John O'Nolan, Founder of Ghost - the Open Source blogging tool making $7.2m ARR

Okay. You might have a point. And, and that's nice.

And the probably the base inflection point that fuels most of our growth for the past few years is when we added, email subscriptions. Sorry. Email newsletters and memberships along with paid subscriptions. As soon as we did that, which was, we built it before Substack existed, but then Substack came along and made that very popular. And as soon as those two things sort of happened in unison, the the creator economy suddenly popped out of nowhere as a term.

Right? And, that drove an enormous amount of growth through COVID. Turns out a lot of people who were in lockdown were starting blocks in newsletters and membership sites. So that was our first kind of proper hockey stick. And we're still kind of riding that wave, to some extent. I think the creative economy sort of had its big spiky trend, and now it's calmed down, but it's there's still a lot

They have an easy to use API and they're trusted by big dev tools like Vercel as well as smaller fast growing dev tools like Nock. So if you're looking to cross the enterprise chasm and make yourself enterprise ready, check out WorkOS. We've also done an episode with Michael, the founder of WorkOS, where he shares a lot of tips around crossing the enterprise chasm, landing your first enterprise deals, and making sure that you're ready for them. Thanks, WorkOS, for sponsoring the podcast and back to the show. I know that, one of the reasons that Ghost resonates with developers is, is open source.

The fact that Ghost is open source, and how does that like, do you try to communicate that with with journalists as well?

The open source piece of it means that no one can ever take it away from you, which is what happens again and again on the other platforms they're used to. Right? Medium suddenly redesigns or puts on a paywall that you didn't ask for or, completely pivots to a new direction and the algorithm changes and no one sees your content anymore. That doesn't happen with open source because you you fundamentally control, your destiny. And if you don't like what we do, what ghost does, you just take the code, move it to a new server, and do whatever you want with it.

When people when publishers understand the independence angle of open source and the significance of it to their business and then they're not gonna be rug pulls again like they have been with so many other platforms then it starts to resonate but it's, it's a challenge to get them there. Open source is inherently kind of a technical phrase you know and explaining it is tricky. We're always kind of trying to figure out better ways to tell that story, because the when people get it, they it resonates, you know, they click and they go, oh, this is amazing. Like, why haven't I heard about this before? But up until that point, the it's it's challenging to get them to to fully buy into the idea.

We use independence. We talk about being a nonprofit organization, what that means, which is very related to being open source because that's kind of that's what they're interested in. Then there's a giant developer section, which is full of docs and install guides and GitHub links and everything. It's just one, you know, it's one level deep in the navigation. It's it's still right there, but we have we try and shape the message depending on, who's reading it.

Right? And for the most part, people landing on the home page are not developers, and they're not trying to find the information about developers, developer topics. But it's still readily available. But, yeah, trying to change it depending on on who's reading is is kind of the the main thinking.

I don't understand what's the value of that. Everything else. Right? Because unless you have, like, a lived experience of being locked into a platform or a network or, like, having the algorithm changed and feeling very stuck and having all the sunk cost fallacy about the amount of time you have invested, then you might hear the cure, but you don't know what the disease is that it's curing. You know?

Yeah. So it's all kinds of less. But what's nice, what's useful in 2024 is A lot of people have much more of a sense of those things. You know, 10 years ago, you'd say that kind of stuff and people would be like, medium would never do anything bad because it would kill their business. Whereas now, you know, there's been enough cycles that people are like, oh, I've seen this movie before. I I see where this is going. Okay. You might have a point. And, and that's nice.

Yep. One of the questions I wanted to ask you is, like, something that a lot of dev tools think a lot about and struggle with is, you know, how open source should you be? How are you still gonna make money? I think that your approach is that you are basically very, very open source, and and someone has no obligation. They can go host it themselves, and yet people like me still pay. So, yeah, what what is your take on that and and how how has your approach been?

