Today, I’m talking to Anthony Casalena, the founder and CEO of Squarespace, the ubiquitous web hosting and design company. If you’re a podcast listener, you’ve heard a Squarespace ad.
I was excited to talk to Anthony because it really feels like we’re going through a reset moment on the internet, and I wanted to hear how he’s thinking about the web and what websites are even for in 2023.
If you’re a Vergecast listener, you know I’ve been saying it feels a lot like 2011 out there. The big platforms like Facebook and TikTok are very focused on entertainment content. Twitter is going through… let’s call them changes. People are trying out new platforms like Instagram Threads and rethinking their relationships with old standbys like Reddit. And the introduction of AI means that search engines like Google, which was really the last great source of traffic for web pages, just don’t seem that reliable anymore as it begins to answer more questions directly. It’s uncertain and exciting: a lot of things we took for granted just a couple of years ago are up for grabs, and I think that might be a good thing.
Anthony founded Squarespace in his dorm room in 2003 — and over the past 20 years, he’s seen a lot of web ideas come and go. My questions were pretty simple: why would anyone even make a website in 2023? He told me that right now, a lot of Squarespace clients think of Instagram and other social sites as their homepage — and they bring people to their websites just to complete transactions because they have more payment options on the web. That’s a pretty huge shift in thinking about the web and what it’s for.
The other huge shift is thinking about where all the content on a website might come from and how much AI-generated content might pollute the web. It’s already happening — and Squarespace is in the mix, with new AI tools for generating sites and copy with OpenAI tools. Is that good for the web? Is that good for business? Is it good for people? I think these questions are pretty open, and Anthony and I got into it a little.
Squarespace also just made a pretty big acquisition, buying Google’s domain registration business, which will make it the fourth-largest domain name registrar on the web. I wanted to know how a deal like that goes down, how it works on a technical level, and, of course, how Squarespace is structured to support it.
I love talking to people who’ve been building on the web for this long, and Anthony was no exception — we had fun with this one. Also, I think this is the most we have ever talked about pressure washers on Decoder.
Anthony Casalena, founder and CEO of Squarespace. Here we go.
This transcript has been lightly edited for clarity.
Anthony Casalena, you are the founder and CEO of Squarespace. Welcome to Decoder.
Thank you. Pleasure to be here. Thank you for having me.
I am really excited to talk to you. Squarespace is one of the OG web companies. It seems like there’s a few parallel revolutions going on with the web. The social platforms are all changing. Some of them are even in crisis. They’re not sending traffic to websites anymore. Something’s happening with Google and AI and how they’re going to send traffic. And then AI itself, if there’s a text box on the internet, people are shoving AI into it, and it’s going to flood us all with stuff. It seems like a lot of things are changing around the web, around how we think about the web, how we might navigate the web, and why people might even make websites. You’ve been at it for 20 years with Squarespace. How are you thinking about all this change?
We celebrated our 20th anniversary in April. So we’re used to a web, pre-social network phase, almost pre-YouTube, pre-iPhone. The predominant browser was Internet Explorer. So we’ve seen a lot. Blogging was a word I used to have to explain to people what it meant when Squarespace launched. So we’re no stranger to change on the web. It’s with that that I’m actually super excited about what it means for the future. When Squarespace started, publishing on the web was an intimidating thing, so we started as a blogging platform because starting a blog was easy.
So from that, over the years, as browsers got more sophisticated, we transitioned into more and more graphically rich websites. A lot of portfolio websites and artist websites started on Squarespace about a decade ago. Since then, we’ve been in an era of the proliferation of a lot of different types of commerce on the web and especially commerce that’s in the hands of people who couldn’t have built an online store, a services-based business 10 years ago on the web because technology’s too difficult, but now you can use the web for all kinds of things.
“…having a space that you own on the internet right now that’s authoritative is almost more important than ever.”
So I think having a space that you own on the internet right now that’s authoritative is almost more important than ever. This is your online real estate. You have a domain that you own. Squarespace doesn’t put anything on your domain or website that you’re not putting there. We don’t monetize through ads — nothing like that. And it’s a way to transact. So Squarespace supports a myriad of ways to transact, from selling physical goods to selling services to booking appointments. We’ve acquired companies that let us get into the hospitality space and with reservations.
So a lot of what we’re focused on is, one, fundamentals, just being the best place to go for a website in terms of ease of use and expressibility, but also really helping our customers make businesses, helping them transact and really being part of the future of entrepreneurship.
So that’s a big spread. You start with, “I want to have a business.” You sign up for a Squarespace account. You set up a website. You’ve got to figure out how to get some traffic to it, which we should talk about. Then somewhere down the end of that road, you’ve started a restaurant, and you’re using Tock to manage reservations and bookings and stuff, and now you’re inside the walls of the business. You’re running some of their core functionality. That’s a big spectrum. You start with, “Okay. This is a marketing platform,” all the way to, “Now you’re running your business.” Where’s your focus?
It really is toward the latter part. Most of the time, when people have a website up, they have a website for some reason, especially a paid website like you would have on Squarespace. Usually, it’s to facilitate some type of transaction. You want someone to contact you. You want to book a reservation. You want to book a hotel room. You want someone to book an appointment. You want to sell a product. You want to sell a service. You want to sell a digital download, a good. So a lot of our development efforts remain on this, I would say, enablement for entrepreneurs.
Some of those entrepreneurs may not have a website with Squarespace, and that’s just fine. We have a lot of tools for entrepreneurs that… it works better with Squarespace as a website, but you might have your website hosted elsewhere. That’s okay, too.
