K chat refresh requests and maintaining good level performance.
Receive our monthly round up of what's new with WordPress.
Convesio is not just another hosting provider. Our infrastructure was designed from the ground up for hosting high performance, scalable WordPress sites.30 DAY FREE TRIAL
All right. So we're live. How are you doing Tom?
Good morning Vito, or afternoon. Wherever you might be listening from.
Yeah. Kind of in the middle.
Yeah, that's right.
The past couple of days, during the event, everything seems like a bit of a blur. Do you know how it is?
But I'm excited to be here with you and excited to start Day Three of the Web Agency Summit. So like with every morning, with the intros, we're also broadcasting this out to social media, just so that people can get a bit of a feel of what we're creating here, what is happening inside. And what we're going to do today, or over the upcoming hour, we're going to explore, along with you Tom, we're going to dive into how we're powering this event, because the level of scale that we managed to create, really it's new totally to me.
You probably have seen some things like that before, but I'm sure that a lot of the people that are joining us, within the event itself, and outside within the Facebook groups, are going to be shocked by the number of requests, the number of visits, the number of resources, really, that it takes to pull something like this off, and the level of the infrastructures that we had to put in place to power up the biggest event in the WordPress space, really.
So how about we start, Tom, by, I'm going to give you the space to introduce yourself a little bit so everyone can know why you are the top expert to get something like this done. And then we're going to see some numbers, we're going to go down into the technical aspect of it a little bit.
Not getting too technical, just so that people can hopefully learn from this, and if they're encountering any platforms or any websites for them or for their clients that they need to scale up rapidly, they can hopefully implement some of those things in place. Right after that, we're going to dive into going through the schedule of the day, and to see what we have planned for today. There are all kinds of awesome things that are coming up.
Just to give you, if you're going to look at the title that is right here above me, just so you see a little bit of what's coming for the rest of the day, and just going to do a quick kind of this one. Going to talk about Alexa, we're going to talk about WPaaS, we're going to talk about accessibility, accessibility people inside the project, how to get customers from LinkedIn, how to measure the value of the agency, how to translate the website, but not only how to translate them, but how to use that as a revenue source for your business itself, as well as how to transform your own agency website into a marketing machine.
So that was a quick one, just to get everyone excited about what we're going to dive in and talk about, and speakers and so on. So, Tom, how are you today? Please introduce yourself, and let us know who you are. Yes.
Yeah, yeah. Doing great. I am so excited about the schedule. I've been, of course, glued to the summit for the last couple of days. So I've been in almost every chat so far, and it's been great.
Yeah, so my name's Tom. I'm the founder of a company called Convesio, and we really focus on scaling, high-speed... I'll kind of unpack this, infrastructure for WordPress. And it's all container-based, clustered databases and multiple servers. So it's very unlike traditional solutions where your website might be on a single VPS, or a single dedicated server, or perhaps a shared service infrastructure, and there is a limited number of resources available to that. And I thank that's really where the magic of hosting an event as this comes from, is that many of these events, need too many resources all at once to be hosted on a single machine. And that's the notion of distributing the load is really important.
But a lot of people don't really understand all of the stuff that goes into this, and one of the things I was telling Vito is: We get asked a lot, "how much capacity or servers do I need to host an event of X people?" And Vito, as you know, every website is different, and every website requires different resources. So one of the things you've got to do effectively is planning, right? You've got to do load testing, and you've got to look at, "Did we have this event in a prior year? And how did the performance of that event go?" Because you don't want to overpower and have more incurred costs.
And you definitely don't want to be underpowered. I suppose if you have to be one or another you might choose to be overpowered. You don't want to be underpowered because if things go wrong in the middle of the summit, you've got a situation where sometimes resolution to those things is not fast, right? I mean, worst-case scenario code change has to happen, because something is inefficient at scale, and you didn't realize it because you didn't do any load testing.
And so we planned for weeks before, and for the week leading up to the summit, we were doing load testing and all sorts of things to make sure that the performance of the platform didn't have any gotchas. And so I'm going to share some stats, and Vito I'd love for you to chime in on this. Let me just share my screen here. And I'm actually going to present.
So how about we get into the game of... First, let's start with... So you talked a little bit about the planning, how we're getting that. So now you're going to talk a little bit about how... Are you going to show us a little bit about load testing and how that came into place?
Yeah absolutely. I can definitely do that, and we can go in any order that you want. But let me just share my screen here, and I'm going to...
Let's show some big numbers so people can really understand what we're dealing with here.