It's kind of the slowest pace of updates that we would release. Then you can use our hosting. It all happens in the background, all happens automatically. All the kind of third party services you need to do things like email and payments and image editing and other libraries we integrate with are all preconfigured. So there's a convenience factor that is kind of what we're selling alongside the the hosting.

If you can do that model, I think it's a great model. It's worked really well for us. Building, hosting infrastructure is not easy, but it it does have decent margins, and it does have very consistent sustainable revenue. You know, we've never had in any enormous spikes or any enormous dips in our revenue growth. It's just been very stable.

And having that, consistent monthly revenue kind of knowing how much we made this month, knowing how much we made next month makes it very easy to hire people because you kind of you always know what your budget is, which is not the case if you're selling one off, licenses or products. You know, you might have a really great January and a really bad February, and that makes it quite difficult to wrestle with the idea of should I hire someone or should I not? I don't know if I'm gonna have a good month or a bad month. So I think for anyone who is doing open source where there could be a hosted or a cloud version of it, I I think it's a great model. Of course, not everyone can do that if you're making a search library or something.

You know, there there might not be a cloud version of it that is conceivable with that model. So I I totally know that it's, it's not available to everyone. But if there's an angle, I think it's a it's a pretty good model you you can do.

Cool. You're gonna be cheaper than us. No problem. Okay. But now if you want email newsletters, you need to go and sign up for Mailgun. And because we send 100 of millions of of emails, we have really good volume discount pricing with Mailgun. So we pass on, that pricing to customers as much as we can. But you don't get that if you're signing up from scratch. You know? You get the your low volume, therefore, the price is higher, and you have to manage all of that.

And the thing about managing, you know, the website has to stay online, so now someone's on call. That's a salary. The emails have to get delivered and if the deliverability goes down, someone needs to investigate it. That's salary. Yeah.

Now you get hacked so you get a DDoS. Okay. So now you need, like, a CDN and a firewall. So we have Fastly in front of, Ghost Pro and again volume discounts so we can get that for everyone. But if you do that by yourself, it's gonna be very expensive to to get all that set up and pretty complex.

So it's probably, another salary. So I'm certainly not saying everyone needs all of these things, but if you do as a larger organization and you start adding them up, the the costs, just of of running an app of the complexity of ghost. So it's easy to see it as just a simple little blogging thing, but once you start adding in all the authentication of members and the email notifications and the email newsletters, which is bulk, not transactional, and and the payments and the webhooks from Stripe that are flying back and forth and queues for managing scheduling, it quickly turns out to be quite a substantial piece of technical architecture. And one of the challenges I think we're gonna have, for the next few years is we actually don't want to have everyone using Ghost Pro. We don't want to be hosting all of the ghost sites because I don't wanna grow a giant company.

We have this idea of not having about more than 50 people working at Ghost. So that could come a point where we have too many customers. We are not gonna hire more people to be able to deal with that many customers. So we actually have to figure out making self hosting easier and cheaper so that we can offload some of the market demand, to other companies and have them succeed and have them do well with Ghost. So we've got a big incentive to to try and, address this in future, but at the moment, the convenience factor of Pro is so high that's, a lot of people a lot of people who are running kind of the serious sites that are using a lot of the features, they will just use pro because it works out better value.

Now if you're a developer, you just want a blog on the Internet, you don't want any of those extra things, you can self host and you don't mind spending, you know, a few hours a month, you can self host. It will be cheaper. You'll have an absolutely fine time. There's no difference between ghost Pro and the self hosted option. It's the same code based on both, so we don't have, like, special features that are only available to our customers.

It's just a case of how streamlined, how automated, how many external things you have to depend on or not.

And so when setting up Ghost, I really wanted to have a business that could support the open source projects, not be in conflict with it. And so the idea I came up with was a nonprofit organization that could make money from hosting to sustain its operations, but that would be fundamentally independent. So I don't own Ghost. My cofounder, Hannah, doesn't own Ghost. We're just trustees, which means we can, control, steer the ship, decide what Ghost does.

