Derek Sivers started CD Baby in 1998 because he wanted to sell his own CD online and nobody would let him. It became the largest online seller of independent music, over 150,000 musicians and $100M in sales. A few months before selling in 2008, he moved the whole company into an irrevocable charitable trust, so the buyer bought it from the trust rather than from him. That turned into $22 million for music education.
The book is that story in 40 chapters, most of them a page. It takes an hour. Every chapter is also free on his site, which is unusual and worth knowing before you buy anything. If you get something out of it, buy it directly from him rather than through a store. It’s $15 for every digital format including the audiobook, or $19 for the paper copy with the digital formats thrown in, and buying direct means the money goes to the person who wrote it. Feels like the right way to pay for a book that gives itself away first.
I read the 2015 Penguin edition; the 2022 third edition adds eight more chapters.
It’s a strange book to review because it keeps telling you not to take it seriously. Sivers says up front that he’s unusual, that what worked for him probably won’t work for you, and that he hopes you disagree and email him about it. Most authors saying that would be doing fake humility. It’s harder to dismiss from someone who gave the company away before he sold it.
Your business is a small world where you write the laws
This is the spine, and everything else hangs off it.
Sivers didn’t want CD Baby to get big, because it would eat his music career. So instead of being ambitious he went utopian, on the theory that idealism would keep it small. He wrote down what a distribution deal would look like in a perfect world, from a musician’s point of view. Pay me weekly. Show me who bought my CD, because those are my fans and not the distributor’s. Never drop me for selling too little. Never sell placement, because it isn’t fair to people who can’t afford it.
Four lines, written as a daydream, and they worked as a constitution for a decade. When an ad salesman called about banner ads, the answer was already written. He didn’t have to think about it.
The pricing came the same way. He had no idea what to charge, so he walked into the record store in Woodstock and asked how it worked if he sold his CD there. She said you set the price, we keep a flat $4, we pay weekly. He went home and typed that onto his site, plus a $35 setup fee because adding an album took him 45 minutes by hand. Six years and $10 million in revenue later, those were still the only two numbers in the business.
His claim is that a business plan should take a few hours at most, ideally minutes. A glance and common sense should tell you whether the numbers work, and the rest is detail. Hard to argue with a guy who ran a $10 million business on a number he got from a record store clerk.
The same instinct produces his position on corporate formalities: no lawyer, no Terms and Conditions, and a flat no to employee review plans and sensitivity training when he had fifty staff. His argument is that the people selling you that apparatus are selling fear. I think the instinct is right and the specific advice doesn’t survive contact with 2026, when you’re handling other people’s data by default.
Start with one percent of it
He had no funding and calls it an advantage. Couldn’t afford a programmer, so he bought a $25 book on PHP and MySQL and learned it. The desks were planks on cinder blocks for years. He watched funded friends spend $100,000 on things he built for $1,000, always justified as needing the very best, and never saw it help a single customer.
The transferable part is his rule for starting: take one percent of your grand vision and do it this week. Want to build a chain of schools? Teach one person something for money. Want to start an airline? Next time a flight cancels, offer everyone at the gate a chartered small plane if they split the cost, which is roughly how Branson started Virgin Atlantic.
CD Baby v1 took a few days. A list of CDs with buy buttons, and when you entered your details the site emailed them to him and he did the rest by hand. That was the whole product for a year. He spent $500 and was profitable in month two.
The related chapter is the one people quote most: ideas are a multiplier of execution. An idea scores between -1 and 20. Execution is worth between $1 and $10 million. You multiply them, which is why a brilliant idea with no execution is worth $20, and why he won’t sign an NDA to hear something simple.
If it’s not a hit, switch, and the broken test underneath it
Before CD Baby, Sivers spent twelve years promoting his own projects. Networking, pitching, marketing. He made progress, but always by force, always against doors that were locked or closing. Then CD Baby happened and the problem inverted: instead of manufacturing demand he was managing it.