That’s a split for me that is particularly interesting, that the growth and the activity is happening. You’re running your business, and people are going to sign up, or they’re going to book calendar slots, or they’re going to buy something from you. You’re launching a payments business in the fall. All that is away from you’re going to start a website. There’s a break there that I think is just utterly fascinating. If I wanted to start a business tomorrow and get customers tomorrow, I’m not sure that starting a website is the way to go. I might start with making a bunch of TikToks about my pressure washing business. I needed a guy to come and cut down a tree, and I went and looked on Facebook before I went and did a Google search, and I found the guy on Facebook in four seconds in my area.
That seems like the big split, that the marketing function for new businesses is happening on social platforms, and it’s not happening at the point of, “we should start a website.” Do you see that split, or is it “we just want businesses that are a little bit more mature,” and there comes a point when you will always need a website?
I like the beginning with the pressure washing business. That was not something I’ve heard anyone lead with before on the small business spectrum.
Small business TikTok is my absolute favorite side of TikTok.
“…when you’re within a social network, you’re beholden to them.”
It fits perfectly with Squarespace, but no, to answer your question, going back to that 20-year history, we are very used to social networks being around. They’ve certainly been around in parallel from every iteration of them, from Myspace to Friendster to Tumblr to Facebook to Instagram to TikTok. Sometimes they come and go. Sometimes they have more staying power. We actually see more demand than ever for websites right now and the importance of owning that URL because, as you know, when you’re within a social network, you’re beholden to them. You’re beholden to them in terms of reach. When you’re posting on these social networks, it’s not guaranteed that all of your followers you reach when you post. Again, they come and go.
So if you’re really locked into an audience there, if you’re serious about what you’re doing at all, that becomes dangerous. That being said, they’re great for distribution. We encourage all of our customers to be on whichever social networks are relevant to them, including incredibly niched ones depending on where people start power washer businesses and how they all interact and collaborate.
By the way, power washing is a business that you should have. I think-
Yeah. It just feels like that is such a creation of TikTok.
No, but that’s so wild to me. Here’s a new social platform that showed up. I very much doubt that ByteDance engineers in China built a platform with the intention of a bunch of 20-year-olds in America starting pressure washing businesses. But that’s the content that started to go viral. Now, we’re at the point of the cycle where it seems like the money in pressure washing is not actually pressure washing but selling masterclasses about pressure washing.
That cycle is nuts to me, but it’s a function of a distribution platform.
What’s really interesting is you see a different kind of content resonate across these different social networks. It’s defined by the medium. A certain content finds its way to Twitter, to Facebook, to Instagram, to TikTok, to any number of ones that have gone away in the past. I’d say two things just to also build on what you’re saying. One of the actually big initiatives we have that we’ll be launching in a couple months is our classes and courses business, so I completely agree with you that there’s a great amount of money to be made in selling classes and courses.
Then the other thing I would say is, toward our portfolio of brands, Squarespace bought a company called Unfold about three, maybe four years ago now. Unfold was an app for creators on social media to basically do formatting around Instagram Stories. The thesis there was that your homepage may not start as a webpage, but it may be your Instagram feed is the beginning of where you want to start, and we want to be around you and help you with the tools you need, whether it’s a link in bio with our Bio Sites product, a full-fledged website, which might be too much for certain people or getting into the flow with commerce. So that’s something we’ve definitely contemplated and certainly have been watching over the past two decades as we’ve coexisted with social networks.
Would you describe Squarespace today or in the future with those kinds of products? It’s still primarily a website company?
I think the brand Squarespace, we’ve spent a considerable amount of money associating with the word websites and online presence and domains and all the things to go along with it. As you get further away from the core of what Squarespace does, the other brands can resonate in a way that is just easier to explain to people. I don’t need to explain to people that Squarespace actually does everything, and it’s for every entrepreneur. It just gets overwhelming for people, and we’ll probably be launching more brands in the future.
So that leads into the Decoder questions here. That’s a lot of brands to manage. You’ve been at it for 20 years. How is Squarespace structured now, and how have you changed it over time?
As you might imagine, it’s in transition. It’s always in transition in some ways, but really, this move from just the brand Squarespace to these other brands within a portfolio — and it’s not that many of them, and they’re hung together in a number of ways. They’re all in service of entrepreneurs, and they’re shared services like our payments platform, which you mentioned that they’ll all use together. We just started buying these brands and launching them probably only four years ago. So, for the most part of our existence, Squarespace was structured very, very functionally.
My background is product and engineering and design. While we’ve had people running those functions here for quite some time, that’s where I was oriented and, of course, mostly toward the Squarespace product. So we grew up very functionally. So around me would be an engineering head, a product head, a marketing head, a creative head, a customer operations and service head, and all that sort of thing.
Now, with the acquired companies and with the brands we’re launching, we’re experimenting more with what would be considered a general manager model, for less of a better way of putting it, just to make sure that these independent work streams and products can do what is best for them without having to always roll up through one centralized point, which Squarespace is a multi-hundred-million-dollar, almost billion-dollar now, revenue run rate company that’s public. Do the leaders of that company have time to focus on five different other brands? I would say they don’t. So you move to this GM structure to give those brands more autonomy so that they can pursue what’s best for their customers and not roll up to just what would otherwise be a corporate bottleneck.
So you are going into some divisional structure now, right?
We’re partially there now.
Are you splitting up so that you have, I don’t know, multiple designers in multiple places or multiple product leads in multiple places, or are you still centralizing all that?
Depending on what’s appropriate for the brand and who the leader is, sometimes we’ll be centralized, sometimes we’ll be dotted line. There’s no hard-and-fast rule. It’s just whatever’s working best. But there are certain things that I think are obvious to be centralized — HR, legal, finance — and then there are certain things you want to have centralized, like payments. Then there’s certain things that Squarespace is special at, and it should have centralized, and those brands can use those services, and that’s our internal creative agency. So when Acuity goes out to do a rebrand, they don’t need to go externally to do that. The people who work on the Squarespace brand are more than happy to help those leaders make something that looks fantastic. That’s one of our core strengths.