But all of the revenue that Ghost earns is reinvested into the project. It's reinvested in the form of salaries paid to the team, but, and funding our operations. But if, Ghost makes a profit in any given year, that profit stays within the company. It stays dedicated to the mission. It's not going out to investors, shareholders, anyone else like that.

Nonprofits confusing term. People think it means no money, but it actually means no money for the owners because there are no there are owners. Sorry. There are no owners. So it's not for profit.

It's being built for something else, not for profit motives. And there's not many nonprofit tech companies out there. In fact, I'm not sure I can name any others that have quite exactly our model. There's kind of nonprofit foundations that do open source, but they don't necessarily have a a commercial business that funds it. It's more donation based, but it's worked remarkably well for us.

And and one of the best things about it is, we just have this very clean model of, like, hey. If you use our hosting, you're contributing to the project. Thank you. You're funding us, making new features for you. And when we do things or announce things, whether they are positive or negative, it it brings a lot of goodwill.

People trust that if we're making a decision, the motivations are to, do it for the the good of the community, the good of the user base, the good of the projects, not because it's gonna make me on my side for a wealthy, which which often is the case with, traditional startups. Right? They all wanna change the world until Google comes along and offers them a $1,000,000,000 and then suddenly, you know, you're a Google branded product that's gonna be shut down in a year. And then where did the world changing ambition go? Go.

Okay. Yeah. That was the marketing side quest. The real ambition was to sell. And that's fine. There's nothing wrong with that. That's just not what I wanna do. I wanna build something that lasts. I wanna build something that's shared infrastructure, that's good for the Internet, that's good for good for publishing good for journalism that's that's what I want.

It's we're in Singapore, but it's British common law, and those are a little more, strict about what things they do and don't allow, and I think it would be a lot harder a lot harder.

And sometimes that would be Tumblr and then post formats would get added and sometimes that would be Squarespace and I've got Gutenberg and sometimes it would be Shopify and then they'd acquire WooCommerce and things that would support WooCommerce, would get prioritized. Some of those were after my time, for the record, but there were others that were during, my time and it Mhmm. I just didn't I didn't love it. And I think, you know, the recent conflict is it's the same thing. It's just more people are aware of it now or more people are becoming aware it now it's not my favorite I would like I think open source projects needs more transparency and more integrity if they're particularly if they're going to be volunteer based and contributor based and I think what we're seeing is you know capitalism meets capitalism meets community, capitalism meets open source development coming coming to a real head there.

I I really don't know how it's gonna end it so it's been like watching a bad movie the the the last few weeks.

If if someone's raising venture capital and they're building an open source project, Do you have any advice that you would say to them around, like, being able to keep the integrity and, you know, build build something great for the Internet, while also, you know, pleasing investors if it's possible.

You know, what would make that better? What would make that feel better? Not as, like, messy as it is. And there's some I think there's some really easy answers there. I think automatically be in a less tenuous position, if if the relationships were cleaner, if the WordPress Foundation owned everything to do with, the evidence source community.

So wordpress.org, the trademark, you know, all of those assets, including the commercial trademark, by the way. And automatic had a a right to to wordpress.com and could run its service and could do all those sorts of things. And the 2 had different leadership teams. So a independent board of directors on each that would be solely focused, not have a conflict of interest, solely focused on what was best for each of those organizations. I think that would be a huge improvement.

Without that, maybe it would have been a lot harder to raise venture capital, but I do think there would be a lot more longevity in, the goodwill of the community and people's understanding of of the model. It's this kind of tangled mess of who owns what and who are they speaking on behalf of that is difficult to reason about.

If you believe those things deeply, I I think it is easy. I think it's just a decision. You just make it or you don't make it. Like, I don't own anything to do with Ghost. The foundation owns trademarks, it owns the domains, it owns the assets, it owns everything, and there's no part of me that wishes like, oh, I wish I'd hold on to one of these assets so I could use it as leverage because that's not, what it's for.

And, you know, I'm I don't know what's what's in Matt's head, and maybe he has great reasons for why this is the correct way to do it. But I don't know if it's that hard. I think if he wanted to do it, I think he could.

