What this covers
This is a conversation in which Dan Shipper talks with Matt and Dom about how his media company, Every, has been built and rebuilt. Much of it is a post-mortem on early structure. A bundle of newsletters written by halfway-committed writers didn't build real audiences, and Dan's attempt to step back and manage writers the way a software company would failed because the product depended on the creator. What replaced it is the Creator-Team Media Model, which he ties to Bill Simmons, Sam Harris, David Portnoy and MrBeast. He describes hiring an Editor-in-Chief so that his hires leave him bored.
The business mechanics get the most space. Dan lays out a Monetization Pyramid running from free content to a subscription, then courses, then consulting, with each tier roughly ten times the price of the last. He explains audience equity, under which a departing writer can buy their list on seller-financed terms, and Dom questions whether it really fixes the platform-creator incentive problem. Other topics include why Lex was spun out, Think Week borrowed from Bill Gates, and why a founder who both writes and builds has a cost advantage.
The last stretch turns to AI. Dan compares today's reaction to language and diffusion models with the early scorn for the typewriter and recorded music. He also describes his own swings between alarm and acceptance. The contested predictions concern the destination chatbot: whether consumers settle on one interface, OpenAI's head start, and the threat from models built into Google, Facebook or Apple. He ends on why narrowing to ChatGPT gave Every a Legible Promise while he now plans a broader rebrand.
Dan Shipper argues that media businesses succeed by pairing a personal creator with a product-building team, monetizing through a wide pyramid of offers, and narrowing to a legible promise, while AI tools amplify creative capacity and the business-model problems of writer alignment remain unsolved.
- Media products are personal, so removing the creator from a writing business destroys product-market fit.
- Media is a hard business, so creators must monetize across many tiers (free, subscription, courses, consulting) rather than rely on one stream.
- Owning a legible promise in a niche helps audiences keep a slot in their heads, which is why ChatGPT focus worked.
Media businesses work when the original creator stays central and builds a team around them.
- Programming is the only way to build a business where the only cost is your time, which made software an attractive path for a young founder who could not yet be a professional creator.
“programming is the only way that you can build a business where the only cost is your time”
- Dan has held a consistent writing identity since childhood, and he chose founding businesses partly because he believed he lacked the life experience to write well, while running a company and writing seemed structurally incompatible.
“running a business, it's just really hard to run a business and write at the same time”
- Dan delegated operational and managerial work to an Editor-in-Chief and an entrepreneur in residence so that he can focus on writing and podcasting, adopting the goal that his hires make him bored, as Bill Simmons does.
“for everyone that we hired that's sort of at that level, I'm always like, your goal is that I'm bored.”
- The effective media business pattern is for the original creator to carve out time to create and then build a team around them, which Dan argues is both a better business decision and a proven path used by Bill Simmons, Sam Harris, David Portnoy, and MrBeast.
“The core of the set-up is, the original creator carves out time for themselves to do the creating and then builds a team around them”
- Dan views Every as a way of backing into being a writer, since founding a media company gave him legitimate reasons to write constantly while also running a business.
“Every, which is the company that I run now, was sort of my way of backing into being a writer”
- When Every hired writers and stepped back to manage them as a software company would, it failed, because most media products are personal and removing the creator means losing product-market fit that must be found again.
“if you have product market fit with a particular writer, a particular creator, and then you remove that creator, you lose product market fit, you have to find it again”
AI should be approached through curious experimentation, since it speeds up building long-held ideas.
- Dan's view of AI has moved through cycles of alarm and settling, and he now sees these tools as giving tremendous creative power while acknowledging that some people whose skills lose value will need social support.
“these tools give us these tremendous creative powers... to a significant degree, we're going to have to have really important social conversations”
- A person's initial reaction to a new technology says very little about the technology's usefulness or how well it will fit into their life, so skeptics may benefit as tools mature, and curious experimentation is the key attitude.
“your initial reactions to technology say very little about the technology and how useful it is and how well it will fit into your life.”
- Historically, new creative tools such as the typewriter and recorded music were initially derided or viewed as inauthentic, but people who used them anyway became the progenitors of new art forms, and society eventually accepted them, which Dan expects will happen with language and diffusion models.