And really, I've been building websites for more than a decade, or really almost 20 years now. I'm underestimating how old I am. A little more than 20 years now. And I've never had to deal with anything at this level of scale, and not only what we did here for the event but what we're doing together for [inaudible] as well, which is the platform through what we're running the event from. So I'm going to allow you to share your screen so we can get some of these numbers in place.
Can you see that as a full screen? Or are you looking at mine?
You can see what we are seeing, it's black bars on the side, and I can see.
Okay perfect. Yeah great. Okay. So yeah, so just to level set where we're at in the first two days of the conference. And I was pulling up this data, and this is edge data from Cloudflare, so we'll kind of bounce around and talk about the infrastructure we're running and a few things. But this is really the highest level view of the performance of the conference. And in two days, we've had two million requests processed, and we've had about 6000 visitors from 91 countries.
And Vito I think that's a huge testimony to justify the huge reach of this. We were on a chat in the lounge with someone video chatting, and they were expressing, they were I think in Pakistan, expressing how happy they were that this was available to them because, with COVID and everything going on around the world, there just wasn't the availability of WordPress, meetups, conferences, and all of that stuff.
Even in specific... We're very fortunate to live in... I'm in the UK, you're in the US, and things come here. Things come to these countries. But I remember when I was living back in Israel, and a lot of other places around the world, these things just don't happen. You're just constantly being left out of the game just because of your geographical location. So I agree, and this was one of the things that we tried to push toward.
So even though that we can see that the top countries here on the list are English-speaking countries because the summit is delivered in English and our circles of networks and so on are mostly in English-speaking countries, we still manage to serve 89 additional countries on top of these ones so that people from all over the world can come in, learn from what's going on, and what are the industry standards that we're doing right now, stuff that really they usually couldn't do, and I couldn't do when I was back in Israel as well.
Yeah absolutely. And I think that that's... I mean this is data that I think a lot of people should take note of, whether you're looking at sponsoring an event like this, or you're looking at participating in an event like this, or in any way getting behind it, that the ability for this to scale, to reach the audience that it reached, I mean yes you can have 6000 events at a physical event for sure. But you're not going to have 6000 people from 91 countries most likely. And so the ability for this to scale to this kind of reach is, I think, really... I was actually really surprised when I saw that, because I was like, "Wow that is a number from the..." It's a pretty big number for two days.
All right, well jumping ahead, I'm going to skip a couple of slides. I'm going to come back to them, but we talked about load testing, okay?
What is load testing? Just [crosstalk] this.
All right, great. So when you put up your website, there's going to be a certain number of people that can hit that website before the server is going to run out of resources. Okay? Now sometimes that number is 10, sometimes that number is 10,000. You don't know because every hosting platform and every website is different. In load testing, there are several tools that allow you to do this. Basically, this is a way where you can simulate visitors coming to your website and browsing around various pages.
And there's a lot of different tools out there. I have a couple of favorites I'll share with the audience. So the first one is a product, and I've got some charts from it on another slide, it's called RoboSwarm. And I think the URL is roboswarm.dev. And the guy there, his name's Jack. He's a great guy. He has another product called Kernel in the WordPress space that does plug-in updates and all sorts of stuff.
But that product is designed specifically for WordPress. So you can do some really cool things with that product, like simulate WooCommerce checkout. You can simulate logged in users browsing your site, which was really important for us because one of the things that's a big load on the site and breaks caching, and we'll get into that in a moment, is whether or not people are going to be visiting and browsing your site anonymously, not logged in, or whether they're going to be logged in.
If you have to log in 10,000 people, that's a lot more load on your server than people browsing cached pages. So this concept of load testing, we ran several different scenarios in load testing. And so this one here is no cache, no CDN, but no people logged in. So we ran this, and if you look over here we did 3000 clients over five minutes. That had an average response rate of 925 milliseconds and no errors. And that's what you want to figure out.
Now one of the things you want to look for when you're load testing is: What is my breaking point? So it's not uncommon that you'll see people test. And when you do a load test, you have to make assumptions, right? If those assumptions can be rooted in reality, that's a lot more helpful. Vito, we didn't have, because we weren't working on the summit last year, we didn't really have exactly the data that we needed, that we have this year, because we've been watching it. So we had some Google Analytics data, we had some CloudFlare data.
[crosstalk] last year. But really, now that you're mentioning this, last year we didn't go as deep as we did this time, a lot with your help Tom. And what happened on the first day? We were shocked that we got 220,000 requests in an hour, and the whole thing just crashed. And so this time, we came prepared and everything was nice and smooth through the transition. And really, when that happened last year, this is above my pay grade, all of this stuff.