Having that own the the WordPress domain name .org, having, a board of trustees or directors that includes, community elected positions, which is something you do see in other open source projects, Drupal being like a really good example. I don't think there will be any restrictions on its ability to do that. I think the trademark is pretty too messy to socially unpick at the moment. I wouldn't I wouldn't wanna try and reason about that one, but there's definitely some things that that would be a huge step in the right direction Yeah. As far as I can tell.

And then the other is being funded versus not funded. So on the architecture side, a centralized app, you can deploy to production a 100 times a day. You've got one database, and the marginal cost of each user is just a row in the database right it's very low it's almost zero so scaling that updating it distributing it it's all on one domain name Very easy. Relative. Decentralized, you have none of those things.

We just don't press pro. We have 26 and a half 1000 individual databases, which is, it turns out, much harder to scale than, you know, one up until a certain point. Eventually, you have to shot it and do all these things anyway. We have 26 and a half 1000 different domain names we have to support, but more broadly, you know, 100 of thousands if you include everyone that's not hosting. And so then there's things that, you know, like, oh, what's the big deal?

It's just different domain names. Well, OAuth works based on a callback URL to a particular host name. Right? It OAuth, the standard, which has become very popular, assumes, a centralized architecture. And the way the web has of has evolved in terms of APIs and webhooks and OAuth has assumed centralization as a given.

And when you're building a decentralized app that needs to run-in all these domains, all these different servers, all these different instances of the app, what you quickly find is that the building blocks of the web today have not really been optimized for that use case anymore. So almost everything we build is in some way harder because it's distributed. When we want to do an update, we have to ship a GitHub release. People have to go and NPM install that GitHub release. All that has to happen.

We don't have access to logs on anyone else's servers to see if there's bugs. Obviously, we do on our own, and we monitor that closely. But all the engineering seems to be a lot more complex for a decentralized app than a centralized app. And then on the funding side, it's it's really the, the economies of scale and the things that you can afford to do just by throwing silly venture dollars down the toilet. You know, Subset can have a free plan.

They can burn $1,000,000 a month, and the free plan attracts mass attention. Right? It's if you have a a decent product, good concept, and you can get started for free, lots and lots and lots of people are gonna are gonna come in the door. And so you can afford to lose money until you get kind of popular enough that, some of those people will upgrade to a pain plan or you kind of hit this cultural moment where enough you become a brand that everyone knows about and then you flip that into a business somehow. That's pretty much the the Facebook play to some extent.

We can't afford to do that. We don't have 1,000,000 of dollars we can burn through every month. So the we can't have a free plan. We can't pay a bunch of people to use Ghost which is another thing subsector is they just offer these giant checks. So you see us they announce like celebrity so and so is on Subsect now and you think, wow, that person has a good taste.

They're famous if they're using Substack. This discerning good decision making individual must have, carefully analyzed which product to use and determine Subsect the best. And you say, yeah. It's that or it's the $300,000 that they got, to come over and set up a new site. So they pull different marketing levers, which is definitely an advantage they have that we can't do.

Trying to think if there's a good counter example of an advantage we have. I think so one of the advantages we have being decentralized and open source going the other way is customization. So any centralized platform, there's a limit to the 3rd party code you can allow to run. Right? Because it's gonna impact all other customers.

Whereas we can have fully custom themes, integrations, much more deep features and customizations that publishers typically want as they get bigger. Beginners often don't care about those things, but once people start building a serious brand on the Internet, they often, start caring about those things. And centralized platforms, closed platforms, funded platforms, they tend to have a harder time, until you get to kind of Shopify scale where you can do some of those things in a centralized way. That's much more of a challenge for them. So I guess that was a lot.

Those are all the things popped into my head. Yeah.

And the reason they do that is because their business model went away. They used to make 1,000,000,000 of dollars. Now they make what I think less than 10% of what they did a couple of decades ago. And that's because Google and Facebook took all the advertising dollars, the more compelling ad platforms. And so what's happened is the lack of the business model has eroded the fabric of journalism as a whole, as a market, as an industry.