“when you get new tools, you get new forms. The forms of art are initially like derided or looked at with suspicion or thought to be inauthentic.”
- Watching the AlphaGo documentary about a model beating the best Go player made Dan feel sick for a day, illustrating how his feelings about AI swing back and forth before settling.
“I remember a year ago, I watched the AlphaGo documentary, where it's like about the AlphaGo model beating the best Go player in the world, and how I was just sick for a day.”
- AI displacement of skills will not be as overnight or widespread as alarmists fear, but some people will be affected, and those people should be supported through social conversations about how to help them.
“I don't think that, that's going to be as overnight and as widespread as like the alarmists fear. But I do think that it's possible that it will affect some people.”
- Using AI, Dan built an internal app end-to-end in two days that would have taken weeks otherwise, and he says long-held ideas are finally getting built.
“I built this app end-to-end that we use as an internal tool in two days, a couple of weeks ago. I would have been able to build it, but it would have taken me weeks.”
- The AI podcast grew out of Dan's viral Superorganizers newsletter, where he interviewed people about their productivity systems by screen sharing and writing up screenshots, and the podcast format adapts that concept to AI.
“it is the podcast form with an AI flavor of a very successful newsletter that I wrote for a long time that became the basis for Every called Superorganizers”
Media businesses should monetize through a layered pyramid running from free content up to consulting.
- Because media is not a great business, creators must monetize in as many ways as possible, but focus on a core still matters; diversification gives legs of the stool that let a business adapt when subscriptions or ads are weak.
“So this is one area where I think focus isn't less important than diversity, because when subscriptions are going well, ads aren't going to be going as well.”
- Consulting, speaking, and incubations are meant to fund and funnel back into the core writing business, so that Every can hire more writers and create a creative institution in business writing, though the adjacent businesses will also be grown on their own.
“ideally, the incubations and the consulting and the speaking and all that kind of stuff, is funneling back into that part of the business”
- Consulting has been surprisingly profitable for Every, to a degree Dan describes as crazy, and he sees it as a strong upper tier of the monetization pyramid.
“We're also doing a lot of consulting and that has been so far crazy, how profitable, I think it can be.”
- A teenage app business sold software as a one-time purchase rather than a subscription, and Dan says he would have earned far more if Stripe had existed to make recurring billing easy.
“it was a one-time purchase instead of a SaaS fee. I would have had a lot more money if I could have done a SaaS fee. And the reason why I couldn't is because Stripe didn't exist”
- Media businesses should be structured as a pyramid from free content to a $10–20 monthly subscription, then courses, then high-priced consulting or speaking, because a fraction of each tier buys the next tier at roughly ten times the price, and businesses that skip upper tiers dramatically under-monetize.
“if you're not doing that pyramid thing as a media business, you're dramatically under-monetizing, and the business is hard enough that you probably shouldn't be under-monetizing.”
- Having product-building DNA at the core of a media company is a rare secret weapon, because most media companies struggle to build tools and products, and launching small tools is a good way to get people to subscribe.
“having product DNA at the core of Every is, I think, a kind of secret weapon. It's sort of rare for media companies to actually be good at this.”
AI markets reward early leaders and embedded platforms over destination chatbots.
- Lex was spun out because it was structured like a venture-backed business that could grow fast and had viral capability, whereas Every's core media business is a different kind of business, and Dan would make the same decision again.
“Lex is styled like a venture-backed business. It could grow really fast... It's two different businesses.”
- Nathan built Lex, an AI writing tool resembling Google Docs with AI built in, and it went viral early in the GPT-3 wave, prompting Every to spin it out, where it raised a $3 million seed round.
“it just like went super, super, super viral. It was at the very beginning of the GPT-3 wave”
- Ordinary consumers will not switch between AI models but will settle on one interface and interaction style, and OpenAI has a head start as the first real consumer use case with the best mobile and web apps.
“I actually don't think that, that will happen for consumers, just normal regular consumers whose job is not to use AI all the time. I think people will get used to a particular interface”
- Yahoo built a PageRank-like system about 1.5 years after Google, yet never figured out how to win search, which Dan finds puzzling and contrasts with the current AI market where OpenAI is not facing that same kind of lag.
“it sounds like Yahoo! built PageRank 1.5 years after Google did, they just never figured it out for some reason.”