So what I ended up doing is I just hired... I'm relying on the community, so some sponsoring, you were sponsoring as well last year, and the guys from [inaudible], and got some people from AWS to work together with some of the speakers. So luckily we're in the right community for something like this to happen in real-time, and everyone just chimed in and helped us get back on track.
But really, what I was doing was I was running behind... I got out of the seat and I was just running in circles. So now, understanding, or after learning about: What are all of these things, and where things can break down when you're scaling things up unpredictably, which I might say, allowed us to prepare a lot better through load testing and through some of the things which you will show us now.
Yeah absolutely. And I remember that, and I felt your pain because I talk to a lot of people, and I've certainly been involved in things prior to Convesio where stuff has gone down and people underestimated the load. And there's only so much technology can do, right? It's like, if you're not expecting and you're pleasantly surprised and you've got 10X, you're probably going to be stressing a lot of infrastructures that you might've planned that you thought would be really good. And by the way, just as a side note of levity, part of the thing that made this event stressful for me was the fact that this is all of our peers, okay? So it's kind of like, "Okay we have to make sure this goes off with any issues."
I was obsessed, personally, with making sure we had the right stuff in place for this. So far, I think we've had great results. And part of the thing is being able to know what's going on. I think last year you didn't have the visibility that you've got this year, and we'll share some more of that.
But the other thing we look for, going back to load tests here, is how many sustained users can we have simultaneously? And one of the metrics that are really important on this is your requests per second that the site can handle. And this is very, very different than if you were caching data. If you're caching data, you can serve that data extremely fast, with very high throughput. But if you've got to process a bunch of database requests, then you have to make sure you have a scalable database layer.
And so one of the other tests we ran, and this was where we said, "Okay let's put a couple of hundred users simultaneously in the system just aggressively browsing continuously. Every second requesting a page. Okay?" We had, I think, a path of like five pages that we thought would be primary pages, and they're the pages like the main session, the lounge, this, that, and the other, the exhibit halls.
Registration, the login.
Yeah, registration. Booths. Yeah. So we ran that test, and we got to the place where we're like, "Okay, we're processing 123 requests with no caching, no layer of optimization, no CDN at this point." And we're like, "Okay, this is performing pretty well. This is very unlikely that..." And the performance, I don't know if I have charted on this, one of the things we looked at was: What is the experience of the load time of the pages? As you can see here, they're still in the hundreds of milliseconds here. So the whole point of this is: Can I have a good performance window? I mean if the site takes 30 seconds to load a page, that's a fail. You might not get any failures, but things are becoming so hard for users to navigate that they get frustrated.
So you have to watch, "Okay, what are my response times? And what are my failures?" Because you don't want errors, and you also don't want users that are frustrated because it's taking a long time to load. Those are really the things that we look for primarily in the end-user experience. And so part of this load testing scenario was pushing the envelope and seeing: What's taking a long time? What can we cache? We also went in because as we started to layer things like a content delivery network on top of this, we had to go in and make sure: Is everything working? Does chat work? Do people's avatars update? There are all sorts of things that caching in a logged-in environment can cause difficulties with. Now we took special care to make sure that wasn't the case, but we definitely have to test that as well.
Tom, what kind of tools are you using for this load testing?
Oh, so yeah. So there's RoboSwarm, and it's roboswarm.dev, and I think that's really good. This is a chart from there. And then this is a screenshot from loader.io, and loader.io is a little bit more... It's not complicated in the sense that it's hard to use. It's just not as customized for WordPress. So WordPress is by far more compatible. RoboSwarm was really designed for WordPress. That's their niche, and Jack's done a great job. So if you want to test commerce checkouts or things like that, logged-in users, far easier to do on RoboSwarm than it is to do on loader.io, in my experience. Yeah, so that's these tests. So at the end of all of these tests, we kind of were like... We had some infrastructure we had rolled out, and I'm going to jump to that real quick.
So this is a screenshot of the Convesio dashboard, and I had mentioned-
Brand new Convesio dashboard, right? You're just releasing this.
Yeah, this is going to be new. This is actually going to be... This version's going to be live. You guys were absolutely the first people to use this new version of Convesio, and this is going to be live next week for general use. And what we did, and this is where I'll dive a little more into some of the technical stuff, just so people understand. And Vito, you know this, right? This is the problem we're trying to solve, right? If you have to hire a sysadmin, dev-ops person, to stand up for you multiple servers, load balancers, figuring out how to run Docker, WordPress, and Docker, a clustered database. I mean you did that for some of the video streaming technology. I mean, what was that like, Vito, when you did that? That's not an easy, quick process, right? And it's not cheap either.