But when journalism works well, it has a real purpose. It's not clickbait and out clickbait and outrage. It's serving communities. It's helping people understand what's going on. Yeah.

You know, with their neighbors or their schools or their community. Like, what's happening that they actually need to know about? Who is who is gonna be elected, not at the national level, but in the direct kind of community you live in. Who's making decisions and how are they making decisions and who should we vote for? And, like, what's going on?

There's having more informed groups of people as a society is generally a good thing. Yeah. But it's not possible without funding it. And at the moment, that's kind of the biggest thing we're trying to fix with Ghost, with the memberships, with the subscriptions is to have a business model that's actually sustainable. Your readers pay you.

In doing so, you must deliver them high quality content, otherwise, they'll cancel. It sort of Yeah. It sets journalism back to the model it used to have with newspapers, which by all accounts seem to work a lot better than what we have today. So so the long winded way of getting back around to your original question which is, if I had a spare million lying around, I would probably try and incubate and fund more new local news organizations to start up, get off the ground, get sustainable. Because the state we're at now is while the big media is distracting themselves with, outrage, all the local news outlets have been steadily dying for about the last 10 years to the point where there's large places in the US that are total news deserts, which is to say there isn't a local newspaper.

There's simply no one covering covering that geography at all. The only thing you can you can do is kind of read the national stuff, but all of the issues that are actually relevant to you that actually make a difference to your day to day life, blackout. No information. So I think that's something long term we would love to solve. And if you gave me a $1,000,000 today, I'd get a head start.

It's about, you know, you've you've got this purpose to make the Internet better, make journalism better. So I think it's great, and thank you so much on the on on one user from one user of the Internet. Thank you. That's the one. It's it's great.

So he's like the I I say he's like Paul Graham is to y combinator. Ben Thompson is to kind of subscription based media on the Internet. The problem is I don't think there's any price, that you could buy Ben Thompson's favor with. He does very well from, his publication. It's called Stratechery. To the extent that I don't think he has a price. You know?

I wasn't I didn't set out to build a product and a company. I set out to get a blog post idea, just an idea out of my head in the form of a blog post, and I didn't really think anyone would be very interesting interested. But with hindsight, that was the MVP of the products. The MVP of the products was marketing the idea and seeing if anyone cared. Because if no one had cared, if if that blog post got 10 page views and went silent, it would not have been a good thing to invest 2 years of my time into to building and getting going with Hannah.

But because there was that immediate demand, it gave me so much signal of, alright. This is worth it. Like, this is worth investing time in. It it took away that uncertainty. And so whatever idea you have, if you're trying to start a project, I think just keep constantly trying to find the smallest version of it and ship that.

And the smallest version of it could be a tweet. It could be, here's my idea. Does anyone think this is a good idea? And then it's right up. But the the 0 for 1 hurdle of, you know, what I need to I need to announce something, but I want people to like it. I don't want them to hate it. Almost everyone gets stuck there, and so just relentlessly tell yourself, I need to get something out the door, I think, is is the most useful thing you can do.

But the the most interesting thing I think with AI coding, whatever the tool and so maybe this is more targeted at the the front end developers of your audience, is the AI makes a person who knows a little bit much smarter, and it makes a person who knows a lot doesn't really help much. So I can now code side projects myself that are full stack applications that work, that is not something I could do before. And that excites me no end. I can sit down with, you know, a chat box and a code editor and I can ship an end to end working app with a my design and front end skills augmented by a functional API back end database and a deploy pipeline. And the the sphere of what I'm able to do has grown so dramatically that I'm still, I've been excited by this for about a year and it's it's not abating that's kind of my my thing that's that's still there at the moment but I know great developers say yeah the the answers are shit but if you don't know how to do any of that stuff and it lets you do some of it the answers are pretty great and again like if I could get any of my projects to the point where people actually care about them well that point then I'll hire a great developer to come and fix all of ai's bullshit So that that problem solves itself.