One of these days, I’m going to have a CEO tell me that they’ve decentralized HR, legal, and finance, and I think that might be the end of Decoder.
No one does it. It’s the one thing that everyone definitely centralized, but the difference is where do you put design? Where do you put product? Where do you put marketing? And everyone seems to have very different opinions about this stuff.
Well, there are examples of decentralized, all those things, and you just are called a holding company. So actually, holding companies have brands where they don’t attempt to mix those at all. We do. Maybe there’s a size where that’s not appropriate. I’m not exactly informed of how Berkshire Hathaway works, but I think they wholly own those companies, and I think they got 50 people in their corporate office.
Do you think that you would get so big that Squarespace has a website company and a scheduling company and your design services company?
The first couple of those, sure. It already does. I’m not sure we would ever get into using our agency externally. We would try to help it with the portfolio brands than going externally with it.
How many people are in Squarespace right now?
We are a little over 1,700, I believe, a little shy of 1,800.
How are those people organized? What’s the biggest part of it, and what’s the smallest part?
The biggest part by headcount would be customer operations, but we’re pretty lean across the entire company. If you compare a company of our size, 1,700 people, to — call it 1,750 — to our revenue level, which is right under a billion for this year, it’s a pretty lean company. So we’ve always had lean design teams. There’s a very large engineering team, a medium-sized product team, a pretty tight marketing team, and then smaller legal and finance and support functions.
When I look at the chart of other big website companies, Automattic / WordPress, I guess Automattic is a holding company.
Even probably more than us.
I’m looking at the market share charts of different CMSs. WordPress obviously dominates the internet. 64 percent of websites are on WordPress. Then there’s Shopify, Wix. Squarespace around 3 percent. When you think about growth, is it pure market share, “we want more websites on Squarespace, we want to take share away from WordPress,” or is it “we want to make more money from our existing customers”?
It’s a variant on your latter idea around money. You can look at all of the URLs out there in the world and think, “Well, okay, which ones are even appropriate for us to host?” So some are apps. We’re not hosting apps. Some are large companies. Some are large content-based sites. Really, they’re just all across the board and what those URLs are out there. I think that there’s a certain subset of those URLs that we are really good at managing. The ones focused around small business, the ones that are more creatively oriented, the portfolios, then websites — that stuff is really in the sweet spot for Squarespace.
Also, it’s not a free product. We’re never really going for just total count of URLs because we want a more serious user. I think Squarespace is in no way expensive for what you’re getting from it. We’re talking under $20 a month for just so much functionality that’s been developed over those two decades and more every day. So it’s not a URL count thing that I’m going for. It’s which URLs and which are the more valuable URLs for us. So that gets us into: how are these URLs transacting, do we have permission to help them with the transaction, is the transaction even happening online, and how much of that transaction can flow through us?
You mentioned the payments platform we’re launching later in the year. That’s a big thing for us. A lot of people for smaller URLs, they buy the URL, and bandwidth and storage were commoditized long ago. You’re not really paying attention to that stuff anymore. So how do we grow with our customers? If it’s not functionality and features or customers they’re managing, it’s probably transaction volume.
So by transaction volume, you mean you’ve got, I don’t know, all the dentists in New York, and you just want them to do more dentistry? You wanted to help them market to more customers?
Well, that’s an interesting example because do the dollars flowing through when you actually go to the dentist — would that actually flow through us? It probably wouldn’t, versus if you’re on Tock, you’re booking a prepaid reservation, those dollars do flow through us, or if you’re selling a service online and you check out online, those dollars do go through us. So it’s really a really interesting question around how many dollars are floating around Squarespace. Unbelievable, billions, tens of billions, but how many do we have permission to touch and make that transaction easier for the entrepreneur? It’s a smaller number. But as we think about the product roadmap, we’re always thinking about how do we get more in there.
This is a fascinating way of thinking about Squarespace as a business I had not considered before. You’ve got categories that you’ve put URLs into. The best part of this conversation is I keep coming up with hypotheticals, and you’re already in it. So dentists are a bad hypothetical, but restaurants are a pretty good hypothetical in this case because you might be able to take some percentage of their transaction or build a tool and say, “We’re going to take a percentage of the transaction, but we’re going to get you more transactions total.” Have you segmented the customer base like this and said, “Okay. Here are all the URLs in these segments. We’re going to go try to conquest them one by one”?
“Squarespace has always been built as a general-purpose tool. I didn’t care what your website is.”
Look, Squarespace has always been built as a general-purpose tool. I didn’t care what your website is. It’s like if it’s fitting into these patterns, we want to host it, whether it’s a dentist website, an event website, or whatnot. Even though the dentist website is not transacting, you’re not paying for that thing through Squarespace. It still doesn’t mean they can’t be a good website customer for us, an email marketing customer for us, and all that sort of thing. It’s just that our upside will probably be a little bit more capped than if we were truly running back office things there. Dentist is not a—
I don’t think you want to do dental insurance billing.
We’re not currently going after that one, but in a way, what’s interesting is it is an appointment-based business. So some of the appointment booking side of it could go through Acuity. So it depends on what part of it we’re going after.
It just seems like more of your growth is inside the walls of the business. It’s not that, “We’re going to go out marketing.” I think of Squarespace as “I’m going to put up a beautiful portfolio for my work, and then you’re going to come to me for a consultation, and I’ll book you, and something else will happen, and I’ll run my business out of QuickBooks.” Then there’s a part of this that you’re saying, which is you show up in the office, or you show up in the restaurant or whatever, and the point of sale is Squarespace or—
That’s not where we are particularly right now. I think Tock is the example where we are much deeper into the operations within the walls of the actual business just due to how Tock is created, but that’s unique because you’re booking the reservation online, you’re prepaying online. So that makes a lot of sense there. So most of our transactions and transaction volume and the way we’re thinking about expanding is an online transaction first.