His correction to the usual gospel about persistence is good. Success comes from persistently improving and inventing, not from persistently pushing the thing that isn’t working. Don’t spend years fighting uphill against a locked door. Go build something else.
But then he gives you the test for whether something is working, and the test is broken.
His instruction is to present each idea to the world, and if multiple people say some version of “yes, I need this, I’d be happy to pay you to do this,” go do it. If the response is anything less, don’t.
That is, almost word for word, the sentence The Mom Test identifies as the deadliest fluff in the world. It’s future tense. It’s hypothetical. It costs the speaker nothing to say and it’s the polite thing to say to an enthusiastic person who just described their project to you. Fitzpatrick’s first company took that exact signal at face value and burned around ten million dollars before finding out nobody would actually buy. Sivers is proposing the worst available signal as his go/no-go criterion, and he’s proposing it confidently, in a book people use to make real decisions.
The defense is that Sivers didn’t actually do what he says. What he measured was musicians mailing him $35 and CDs showing up at his house. That’s money changing hands, which is the strongest currency of commitment there is. He passes the Mom Test in practice and fails it in prose, which is a common shape for founder advice: people describe what they did in the language of what they now believe, and the language drifts.
The same drift shows up in his advice on prioritization. He says when you’re unsure what to work on, ask customers the open-ended question of how you can best help them right now, then focus on satisfying those requests. Fitzpatrick’s whole chapter on feature requests exists to stop you doing that. Requests are symptoms. The thing you want is the motivation underneath, which you only get by asking why they want it and what it would let them do. Sivers says obey. Fitzpatrick says understand, then decide.
Read together, the honest synthesis is this: Sivers is right that you should stop pushing things nobody wants, and Fitzpatrick is the one who can tell you whether anybody wants it. Sivers got away with a bad test because his customers were paying him before he’d asked anyone a question. Most people don’t get that.
The personal-life version of the same chapter has no such problem. No “yes.” Either “hell yeah” or “no.” Saying no to most things is what leaves room for the rare thing you’re actually excited about. That one’s about your own enthusiasm, which you can observe directly, so there’s nobody to lie to you.
Proudly exclude people
Most businesses try to be everything to everybody and then wonder why nobody notices. Sivers says to state clearly who you’re not for.
His example is the Hotel Café in LA, a venue with signs saying no talking during performances, where artists are encouraged to stop the show and tell a talker to go to any other club in town. Because it’s the one place you can actually listen, it became the most popular venue in the city. At CD Baby he told record labels their acts weren’t allowed, because the site was for musicians who hadn’t signed their rights to a corporation.
It’s a big world, you can loudly leave out 99 percent of it, and your 1 percent comes to you because you’ve shown them you value them.
He pairs this with an argument for many small customers over a few big ones. Landing one huge client means custom-building for a handful of people who might leave, and it means that client is effectively your boss. Thousands of small customers means nobody can make demands and you hear a real spread of opinion.
You don’t need a plan or a vision
A year and a half in, running CD Baby out of his house with one employee, Sivers sat down for a night of serious long-term thinking. What came out was an email to that employee speculating that someday they might have a thousand artists, might need a third employee, might need to network three computers together, might have to start moving CDs into the garage. He apologized for how grandiose it sounded. Years later, at a hundred thousand artists and eighty-five employees, that email was a running joke.
When journalists asked his long-term goal, he said he didn’t have one, that he’d passed his goals long ago, and that he was just trying to help musicians with whatever they needed that day.
This is the chapter Zero to One is written against. Thiel’s entire argument is definite optimism: have a specific plan, know your secret, commit to a future only you can see, because indefinite optimism produces a generation of people optimizing without direction. He’d file Sivers under indefinite optimism immediately, and the lean startup movement with him.
Sivers would say Thiel’s founders are pattern-matching their own luck into a plan after the fact, and quotes Steve Blank on the way past: no business plan survives first contact with customers.