- Chatbots carry more risk than search engines because they present a single canonical answer rather than links to sources, so OpenAI can take less legal and reputational risk than Google, though workarounds like user-built branded GPTs may mitigate this.
“search engines aren't saying, here is the answer from Google. They're saying, here are a bunch of links that have the answer.”
- Dan's light-held view is that Claude and Google models have improved and caught up faster than he expected, and he suspects OpenAI has unreleased capabilities since GPT-4 was trained earlier than its competitors' models.
“I think that the models from Claude and Google are better and caught up faster than I maybe thought that they would.”
- OpenAI's strongest position is as the destination chatbot that people go to in a browser or app, but language models embedded in ubiquitous platforms like Google, Facebook, or Apple could reach far more users and reduce traffic to chatgpt.com.
“I would make a distinction between a destination chatbot, like a thing that you type into your browser to go chat with, and just using language models generally.”
- Dan uses Claude for writing and GPT-4 for programming, splitting between models, but expects ordinary consumers to settle on one interface rather than switch between models.
“I kind of just know that Claude's a better writer and GPT-4 is better for programming, so I kind of like split them.”
A clear, legible niche promise helps creators earn an audience's limited attention.
- Creative people find it psychologically hard to limit themselves to one niche because they do not want to pigeonhole themselves, and it is also hard to know which niche to choose.
“it's really hard psychologically to be like I'm going to limit myself to this little thing. I don't want to pigeonhole myself.”
- After Lex launched, Dan deliberately chose to focus on AI because it intersected with his core interests in productivity, tools for thought, and software psychology, and AI changes the conversation on all of those topics.
“AI changes the conversation on all of those topics, so it was like both are very, very important.”
- Beyond niche focus, what matters is a legible promise that occupies a slot in someone's head, because people only have limited mental slots, so a creator must consistently deliver on a clear promise to be the one people want to hear from when something happens.
“it's not necessarily about being niche, it's about having a legible promise that sits in a place in someone's head.”
- Ben Thompson is a strong example of someone who has a clear promise and reliably delivers on it through regular value, which Dan emphasizes for himself and for everyone at Every.
“It's like Ben Thompson has a thing, which is regular delivery of value or something like that”
- Narrowing the podcast and coverage to ChatGPT was a deliberate and effective branding choice because it creates a legible promise, and Dan says it was very important early even though he now plans to rebrand more broadly.
“it's not necessarily about being niche, it's about having a legible promise that sits in a place in someone's head.”
Profit-share writer economics and long-term incentive alignment are hard, often leaving writers underpaid.
- Most writers under Every's profit-share deals never exceeded the guaranteed minimum payment, with Evan being the only writer to come close and eventually surpass it, meaning writers did a lot of work for little payback.
“they just almost never hit, got above the minimum. Evan, who is still our Lead Writer for Napkin Math and then for Every, is the only person that ever came close to that and eventually surpassed it.”
- Audience equity lets a departing writer buy the list of people who signed up during their tenure on a seller-financed basis, paying a share of revenue from it over time, so the company benefits when a creator leaves rather than losing the audience entirely.
“you have the option if you want to buy that list from us in a seller-financed way. So you don't have to pay any cash to us upfront”
- Recruiting writers who were only halfway committed, writing once or twice a month, made it almost impossible to build a real audience, and most of the writers who want to write seriously either go all-in on their own or are rare.
“it's just almost impossible to build actual audience for a newsletter if you write once a month or twice a month”
- A major unsolved challenge for multi-arm media companies is keeping writers and software people aligned and respectful of each other so they support rather than fight one another, and Dan thinks AI makes this more achievable at a higher scale with fewer resources.
“where the writers don't respect the software people and the software people are like, you're just writing”
- Puck's model works because its writers are established professionals who were underpaid elsewhere and now share upside, which makes their supply of writers easier to hire and more committed than typical Substack creators.
“their crop of writers are all professionals. So it's not people who would necessarily normally go off and start their own Substack”
- Long-term incentive alignment with writers is hard because writers who start out grateful for a platform later see that they can go alone once they build their own brand, so alignment that works early often breaks down as people become big.