One way you’re definitely expanding is in domains. You just acquired Google’s Domains business. Walk me through that transaction. It seems like Google launches things, they get tired of it, and they got to flip it, and you were there to catch it. How did that come about?
First off — once in a lifetime opportunity for us. Incredibly grateful that we were selected as the stewards of that business. We weren’t asking them, like, “Hey, planning on shutting down domains or anything?” It wasn’t exactly outbound. I think they made the decision that it’s not a business that they were going to be in. And they contacted a couple of legitimate parties who could potentially even take on a business of that size because, again, it’s not the code or the employees are moving — it’s basically the domains themselves and the hosting services and the registrations, that sort of thing. So that really narrows it down to the number of companies that could even support that.
Then the other thing that was a big factor is we’ve been a huge fan and big reseller of Google Workspace for nearly a decade now, which was very important to them, and we’re incredibly sophisticated in selling Domains, selling Google Workspace, servicing it, and managing that for millions of people. So we were able to find a transaction that worked for us.
For me, it’s really just the beginning. We’re going to be investing a lot more in our Domains product, especially the Domains product for customers that might not use us as a website. That was a theme the whole way through this conversation. We want to just be the best place for you to have your domains, whether or not the website is with us or not, but it gives us the justification, the opportunity to really relook at that product and relook that experience, make it world-class.
Then also, we’re focused on making sure the transition period when we start that is seamless. We’re using a lot of Google’s infrastructure that they’re currently using in Cloud DNS. So if you’re just staying with the product, a lot of the backend will be the same, which is really important because moving registrars is a huge risk there.
Then the other thing in my mind is, and this is funny: I’m a Google Domains customer. I use Google Domains, and I’ve had a number of domains there for over a decade. Why is that? Because Squarespace started very website first and then added domain second. It’s very valid to get multiple domains on Squarespace now, but just due to inertia and Google Domains being a good product, I had left a couple of domains there. So I am extremely interested in making sure that a really good experience exists on the other side for all of our customers, myself, and our employees who use this product. We’re familiar with it, and I just see it as a great opportunity.
I’ve bought so many joke domains over the years that I’m confident that I have some Google Domains. I’ll let you know to make sure you’re transferring most of them.
You let me know, but after me and some of the people here offer us guinea pigs to transfer, but no, we have incredible resources dedicated to this. I’m confident it’ll be a success. For us, we’ve been on the internet for all of our lives. You just pile up domains for some reason.
Yeah, it’s just a fun thing to buy. They’re like the original NFT.
Actually, more utility than NFT.
You said you’re just buying the domains. You’re not buying the people. You’re not buying the infrastructure. Is part of it, “Okay. We’re going to get these domains. We have a suite of services. We can go market to those customers now too,” or is it, “Hey, maybe some of them will actually move to our web solutions as well”?
Look, we’d love it if they use Squarespace as a website, but again, I think that Squarespace domains should be a completely legitimate option. Whether or not you would like to use Squarespace or not, will we try and show you things about our services? Sure, and if you unsubscribe from that, we’ll leave you alone. Again, I was a Google Domains customer, so I am in that seat of understanding what that experience should be like, but we’re using a lot of the same infrastructure Google is using in their Cloud DNS product. So I think it’s going to be a good outcome.
This leads into the other classic Decoder question about decisions. This was a big decision to make. What is your decision-making framework? How do you go about making decisions, and how did you apply it to this acquisition?
“We don’t just sit there and wait for all of our customers to ask us for something to do it.”
Well, this one was complex because it’s very confidential — as it’s going on, very uncertain at various phases of it. This one, for me, after the inbound and talking over with some corp dev and engineering a little bit, was almost purely a business decision. It was interesting because we’ve been in the domains business for almost a decade. So it’s not like this huge build. We’ve resold [Google] Workspace for almost a decade. So it’s not this huge build where it’s like, “Oh, all these new things we’re going to have to do.” There are new parts of this deal that we will have to build, too, and we’ve already got that staffed up.
I think to answer your question more broadly, depending on what the thing is, it generally starts with a much smaller group of people, and then I widen the concentric circles to either stress test the idea or get more people aligned with what we’re doing. Google Domains was no exception to this — had to start with a small group of people because it was so confidential. Then we did that, widening concentric circles. I get more buy-in. I pressure test financial models with finance, with the board, and try and just gain some conviction that this is something that’s smart.
The other acquisitions — same way. Some of the product releases and product initiatives — same way. It’s interesting because a lot of what we do actually starts from insights and feelings and orientation we have for doing something for so long. We don’t just sit there and wait for all of our customers to ask us for something to do it. So it’s an interesting balance between what we feel that the market needs just being in it for so long and external factors either popping up as an opportunistic thing like Google Domains or just something staring at us in the face as just being a giant market that we really should have been in.
Did you send in emails that were like, “We got to keep this away from GoDaddy”?
We are very happy to welcome a large number of customers onto our domains product.
That’s good. I’ve got a whole sequence of questions about AI, and that was a perfect AI—
Sanded the edges right off that answer. You mentioned your board. You have a rare experience here. You’re the founder. You’ve been at it for 20 years. You obviously started before you were a public company. Now, you’re a public company. You’ve been on the public markets for a little bit. How has that changed your decision-making process?
We’ve been public for just a little over two years now, which, as I’m sure you’ve seen and other guests would’ve mentioned, is probably not the most fun time to be a public tech company, no matter if you’re high flying or profitable or anything else. We at least have the luxury of being … We were running cashflow breakeven for 15 years and had been profitable for the last five. So we weren’t in this money-losing phase or anything even close to that while being public, which helped put a floor on things.