They can’t both be generally right, and I don’t think the answer is a compromise. They’re describing different games. Thiel is describing something that has to reach escape velocity, where the customers don’t exist yet and no amount of asking hands you the plan. Sivers is describing a service business where the customers are right there telling you what they need, and the theorizing is what’s in the way. The question that decides which advice applies isn’t “which book is smarter,” it’s “does my thing have customers yet.”
Which, incidentally, is a question Fitzpatrick can answer and neither of the other two can.
The little things are the actual product
The most transferable part of the book and the part nobody copies, because it doesn’t scale.
CD Baby’s best marketing was an order confirmation email. The default said your order has shipped, thank you for your business. Sivers spent twenty minutes replacing it with something absurd about sterilized gloves, a satin pillow, fifty inspectors, a packing specialist lighting a candle, and the entire town of Portland waving the package off on the private CD Baby jet. People loved it enough to post it on their own websites for a decade. Go read it rather than take my word for it; the joke is in the escalation and summarizing it kills it.
The rest are the same shape. Customers kept calling to ask if they’d made the 5pm FedEx cutoff, so he added two lines of code showing a countdown next to the shipping options. Every outgoing email had a customized sender name, so Susan got mail from CD Baby loves Susan. Big favors cost a pizza, and they got one every few weeks. One order asked for a rubber squid, or a real one if no rubber was available, and since a customer in Korea had recently mailed the warehouse a packaged squid filet, that went in the box.
There’s a related chapter on clarity I keep thinking about. With two million customers, one unclear sentence in a mass email produced twenty thousand confused replies, costing a week of staff time and at least $5,000. So writing those emails took him all day. His observation is that people writing websites never get this feedback: unclear copy doesn’t produce complaints, it produces silence, which is the hardest signal to learn from.
Delegate or die, and the three ways it went wrong
By 2001 he had eight employees and was working 7am to 10pm daily because every decision went through him. He fixed it by changing how he answered questions. Instead of giving an answer, he’d gather everyone, explain the philosophy behind the answer, have someone write it in a manual, and tell them they could decide it themselves next time. After two months the questions stopped. He moved to California to make it clear the running of things wasn’t his job. The company went from $1M to $20M and eight to eighty-five people while he was away from it.
Then three failures, and they’re the best part of the book.
He hired someone for the single most critical job, delivering every album to every retailer every week, watched closely for a few weeks, and looked away. Months later he found that nothing had gone to Napster or Amazon since. Trust, but verify.
He empowered employees so completely that when they asked which profit-sharing plan to use, he said pick whatever you want. They designed one that returned all company profits to themselves. Canceling it turned eighty-five people against one, and he never saw the office again. The word he learned was abdicate.
And the $3.3 million one. He’d signed his dad’s paperwork years earlier without reading it, then let a bank teller talk him into making CD Baby an alias on that existing company because it saved ten minutes and $100. His dad’s company owned 90 percent of his business, and the IRS wouldn’t let him buy it back at the original price.
Being, not having
He wanted to be a singer from fourteen. Everyone told him for eleven years that he wasn’t one, including a mentor who listened to his first album and told him to stop trying. He practiced an hour a day anyway, and at twenty-nine he was good, at which point a stranger told him singing is a gift you’re born with and he was lucky.
Same with production, same with programming. He wrote all of CD Baby’s code himself even when his employees were furious about how slowly features shipped, even though he estimates it cost millions in lost business. His comparison: outsourcing the code of an internet business is like a band outsourcing the songwriting.
Having something is a means. Being something is the point. When you sign up to run a marathon, you don’t want a taxi to the finish line.
This is obviously self-serving and he mostly admits it. It’s also the honest answer to why anyone builds things themselves now that you don’t have to.
Enough
The chapter on giving the company away opens with Kurt Vonnegut and Joseph Heller at a billionaire’s party. Vonnegut marvels that their host has everything. Heller says he has something the billionaire never will, which is enough.