“long-term incentive alignment is really hard because you might bring in a writer and when they're starting out, it's like amazing that they get to write all the time. But as they build their brand, they're like, well, I can do this on my own.”
Consolidating many small newsletters into one shared publication cuts costs and focuses attention.
- Every was founded at the start of Substack's popularity, with Nathan Baschez, Substack's first employee, providing insight that newsletter businesses could bundle several business-focused writers into one platform.
“my Co-Founder, Nathan, was the first employee at Substack. So we understood that business intimately.”
- Puck will face the same problem Every faced: running many separate newsletters is expensive and splits focus, making it hard for audiences to hold a clear mental picture of what the brand offers, so consolidation is a likely future move.
“when you have so many different newsletters, it's really expensive. You're kind of splitting your focus.”
- Every's traffic came heavily from Twitter/X, and after the algorithm change page views fell, so it is harder to grow top of funnel and the company is shifting attention to monetizing its strongly engaged existing audience.
“it's just harder to grow top of funnel in this environment for us, specifically a lot of our traffic came from Twitter or X. And now the algorithm is different, so page views are down.”
- The newsletter economy's core problem is that writers peeling off to publish alone make the reader and writer experience worse, since readers must pay for many separate newsletters and writers work in isolation, so Every aimed to make a shared place for writing together.
“one of the core underlying issues in the newsletter economy is that you have to pay for a bunch of individual newsletters”
- Running many small newsletters costs nearly as much as running large ones because expense does not scale with list size, so Every merged them into one publication with columns to focus attention and resources.
“Running each individual newsletter is just as expensive, no matter how many emails you have.”
Shipper's career has run through writing and collaborative media rather than software.
- Newsletters run by writers who write once or twice a month cannot build a real audience, and most people who want to write are only partly committed, so a bundle of many small newsletters with halfway-committed writers was the wrong starting structure.
“we started off in the wrong way by just starting a bunch of newsletters with people who were halfway committed. And they did great work, but it's just almost impossible to build actual audience for a newsletter if you write once a month or twice a month.”
- Shipper decided in third grade he wanted to be a writer, then chose software because he thought he lacked enough life experience to write well, while writing stayed a constant thread throughout his career.
“I decided I wasn't a good enough writer because I hadn't had enough life experience. So that's when I decided to get into software”
- Shipper believes talent is the rarest thing in media, which makes scaling a creator-driven business difficult.
“And talent is the rarest thing.”
- Shipper describes Every as his way of backing into being a writer by founding a media company that gave him many excuses to write, and he found media fun and more appealing than software at the start.
“Every, which is the company that I run now, was sort of my way of backing into being a writer”
- Shipper felt that running a business and writing full-time were very incompatible, so he kept building companies while hoping to eventually make writing the core of his work.
“Running a business, it's just really hard to run a business and write at the same time.”
- Shipper's college company Firefly built co-browsing software for customer service, raised about $20,000 from Dorm Room Fund, and was sold to Pega, where he ran the business for a couple of years.
“we raised like $20,000 from Dorm Room Fund. And so we sold it to Pega”
- Every was founded as a place where writers could write together in an age when any writer could go alone, offering profit sharing and list sharing to incentivize collaboration, starting with multiple individually headed newsletters.
“can we create a place where people can write together that is designed for an age where any one of us could write alone if we chose to, and make money doing it.”
Risky creative work requires unscheduled slack that deadline-driven publishing cannot provide.
- Every's Think Week, borrowed from Bill Gates, is a periodic week without meetings or publishing duties in which staff reflect, improve inefficient processes, and pursue net-new projects, because deadline-driven creativity needs unscheduled slack to produce breakthroughs.
“creativity, it can thrive under those sorts of deadline circumstances. But then also, I think you need times where you're just like unscheduled”
- Experiments that might fail cannot be attempted during normal deadline-driven publishing because a failed attempt would cause a missed deadline, so dedicated unscheduled time is required for risky creative work.
“I just want to try making, and I wouldn't be able to do it during the regular time because it would be too much of a risk.”
- The term Think Week comes from Bill Gates, who took a week away to read books, and Every adopted the idea quarterly as a way to create unscheduled time.
“creativity, it can thrive under those sorts of deadline circumstances. But then also, I think you need times where you're just like unscheduled”