How has being public changed who we are? Aside from just the unfun nature of dealing with the volatility and dealing with all these new actors that are in the public market, I actually think that it’s actually been somewhat of, frankly, a good thing for Squarespace. When you’re private, your employees are waiting for tender transactions to happen. Those generally happen at a discount to your 409A, which is based on public comps over the past two years, depending which comps you pick for us. We’re trading at a premium to those public comps. So you could be unhappy with the share price, but I can almost guarantee you, privately, it would’ve been lower. So that’s been good.
I think after getting into the cadence with the quarterly earnings — I think it brings a discipline to the company that I wouldn’t say we didn’t have before because we certainly prepped for two or three years before going public, including having mock earnings calls and everything else. This wasn’t a giant surprise, but I actually think it’s been a really good thing. The employees can get liquidity. Investors can get liquidity. You have this lovely dynamic where there’s analysts looking at Squarespace all the time asking sometimes good, sometimes medium questions about how the business is going, but in a way, that’s a level of transparency that you don’t have in the private market.
It really forces you to think about, “What are we really doing here? If we’re here for another year, two, three, four years, do we have a viable growing business, or do we not?” I think it puts it in your face all the time. Luckily, because our business is mostly subscription and has been built over the course of 20 years, a lot of our revenue is very, very predictable because we have all these existing cohorts coming over. So it really is about what can we do for growth.
So I think the public markets generally greatly dislike unpredictability. We’re more on the predictable side. We’re not a money-losing business. We’ve been operating this way for quite some time. There’s a million opportunities in front of us in terms of these services for entrepreneurs, the payments business, the other acquisitions, things like Google Domains. I think it’s exciting.
It’s just a different world. I think it’s maybe hard. Look, the past two years have not been fun to be any public tech company outside of maybe three or something like that, but even there, it has sucked. So that’s just been different because I think Squarespace generally plays are used to up into the right, maybe not as fast as they might have liked, but up into the right. So it’s traumatizing to see the value change like that so rapidly, but we’re here to stay and [have] just so many great opportunities coming up. So it’s exciting. You get immediate feedback on that stuff in the public market. It’s just a different equation.
How has it changed your decision-making now? It’s been about two years. Have you perceived, “Okay, I’m making decisions more slowly or more guarded”? Has there been effect that you can call out?
“There aren’t that many short-term things I can do to meet the quarter. There’s not any tricks”
People always seem to want to get into this, “Oh, well, they’re going to do all these short-term things to meet the quarter,” or something like that. There aren’t that many short-term things I can do to meet the quarter. We’re not like a Salesforce basis. There’s not any tricks. So if anything, I think it’s accelerated decision-making about things that aren’t working so that we’re optimizing more for the long term. I think depending on if we needed to do something super risky that would just change the whole model or something, I think maybe I’d have a different feeling, but a lot of what we’re doing is additive.
I think it’s actually accelerated decision-making because it’s like, “Hey, you’re going to make this decision now, or this is what it’s going to look like this quarter, next quarter, next quarter. Do it. Move forward.” It gives you, in a weird way, I’ll say error cover is the wrong word, but you can point to some numbers and say, “This thing’s not contributing to this in year two or three. Do you ever see it?” or, “Hey, do you really want this expense right now or do you really want another point of free cash flow, another two points of free cash flow to get us working in that direction?” which then just opens up even more opportunities for us to be able to fund a transact … There’s only so many companies that can even fund a transaction like Google Domains either out of cash or debt. So that’s really important to us for when these things come around. Imagine another world where we were private, burning cash. Maybe we couldn’t even finance it. So I think it’s been positive.
I have a sense of Squarespace. I have a sense of how you make decisions. I have a sense of where Squarespace’s business would be and where it would go if not for the extremely disruptive shifts happening with AI and distribution on the web. I want to take a second out of this conversation and really poke at that stuff because I’m extremely curious about it. Let’s start with AI. Squarespace, like every other company, you’ve launched some AI tools. You can use AI to auto-generate some text on your website. “Write me a paragraph about pressure washing” — it’ll do it. Great. How does that work? Were you [like], “Okay. We got to go find an LLM partner and pay a license fee”? Is that ChatGPT, or are you building your own? Just that turn of it, how did you integrate that?
First off, just to frame it all for us, I’ll take it from two angles. One is we’re a very tech-focused and forward company. My background is engineering from when I was a kid. So the AI machine learning is absolutely nothing new to us. Obviously, the leaps that the LLMs have provided are really exciting and new, and we’re all excited to either integrate them like we’ve already done in the product for text generation or we are integrating them on onboarding in the form of prompt engineering into an LLM, which would feed back into the visual product of Squarespace or even a little further out for us, just how do we incorporate that into assistance, but we’ve been incorporating machine learning models in Squarespace for a long time.
We’ve had some form of AI-powered support for four or five years now that we’ve been training on our own data sets and getting better with. This will be an evolution on top of that. That’s super exciting. I talked about this extensively in my last earnings call because it was such an overnight interest in all of this. I’m actually not as worried about the impact of the LLMs and Squarespace’s core business for, frankly, a number of reasons.
One is we stopped requiring people to code websites two decades ago. Also, a lot of what we do on Squarespace is not the coding of the website. It’s storage, it’s bandwidth, it’s DDoS protection, it’s CDNs, it’s an SSL certificate, it’s domains, it’s payments, it’s support, it’s design assistance, it’s our email campaigns product, it’s anti-spam — just 20 things that are happening in your subscription for something like $20 a month that there’s a lot of value that we do that’s not just, “Code me a website.”
Even if you wanted to code a website, I would say that while I think the AI right now can get you to a great starting point, I think that the use of a visual tool is super useful even after that starting point is output to you because you might want to just grab a thing and move it an inch to the left and there’s sometimes no better way to do that than grab the thing and move it an inch to the left.
So I’m excited about the future of the core business because of, frankly, the great reception we’ve seen in the past couple quarters on the core product, but then I’m excited to integrate these new technologies and augment the ones we already have and, hopefully, I’m pretty confident it’ll create a tailwind for us.