It's not cheap, really is the key point here, being your own sysadmin, or hiring proper sysadmins. But also, it's a matter of availability, because with what we're doing right now, really every minute counts. We have thousands of people walking around the platform at any given moment. If the platform crashes for an hour, you lose the trust of thousands of people instantly.
And so when it comes to running events, or even when it comes to running e-comm sites, you want to make sure that you have someone on call. And if you want to make sure you have someone on call 24/7, then it's not just hiring the one guy, right? No one can handle that. You need to build a team around this. And really to be honest, because this is not really my forte, I'm more on the website side, I'm more on the business side, I'm more on the marketing side.
So getting into the game of dealing with AWS and the different containers, load balancers, and all of those different things, it's just way beyond the things that I want to deal with, really.
Yeah. And it's a scary place to be when something goes wrong with a setup like this because it's so complicated if you were to hand-roll this on AWS or Google Cloud. It's so complicated that, for the people who are our audience, who are the non-technical site creators, builders, marketers, who just need to have the ability to scale a site for an event. "Hey, I'm doing a big email blast. I want to make sure I can handle it." Or, "Hey I'm getting featured in some really big TV show or social media outlet, and I want to make sure I can handle it." How do you do that? It's like upgrading your site, like "Oh I got to go change packages.
I might have to migrate to a new server." There's a lot of complexities in doing that with existing hosting providers. And this is really what we've made dead simple, like a few pushes of the button, a non-technical person can have all of this type of infrastructure running for their site.
And what we've done, just to go back to the summit here, and what we're running is: You're on a dedicated cluster for the summit. It's running five C2 instances, those are compute-optimized instances on Google Cloud. Each of them has 16 cores, and there are 64 gigs of RAM. There are five load balancers, so each server has a load balancer. There are 10 WordPress containers, and that's an image of this right here.
These are all 10 of your containers, and you can see by looking at this what the CPU load is on each one, and you can see that we've pretty much-balanced requests almost evenly between all of these different containers. This is something that takes tens and tens and tens of hours of a sysadmin to set up a load balancing configuration.
And oh by the way, if you don't do this every day, there are tons of little gotchas that you can run into that are tricky to figure out and to work the kinks out of. And so all of this is the front end of it, and then we have a Percona XtraDB cluster, which is a master cluster that's load balancing all of the database requests.
So we're dropping terms here left and right, and I can understand that some people might be like, "What?" So what are load balancers? Let's start with this one.
Great question and I'm going to give you an analogy, and I've been working on this pitch for a while to explain this. Because it's hard when your closest friends and family don't totally understand what you do. So imagine you've got a cup, and you're pouring water into it. And the cup is your server and the water are your website visits. At some point, you reach the maximum capacity of how much that can handle. So what starts happening? People start overflowing and water starts spilling. Well, what Convesio does is we have the ability to spin up another cup for you, okay? And a load balancer is what distributes the water from... It knows this cup is full, and it says "Start using the other cup."
And so a load balancer sits in front of your website, and it distributes the visitors to the different cups. And as each cup fills up, instead of it overflowing, which would be your visitors getting a really slow website, or your visitors getting an error. Instead of overflowing, we basically have another cup spin up for it.
So here we have actually five cups at this point?
Well, we have... Yes. We have five cups. Those cups could be your individual servers. More specifically in our platform, a cup is a container. So a container is what runs your actual WordPress, your runtime of your WordPress. You have 10 cups. Let's not over confuse people with the other level of the server. But yes, that's in essence what happens, and that's how a load balancer works is: In Convesio your site can clone itself to all of these different servers, and the load is distributed among them evenly.
And is this something that happens automatically? It's going to go up, it's going to go down? Or is this something you need to monitor and be aware of?
Yeah, it depends. You can do both. So in Convesio, you have the ability to manually scale your site up with the click of a button, or you have what's called auto-scaling. And auto-scaling will give you the ability... And this is all in the new release we're pushing out this week. Auto-scaling gives you the ability, we have a screen now, where you can control it, and you can say, "If my CPU instance gets over 80% for five minutes, add another container to the site." And that all happens automatically.
Okay. And then, around this concept. I know that this was a question when I had when I started getting into the game of load balancers. Because in most cases, or from my experience, I was mostly building small to medium businesses websites. We didn't need that level of resources for the vast majority of projects. And I can also tell you that we did need this for some, but I just didn't know about this when we needed it.
But something that really came to mind is: What is going to happen to the WordPress website that you're running? Can I make changes? If it's distributed over different websites or different servers, really, what is going to happen if I make a change on one? What is going to happen if I'm going to update the plug-in from my back-end? How is that going to reflect throughout?