That’s a pretty interesting compare and contrast, given your history. I remember when the first WYSIWYG web design tool showed up, and they basically output bad code. It was just bad, HTML was sloppy all the way around, and the old-school web community was like, “This is garbage,” but eventually, the WYSIWYG editors won, the visual web design systems all won, Squarespace won, and yes, some people still hand code their websites and I love them. They’re my people, but—
Yeah, no, it’s great, actually.
… but the mass market all moved on to the easy-to-use tools. Are you saying this is the same with AI, that a bunch of people are freaking out, journalists, writers are freaking out, but at the end of the day, we’re still going to be in balance?
“…just because everything could be eventually possible, it’s not all possible today and even next week or even next month.”
It’s a funny thing to respond to because I’m going to preface it by saying I’m blown away by the developments in AI. I think that the LLMs and the experience of that are amazing. I think with the prompt engineering, and that, on top of tools we have, is ultra exciting. Do I think people have gone into this like “all the jobs are gone tomorrow, next week” thing a little too fast? It sure seems like it. This is something that’s going to be disruptive to many, many industries and something we’re incorporating, but I think this is a phenomenon where just because everything could be eventually possible, it’s not all possible today and even next week or even next month.
A lot of those things I listed out that Squarespace does, nobody is sitting there going, “Bandwidth will now be completely different because of the large language [models],” or at least not right now. You could paint yourself a way of getting there because all the coders are 10x productive, and then you can get there somehow, but it doesn’t currently seem like outside a number of very specific use cases. Wall Street has modeled in that all of the company’s workforces are going to go down by 50 percent, and thus, all the profit margins are going up by whatever equivalent is, or this business is completely gone because it’s replaced by I don’t know what.
There’s a lot of words, but now a lot of model updating for some of this.
I got you, but there’s one specific place where I can say AI is going to radically change this thing, and that is the web for two reasons. One, flooding the web with text is pretty easy. If you have a Squarespace account, it’s not built into the tool. I can set up a new website and have some LLM, you still haven’t told me which one, but I can have some LLM-
…fill a website with text. That has implications just for the web at whole. Then on the other side of it, there’s distribution. Facebook is not sending a ton of traffic to websites. It’s all Google, and Google’s incentives have really shaped the web for the past decade. Now, we’re at a point where Google is going to start eating some of those search results. Maybe AI is overheated in some places, but on the web, it seems like the issues are fairly clear.
So to be clear, we currently have in production the ability for you to auto-generate text using, in the background, is called OpenAI, and there are LLMs, and we make that accessible to all of our customers right now. Now, if you were trying to, as you put it, flood the web with text, using Squarespace would probably be a pretty bad way of doing that. I think you’d want to script stuff and output it and all that, but they’re being—
No, but I’ll give you the example, just a really dumb example. Every time I pick an example, you tell me all the details of this example, which is my favorite part of this conversation, but I’m going to pick car dealers. Car dealer websites are full of garbage. They are basically SEO honeypots. You search for a feature in a car that you’re interested in, and a car dealer has a webpage that may or may not be accurate designed to just rank and search. That’s what I mean. It’s going to be a lot easier for that set of actors who are doing something that could be described as honest content marketing but what’s actually underlying it is pretty insincere. They’re just trying to get traffic.
So maybe we live on different webs, but hasn’t garbage and content farms on the web been there for an extreme amount of time, maybe not on the scale-
But now you’re handing those people a bazooka.
Correct, but I would wonder what percentage of their articles are actually generating the majority of their revenues. And I wonder how Google is either giving them credibility or not credibility. What I think of more is how the web has been a massive input to these models. I think a lot of disruption can happen to certain businesses where if you’ve ingested the entirety of a reputable set of content, a Wikipedia, a Stack Overflow, that the LLM model can sometimes do a bit better of actually giving you a response on top of that corpus of knowledge. That’s really interesting. I wonder how people are going to feel about the lack of attribution within the LLMs that Google fought with for a while.
Right now, if you type into Google various search terms, many summaries and cards appear that are not websites that are attempting to answer that question for you. Some of them have attribution, some of them are just computations that Google will just do, and that’s cool, and you don’t need to go to the website, or maybe the website is a click later because the transaction is still occurring on the website.
I think it’s really interesting to think about how the web and private data even will flow into these models and for which examples the LLMs will be a better alternative to search and one that’ll be a worse alternative to search. Now, one of the examples that comes to mind is a hypothetical, but a better alternative to search is I’m a coder, or I used to be — now I joke that I’m an HR and comms person, but I used to be a programmer and honestly looking up those coding snippets and getting started, not writing the whole program for me, but getting started with, “How do I do an X in Python if it’s like this in Java?” That’s a magical result it’s giving you. It’s really, really, really interesting. So I think you’ll see reduced traffic to certain kinds of things on the web. Whereas you’ll see increased traffic and usage of the LLMs, but—
Are you going to watermark Squarespace pages that are made with AI? This is a hot topic that you should be able to somehow detect what content has been made with AI or somehow mark content that’s authentically made by humans. It seems like for a provider of webpages in the most abstract sense, Squarespace could say, “Okay. If you use AI tools, we’re going to tell Google the content on this page is made by AI,” or, “We’re going to tell Google, actually, a human made this.”
Is there an effective way of telling if a content block is generated by AI? Because obviously, we know if you click the button on Squarespace, if you went to some other model and pasted it in, I don’t know if you’ve typed it into a text editor or not.
I’m wondering if you had this conversation because I talked to Microsoft or Google, and they’re constantly talking about cryptographic solutions to at least imagery and video.
Imagery and video would be different.
Then even to some extent, they talk about text. You can, to some degree of confidence, detect when an AI has generated a piece of text.
Not to make a joke about it, but what if the AI-generated stuff is better than some of the human-generated stuff?
I’m not saying that never happens.
Caution: this one’s generated by a human.
I asked this because this seems like where you would impose a regulation. It’s on a vendor like Squarespace that’s making the webpages. The reason you would want to impose something like that is, like you said, right now, these LLMs are being trained on data that the majority of which is generated by human beings, the internet up until now, basically, and we’re about to hit a point where Squarespace is going to publish a bunch of content generated by AI. WordPress or Wix or whoever, they’re all going to do it. Then the models are going to start training on that, and then you end up with a number of bad outcomes, one of which is model collapse, where the models start failing.
I have two responses to that. If you’re looking to generate a large number of webpages — call it 10,000, 100,000 — making 100,000 Squarespace trials and injecting that in is probably a really bad way to go about that. So that being said, from an AI perspective, though, what I’ve started to contemplate — and it’s more interesting — is for a long time, the internet has had robots.txt, which tells crawlers what they’re allowed to do with the content on your site. We’ve also had creative comments, licenses, and other things you should put on your site so that humans know if this is free, if this requires attribution, all that sort of thing.
So where I think is a bit of the Wild West is, have we equipped people or even equipped the LLM creators to understand what is allowed to be used, who is restricted, what requires attribution, because that’s an interesting one. If I’m asking an LLM a question, I would love to know if it could tell me whereabout some of the sentences were sourced from, like, “Is this 80 percent Wikipedia-type stuff? Is this 80 percent Mayo Clinic?” or whatever — pick your company that has a large number of URLs. So I was thinking more about it like that from a user perspective and less about it from all of a sudden we’re going to be the host to 100,000 AI-generated articles that … I’m sure somebody’s already going about doing that.
Just to be clear, I don’t think it’s a single bad actor. Although if somebody tries to start 100,000 Squarespace trials and do AI, I respect the hustle.
They would hit a big anti-bot filter.
I don’t think that’s the bad outcome. I think the bad outcome is that all of your customers start using the tools, and then, on some timeline that doesn’t seem that far out, you will be serving 100,000 AI-generated things.
I see what you mean. Basically, you’re worried that there’ll be no creative writer or imagery.
Yeah, because it’ll be cheaper and easier to say … Again, I’m a car dealer, and I know … I just installed a booster seat for my kid. This is why I had this example in my head.
It’s power washing and cars. That’s really where—
It’s a very car-oriented entrepreneur conversation.
It’s always in my head, in the back of my mind. Actually, our first set of guests was all car CEOs. It was very obvious what was happening. So our kid got a little bit older. We got her out of our car seat. We put her booster seat in the car. I was just Googling, I need to be able to install this thing right, like any parent would do, and 10 of the first results were just car dealerships. I have a neutral opinion on that. They’re doing content marketing. That’s fine.
There comes a point where the car dealer is going to say, “Look, I am tired of paying for anyone to write this copy. Just have the intern write me five paragraphs for installing a booster seat and put that on the Squarespace page,” and that will be easier and cheaper at scale for more businesses to do for more things. Eventually, that stuff will get indexed into Google, and that will be a recursive loop that leads to bad outcomes.
At some point, someone’s going to say, “We should stop it.” Google could say, “We could stop it in a pretty dramatic way.” Or they could come to you and say, “Hey, start letting us know when this is happening so we can downrank it.” Or the government could tell you to stop it, but at some point, that cycle gets to a place where there’s more garbage in the ecosystem than not.
What you were saying toward the end of that — and I’ll respond to the beginning of it — when you say someone should flag that this is AI garbage and we don’t want to rank it, Google has as much authority or more as a third-party observer to make that determination than we do because then you have to trust us. I actually don’t trust, because we haven’t invested billions into it, our ability to tell them because you can just paste something in if it’s completely AI-generated or not.
I’d say one other thing that, as technology evolves — take Squarespace from 15 years ago, “Squarespace is replacing web developers. There will never be more jobs for web developers.” Lo and behold, there are still jobs for people who help people with creativity and content on the web. There are more of them. They’ve just changed. So if you are capable of coding really generic websites, yes, Squarespace totally did displace the need to do that a long time ago.
So when you talk about copy, or you talk about image generation, first off, there’s a lot of things in that realm that are totally unique and a unique story. You might start with somebody helping you with the paragraph, but you need to write more. Secondarily to your car dealership example, how do you know which one’s good? Well, probably you have some human filter for, like, “No, that’s actually a picture of the real car dealership, I think.” They could lie completely and fool you, but at some point, that will end when you show up at the car dealership, and it’s not the thing it said it was.
“I do not think right now, in their current form, [AI tools] are a replacement for human creativity and storytelling.”
So I think these tools will displace a certain amount of bad writing or something like that, but I do not think right now, in their current form, they’re a replacement for human creativity and storytelling and its deepest of forms. I think they could be an assist on that, but maybe that’s just a romantic me holding out for creativity in the world.
It served you well for the past 20 years. Does most of the traffic to Squarespace sites come from Google?
Actually, I wish I had a better answer for you on that because it would probably depend on the segment. For some segments, it might be Google and Google rankings. As you know, for certain keywords, there are very few sites that rank for those. Obviously, Squarespace sites is too great at Google. We’ve been around for two decades. We know about SEO, but depending on the personality, a lot of your traffic might come from your Instagram page. It might come from where you have a following. So I don’t think there’s any one answer to that based on just the entirety of Squarespace.
When you say you’re good at SEO, this is actually something I’m really curious about. You do a lot of design services. You have a lot of templates. Do you feel the tension between, “Okay. Here’s where we think the web should go,” or, “Here are some experiences we’d like to build, and here’s what Google needs in order to rank”?
Look, I hope that for most people who are not programmers, Squarespace will continue to exist as a place that pushes forward what they’re able to do creatively by themselves and will always have a place on the web for completely custom coded one-off content that is beautiful and creative and amazing. It may be some time before a CMS replaces those sorts of things, but look, both can coexist.