Yeah, great question. Really good question actually. We're getting a little in the technical details, but there are two ways you can scale on Convesio. So we can scale your WordPress resources, which is this container. We can do it on a single node of a server. So let's say you've got a container running. Think of it as a micro-VPS for your website. And we can scale that on a single container. So maybe you get a surge of traffic, but it's not astronomical, right? It's just more than your resources allow.
Your site scales up to two containers. That's totally fine to make edits and changes, okay? You can make that all in that one server. Because the files don't need to synchronize to other instances across the cluster. But if you're running, like you were running in a situation today where you're across many, many clusters, and we're having high-traffic events, then we want to lock your codebase, okay?
Now this isn't your uploads folder, okay? It isn't that type of stuff. It's the codebase of your CSS files, your plug-ins, and all that. We don't want to be in a scaled event where you've got potentially millions or thousands of requests coming in per second, per minute, what has it, and you're trying to push code updates to your site because that's not... We want to try and stay away from that. That can be unpredictable.
That's really what we did here. You let us know, "All right, we're going to scale things up before we open our doors." And we went into code lock. We have our little staging environment where if there is something, we can play around, check things out. And then if we need to deploy any piece of code, really the process of scaling down that to a singular container, and then bringing it back up. Is that correct?
Yes, that's exactly right. Yup.
Okay. So we got to now understand the distribution of the infrastructure... Yeah, please go on Tom.
I have one more thing. You brought up a really good point, which is: What happens... Most people, don't need this, right? The challenge is that when you need it, sometimes you don't know you need it. And when you need it, if you find out during the event, it's kind of too late.
Too late, yes.
Because you're trapped on this thing. And so what we've done that we think is innovative here is this notion that you can have the ability to scale up any time. When you don't need it... In fact, you can totally disable auto-scaling. So if you're an agency and you're like, "Well I have fixed costs per month, where I charge my client X and I can't go back and charge them more. So I don't want to scale on because I don't want the idea of an unpredictable bill, because the client did whatever." You can totally disable it.
And so the thing is, and what we're doing that we made really smooth is: You don't need to change billing plans. You don't need to upgrade, no need to move your site. In fact, you don't need to do any of that stuff. If you know you have an event and you talked to your customer, you can literally turn auto-scaling on, or scale-up manually, and you can run it like that. And you only pay for the resources you use.
That's great. So I have a question here from [inaudible]. Before we're moving on to the next part of this session where we're going to be reviewing the talks that are coming up today. But let me get your take on this. Will you guys be supporting developments or deployment with Trellis? A bit of a technical question here, right?
Yeah. So we don't today. But I will tell you that... And this person sounds like he's pretty technical, and so I'm sure he knows about Docker and all the wonderful things that you get as a benefit of that. And our plan is that today, we deploy a very specific WordPress container. Well, actually we have many different containers based on configurations and all that. Part of the magic of being a containerized platform is if I say "You know what? I want to deploy a container configured for a head list," or, we're enterprise Cloudflare partners and we leverage workers really heavily.
So we're investigating things like, "Okay, are there were ways to do static websites where we can just push your website completely to the edge of Cloudflare and turn off your containers so it can serve statically at the edge?" There are all sorts of future projects like that that we are investigating, and absolutely using frameworks like Trellis, Gatsby and all that, are extremely interesting to us, and we're going to get there.
We officially launched our product to the market in January of 2020, and so we're a little over a year old, and we had been in R&D for a couple of years prior to that. Because as you can imagine, we've been solving challenges here that aren't very simple to solve. In fact, the new version's finally going to support multi-site. So yeah, we're adding many new things, and we're definitely going to be looking at those things in the future.
Okay. So one more question, one last question, from [inaudible]. Does your support provide hands-on help in optimizing WordPress websites? Are you going to go into the website and see what's up?
Yeah absolutely. For sure. We'll definitely help with that. I want to share one more stat, Vito. Where is it here? So this is just to give you some sense of the scale of the conference. So internal network traffic on this cluster has been 47 terabytes in two days. We've peaked out somewhere, last time I look, around 3000 database queries a second, and you've had over half a million chat refreshes. So yeah, I mean you can see the sheer number of data that are served through all of your containers, right? All the disc usage is being pushed through this here. So this is exciting stuff for us to geek out on, but I just wanted to give people a sense of scale.
And quite frankly, this is the kind of stuff that, just gets hard to do unless on one machine. The reason that we like this distributed approach is: We get this level of scale that's hard to do on one particular machine.
Okay. And this is amazing, Tom. And looking at these... These are insane numbers, just to really understand what we're dealing with. And to get to a place where everything is nice and smooth as it is inside the venue when people are jumping between the lounge and the live sessions, or between the different booths, is a huge undertaking. So I want to thank you for helping us with this.
And now I'd like to bring in Andrew to join us. So how are you doing Andrew?
I'm good. I'm good. Tom, just ahead was blown off by the situation with watching all of this activity going on. And I know we're working backstage, and there's some stuff going on, and we've got lots of things that we have to do, almost instant production stuff, and adding banners and doing all sorts of... Answering questions, making sure the sponsors are okay, making sure the networking lounge is okay. And if we didn't have a fast interface or a fast website to do that, we'd be stuffed, basically. I mean it's astounding that we've transferred 47 terabytes of data over the last two days. I mean it's mind-blowing, isn't it? It's crazy.
It's exciting stuff. Yeah. And it's awesome. I'm so excited to see the success of the summit.
So, without further ado, let's just dive in to see what's coming up today. So today's Day Three of the Web Agency Summit, and as you have seen from the numbers over there we've managed to maintain our position this year as well, as being the biggest event in the WordPress space. A lot of this is only possible thanks to our amazing sponsors, the amazing partners, the fact that people are sharing the word, the incredible speakers that we manage to gather throughout this event.
So I'm going to share my screen here so people can really see who we're talking about. So that includes GoDaddy Pro, WP Engine, GridPane, WordPress.com, High Level, Nexess by Liquid Web, GroundHog, Elemental, Shield Security, WP Activity Log, Termaggedon, Weg Lot, Cloudways that we just saw Sam from, Fixed Runner, Yoast, and of course, the kind supporters which is Convesio and [inaudible], that's powering the platform itself.
So now let's see what's going to happen today. So we're going to kick things off in the morning by talking to Thibaud from Weg Lot, and he's going to teach us about generating additional revenue for your agency through upselling, or through selling multi-lingual websites. And I'm just going to direct your view here to the categories that we have here. So we have four categories within the summit. It starts from the build, then we have expanded, scale, and thrive.
That's why you're going to see another session about multi-lingual from the building aspect of it. But when it comes to expanding we're going to mostly be talking about the business side of it. So this is about how to build a website for global audiences through multi-lingual for your clients.
So we're not really focused on building websites for ourselves in most of the sessions. We're focused on delivering websites for clients, which is a different mindset completely. It's a project that has to have a beginning, middle, and an end, as opposed to, as Tom was talking about, building their own platform. You have two, three years of R&D. You can't do that with clients. Get the job done and move on to the next project. So that's why the build category's focusing on that aspect.
And then expand is using what we learn on the build category to expand our business to generate an additional revenue stream. And I can tell you that within my agency, upselling additional languages was always an amazing idea, because people when they need an additional language, deploy all kinds of different tools like Thibaud is going to talk about here with Weg Lot. But this doesn't really mean that you need to translate.
You need to manage the ability to translate multiple languages, and you can create that for additional languages. So a website that started with one language, can upsell additional revenue for every additional language that is being added to it. So imagine a 3K project. If you add one language for another 1000, but then you have seven languages on there. Then you can see how you started with a 3K project, then you moved over to more than 10K just by implementing what Thibaud is going to teach us here at 10:00.
We talked about the build, with Tanya, how to build websites for your clients that are for global audiences. And we talked about Thibaud with expanding, and I'm going to go right here in the middle with scale, by Melissa Rachel Love, that is going to be showing us around turning your own website into a marketing machine. So that's why it's coming through the scale category so that you can really generate a steady stream of leads.
And Andrew, I'm going to direct this one to you, because I know you worked with Melissa. She actually helped you with some of the things on your website as well, right?
Oh yeah absolutely. We're kind of a to and fro. I help her with some technical aspects on her websites, and speed optimization, and server issues, and stuff like that. And she's helped me with marketing, certainly when I had elegantmarketplace.com. She's the first person in the Divi space to make a Divi child theme. So I call her the Queen of Divi. She's an amazing marketer. She has some talents that we all wish we had, like what she's got in her little finger we need in our whole body.
She's just an amazing person. She continually helps people. Part of the thing that Melissa does really, really well is building communities. So her advice on building a community around what you're doing, where you're going, and why you're going there. And lots of free webinars, lots of free Facebook Lives. And she's an expert website builder, UX, and designer. She uses tools like Pro Photo and Divi and Element. She's a real all-arounder on that.
But really the marketing fixes that she's got going on, if you don't join that then you're missing out. There's a lot of free content, a lot of education there. It's great. She's a great person in the community.
Awesome. So actually, if I'm looking at the schedule today, we have a whole lot of ladies, and I love that. And this is actually something that we proactively tried to make sure of. So there are three white men right here on the session that is talking to each other for a few days. So a lot of diversity happening today, and we can see. And I'm going to move over to Kristina here on the thriving category. So we already looked at the four different categories of the summit throughout the upcoming four hours.