Well, I’m just curious because you can have a website. Your website’s not worth a lot without traffic. So a lot of my silly car dealer examples or whatever, they’re just trying to get traffic. They’re looking at what people are searching for, and they’re firing out content to just try to get one click onto their website in the search result. Google is the last big funnel of traffic from what I can see. Maybe some people have links on their Instagram page or links on their … The pressure washer guys all have links on their TikTok pages, but the last big source of traffic is Google. It seems like the influence is getting correspondingly bigger as well.
Buzzfeed, for example, was a Facebook product. They were not organized around SEO. Now, they’re getting more organized around SEO because Facebook traffic has fallen off. That’s just a big example I can give you. Do you see that pressure inside your own business? “Okay. We help people make websites. In order to market those websites or get traffic, we have to increasingly push them towards what Google wants.”
I would question whether or not if you are the new power washing company just starting out.
This episode has done more for power washing than any other podcast. We should just do an entire episode about—
I do not have a power washer.
I’ll send you some TikToks.
Imagine the zen of using it.
You were talking about it with relation to Google. So why are people putting this content on TikTok? Why are they putting it on Instagram? Why are they putting it on Twitter? Because to rank on the first page of that on Google is maybe not where you should start. You should start with something that’s more niched, a community around you. For blogs a million years ago, you would participate in the comment section and leave your link and get authority that way. There’s different ways to get authority on Twitter, Instagram, TikTok — name your social network. I think when people flock to those greenfield opportunities, it is specifically because ranking on a very common term on Google is not where anyone is starting. That’s impossible. That’s more the result of success versus the — for generic term, of course — versus the way you become initially successful.
Have you found Squarespace’s ideas about the web getting more or less influenced by Google over time?
I think less because of what we were just talking about. For instance, if your homepage in your mind is your Instagram profile, how much does your actual top-level URL matter as much as if you were trying to sell a product, the detailed URL that you linked to from your Instagram page? That is something that has nothing to do with Google that we need to really, and we do think about where is the traffic coming from and how are they gaining popularity and how do our URLs and whatnot present themselves in those environments. At the end of the day, most of the transactions that are occurring, maybe almost all of them, are not actually happening on the social network themselves. They’re not happening on Twitter. They’re not happening actually within Instagram chat.
There are some examples where that might be the case, but a lot of the complex things need to occur. It’s still happening at a URL somewhere at some point because there’s a lot of backend logistics, and a lot of things need to happen. A lot of delivery needs to happen, and it has to hit an end point somewhere.
You’re saying all that’s better on the web so people just convert over to the web, and you’re going to be there for them as that provider.
I’m saying it’s only on the web, unless you’re in a walled garden. Unless you’re selling through Amazon, for instance, a physical product, but as sites like the success of Shopify has shown us, there’s a massive demand for people to go direct to consumer and disintermediate those experiences. Otherwise, we wouldn’t even have a Shopify. They’re a great company. They do a great job. We, of course, have ways to sell physical products. We have many other things we’re selling on Squarespace that are not a physical product – service, and appointment, et cetera.
Are you thinking about the next generation of social media services, the decentralized products like Bluesky and Mastodon, whatever Reddit clones – Lemmy, Kbin. You’re talking about your new homepage is going to be Instagram. “We went out and bought a company and made a product to make your homepage better at Instagram.” Are you thinking, “Okay. We got to get ahead of it on Mastodon,” or whatever?
I’m not sure we approach those in any way that’s substantially different than how we’ve approached them appearing in the past because, again, there’s usually this link out somewhere. If there’s not this link out somewhere, people can’t really transact on the platform, and so their businesses are just going to be so limited there. I think it’s going to be very interesting to see whether or not content moderation sits on the server or on the client and what’s more appropriate for that. What I think is interesting about something like a Mastodon from what I know about it or BlueSky from what I know about it, I could be getting this part wrong, is by decentralizing the servers, you create an environment almost like old school IRC, if you remember, which is something I grew up on and programmed.
We used to run the whole Verge on IRC.
Oh, that’s amazing. Pre-Slack, right?
Yeah, it’s a precursor to Slack. So I learned to program from people on that when I was 14, 15. But remember, there were different networks, and it was all the same protocol, but there were different networks. So if you didn’t agree with one, you could switch to the other. They could interoperate, they could merge, they could split. So that was interesting. It’s interesting to see a bit of a return to that. So do I think everyone’s going to run their own servers? No. Do I think, in some context, something more decentralized but sharing a protocol could work? Maybe. It used to work for email until spam would’ve ended that one, right?
Yeah, for sure. You’ve given me a ton of time here. I feel like I could go for another hour on just what the future holds. It is refreshing to talk to someone as optimistic as you about this stuff. Even the AI people who should be the most optimistic based on their evaluations have a twinge of like, “Oh, so it could kill us all.”
Well, yeah, we didn’t get into all those hypotheticals, but I was mostly talking about it in the context of the business and not the context of a dystopian five-year view.
Look, the car dealers are going to be armed with AI, and they’re going to pressure washer all of the—
There’ll always be power washers, though. They’re not coming for that.
I’m going to send you a list of some things to check out. It’s going to be great. They’re mostly TikToks of power washing guys. What’s next for Squarespace? What are we looking out for?
So many exciting things. Toward the end of the year, the new product launches we’ve got for service-based sellers, classes and courses, all the improvements we’re making around Google Domains, our payments products, hopefully some new brands soon, enhancements to the existing brands, and just a really powerful portfolio of products for entrepreneurs. It remains incredibly rewarding to work on that, and there’s just a lot left to do.
Amazing. Well, this was so much fun. We’ll have to have you back soon. Thanks for coming on Decoder.
Thank you so much for having me. I really appreciate it.
Decoder with Nilay Patel /
A podcast about big ideas and other problems