And then thrive is when you already have a sustainable business, that you learn how to do through the expansion and then the scale categories, then we need to make sure that the business actually thrives. We understand what it means to work with the employees, what it means to work through the... How we can estimate, or understand the value of what we're creating here. What does this really mean to build an agency of 50-plus employees, as we've seen on some of the other sessions?
So what Kristina is going to focus on here is measuring the value of your web agency. And Kristina just sold her agency just a few months ago after building it for a few years. She's going to take us through that journey, what that really means. And Andrew, I know that you also sold a few of your WordPress-related businesses.
Actually, I did, and I sold Page [inaudible] cloud and layouts cloud. So it's love. I just, "Said you want to buy these?" And she said, "Yup." Didn't even discuss the price. She said, "Yup, I'll have those." And I sold Elegant Marketplace through Web Ventures, the lovely Mike Demo, and a few other people.
He's also talking tomorrow, right? Do we have Demo here?
Yeah, Demo's great. He's the lead handshake of Web Ventures, does all sorts of deals. And they've just done a massive deal on RAMNode, I think. They bought RAMNode. Tom, you'll know who RAMNode are, I'm sure.
Yeah. Yeah absolutely. And I can't speak highly enough about Kristina. She's amazing, and one of the things that are really interesting, and I'm sure to share this, and if she doesn't we should ask. But there's a lot of people out there looking to buy agencies, and a lot of time it's like: Where do you go to find these deals?
Because there are people with cash to make deals happen, and a lot of times it's very unstructured, and there are certain websites you can go to to find these folks, and of course conferences like this and being in sessions like that. So just the networking of knowing Kristina, since she is the thought leader in that space, and knowing Mike Demo, to be able to connect with people who might want to buy and sell is really valuable.
Yes, I agree. And then right after Kristina, we're going to dive into discussing LinkedIn with Nicole. So Nicole already joined us, Andrew, and Stephanie, our cohost, in the initial kickoff of the event, and we kind of spoke about this a little bit. But the idea is, and you can see that this one is inside the expand category. The point behind this is, really, after you build you want to build a business, right? After you build a product, you want to build a business.
And so how do you find those perfect clients? How do you define who is the perfect client that I'm looking for? And so Nicole is going to explore LinkedIn as a method of getting noticed by them so that you don't need to chase clients. They actually come to you because they understand that you are an expert in what you do.
And then everything here is going to be discussed with organic outreach. So no need to spend money on ads or anything like that. This is a really interesting talk, especially because I know that we didn't, in my agency, we got to a place where we were generating 60-plus meetings, not even just leads, but straight-up meetings, every single month directly from LinkedIn without paying a dime on ads. I'm really excited to share this knowledge, or for Nicole to share this knowledge, and to dive into this topic.
Then, we have two sessions in a row around accessibility, and you can see how one of them is inside the build and one of them is inside thriving. And the reason why we try to group them up or create them in this way is that first, we have to understand how to build accessibility into our website. This is such an aspect that has been... It's not discussed enough, and it's actually becoming part of the law, isn't it Andrew?
Yes, it is. Definitely. Well, it's been part of the law forever. I mean, it's something that we've all ignored for years. So you can, especially if you're a... Governments tend to go for the big corporate. We had the issues with Domino's in the US not having accessibility because blind people couldn't get what was in the pizza, right?
Yeah. Couldn't get a pizza.
And at the very basic level, that's alternative text and making sure that your website is readable by stream readers. I've got a blind friend of mine, which she's lovely and she helps me do accessibility from a blind person's perspective. And also if you have sound and video and stuff, you've got to think about deaf people and you've got to think about people that actually use the spoken word. And what's interesting later on in the day or tomorrow, there's Ask Alexa.
Sometimes you've got people that actually speak to their computer to guide themselves around the website. So you may have to think about what kind of text, how you're dealing with tabs and all that kind of stuff. And not using the website by using a mouse so you can actually use the tab keys to get around the website. So there are lots of things inaccessibility that you can do that are really quite easy, but to be 100% accessible is almost impossible. So these agencies accept that, but-
Maybe. Let's see. Maybe Amber is going to teach us.
Maybe. Let's hope. Let's hope. But we've done some very accessible websites and hopefully, 99% of our websites are very accessible.
Beth Hannon, who's a big accessibility person, told me once, she's like, "Marketers will spend tens of thousands, hundreds of thousands of dollars trying to increase their conversion rates by fractions of a percentage, and 7% of the population has some sort of a disability." I think that's the stat. So it's like you can tap into a whole... There's absolutely a cost and profit and conversion rate argument to be made here with accessibility. So excited to see that.
And this really directs us into the next session about accessibility. If you don't want to do it because it's the right thing to do, do it because it's the profitable thing to do, or it can actually cost you a lot of money in the long run as we've seen with companies like Dominoes. And I just believe that it's going to happen more and moreover in the upcoming years, both on the privacy side as well as on the accessibility side. So then we're going to talk with Colleen from [inaudible] Graphics about how ignoring accessibility costs you and your client.
And you can see that this is inside the thrive category because, as you said Tom, people are spending tens of thousands of dollars, hundreds of thousands of dollars, to move the needle by 0.1% of the conversion rate when it comes to bigger websites, enterprise-level clients. And that's why we included this in the thrive because that's where this impact can really make a whole boatload of difference when working with bigger clients or with high-traffic websites.
So next up, we're going to go back to the scale category with Matthew Rodella. Matt is going to talk with us about why every agency needs a WPaaS. So if you guys are not familiar with a WPaaS model, this is Website as a Service. Everyone knows what SaaS is, Software as a Service, but WPaaS is essentially Shopify, or Wix, or what's coming up by Elemental Cloud. WordPress is the biggest WPaaS in the world, really. And you have Squarespace and massive, massive businesses that have been built based on the model of scalable deployment of templated-based websites. And that's what a WPaaS is, and we're going to see what Matt has to say about this, and why every agency needs this.
And I can even touch a bit of the why. I know that I had a lot of leads coming in, and my mindset was "I don't want to do the small projects." You have someone coming in, and "I only have 500 bucks. I only have 1000 bucks." So my answer to them was, "I guess we're not a good fit. Here is a link to Wix, and talk to me in a few years." Right? Which they never did. So instead, why not generate recurring revenues from, those lower-tier customers that can't really afford the custom-builds that we wanted to deliver. Low maintenance projects that are self-deployable. So Matt is going to talk to us about that.
And this really ties into what Chris Lima was talking about yesterday, Andrew, around... What were the names of the concepts?
Well, Nexess have got their own WPaaS going on, which is a WooCommerce store. They're using, is it AI? I don't know. I've used it, and it's really easy to use.
Yeah, but he was teaching around the concept of the-
Yeah, there's the ladder there. But just to say that there's another thing out that Nexess do, which is their own WPaaS, which is they set up a WooCommerce WordPress website for $20 bucks. But actually, you can do that, follow Matthew's advice and you can do that yourself so you can become a Nexess. You can become a Wix. You can become a Shopify. But the ladders that he was talking about were: Don't miss out on the opportunities that customers present themselves to you.
Even if they're not within your current mindset, or you're remit of doing... You want the over $1000 or the $2000 website, if you've got the opportunity to point them into a direction where you want them to go, take that opportunity, even down to the fact that the guy wanted Facebook developed. A lot of customers have come to us and said, "I want a networking thing. I want a Facebook. I want a LinkedIn." Say, "Well I can sell you an e-book for 100 bucks because you've only got $500 to sell on this thing."
So if you want to catch Chris's talk you can jump into the replays and you can find all of the replays yesterday here, from Ben Pines from Elemental, Chris Lima from Nexess, Miriam from Stratic, Adam from GoDaddy Pro, David from WPEngine. We have Roxanna from Social Bee, Marieke from Yoast, the CEO of Yoast. We have Alex from CloudWays, Chase from High Level. And then there's been a really interesting discussion about hosting by a panel from WordPress.com. So these are the sessions that we had yesterday.
And finally, before we jump off because we just ran out of time, Chip is going to talk about the future, really. So we can see here, going back to close the day with the build category, and Chip Edwards is going to talk about how to incorporate Amazon Alexa, or smart speakers, into your build. And I think that this can have really interesting potential for additional upsells, even for existing clients that want to go into this space. Maybe even... Not maybe, but even local businesses that you can tell your smart speaker, "Where can I find the local flower shop? Where can I get it... Let's order some pizza." So it's going to find the businesses that are within the radius of the smart device. So Chip is going to talk to us about this concept.
So we're going to jump right into the discussion with Thibaud. If you're watching us on social media, please come and join the event. All you need to do is go to atarim.io/summit. We have today starting right now, and tomorrow is the final day before the after-party at the end of Friday night where we're all going to be hanging out in the lounge together. So come and join us here. It's completely free, and you're all invited to join the party. And for those that are already on the live sessions with us here today, thank you everyone for starting to jump in. And we're going to jump right into Thibaud's session next.
Thank you guys. Thanks Tom. Thanks Andrew.