Sivers insists the trust wasn’t altruism and that he sacrificed nothing. He doesn’t own a house or a car, owning less makes him freer, and what he wanted was the pride of doing something smart and irreversible before he could talk himself out of it. The trust pays him 5 percent of its value a year while he’s alive and everything goes to music education when he dies.
What makes this an idea rather than a flex is that enough is a number, and if you don’t pick it in advance it silently defaults to slightly more than you currently have. Sivers picked his early, which is why the decision to sell took one day of writing in a diary rather than a year of negotiating with himself, and why he took the lower of two bids because that buyer understood his musicians better. Almost every founder story I’ve read is about someone who never picked the number.
My take
The thing I can’t get past is the hit assumption underneath the whole book.
Sivers built a store for himself, a friend asked to be added, then two more, then strangers calling because their friend Dave said he could sell their CD. Two newsletters mentioned him and fifty musicians signed up overnight. He was standing in a real structural gap at the exact moment the internet opened it, and his demand problem was solved before he knew he had a business.
That’s why he can be so relaxed about validation, so dismissive of plans, and so confident that if it’s not a hit you should switch. The advice tells you what to do when the answer is obviously yes or obviously no. Almost everything actually lives in the enormous space between, where the response is polite and small and you can’t tell whether you’re early or wrong. The Cold Start Problem is a whole book about that space, arguing that networks look exactly like failures right up until they don’t. Sivers would say having to push that hard is the signal. Chen would say every network business had to push that hard and the quitters quit right before it worked.
I notice I want Sivers to be right, because his version means I get to stop. That’s a good reason to be suspicious of my own reading.
The other thing that doesn’t survive to 2026 is his silence on attention. Distribution was the scarce thing in 1998. Independent musicians had no way to reach buyers, Sivers built the road, and he got word of mouth free from people who’d been waiting years for exactly that. He helped kill the problem he was solving. Now anyone can ship anything in a day and attention is the scarce thing, which he barely mentions because he never had to.
But dated isn’t worthless, and a book being wrong about one thing is not a reason to skip writing down what it got right. The utopia framing, the little things, the clarity penalty, the delegation failures: none of that has a shelf life. The growth chapters are a historical document about a specific moment, and reading them as one is fine.
What makes the book worth an hour is the thing underneath all of it, which almost no business book says out loud: that the point of the company was never the company. He built a small world he wanted to live in, discovered he was happier at five employees than eighty-five, and got out. Most books in this genre would call that a failure of ambition. He calls it the whole point, and he’s the one who wrote the four-line mission statement that ran a $100M business for a decade.
Concepts I’ll carry forward
- Write the utopia first. Four lines describing the perfect world for the people you’re serving, written before any code, treated as constraints. It makes hundreds of later decisions obvious and it’s cheaper than a roadmap.
- A business plan should take minutes. If a glance and common sense don’t tell you the numbers work, more spreadsheet won’t fix it.
- Start with one percent of the vision this week. Not a prototype of the whole thing. The smallest version that actually helps one real person, done by hand.
- Ideas are a multiplier of execution. A brilliant idea with no execution is worth $20. Stop protecting ideas.
- Stop pushing what isn’t working, but check the signal. Sivers is right that you should go invent something else instead of flogging a dead product. Just don’t use “people said they’d pay for it” as the test. Use what they actually did.
- Hell yeah or no. For your own commitments, where nobody can lie to you but you.
- Proudly exclude people. Say who you’re not for. The 1 percent shows up because you turned away the 99.
- Many small customers beats a few big ones. One big client isn’t a customer, it’s a boss.
- The little things are the marketing. The countdown timer, the sender name, the squid. They don’t scale, which is exactly why they work.
- You should feel pain when you’re unclear. Bad copy doesn’t get complaints, it gets silence. Write as if twenty thousand people are about to reply confused.
- Answer the philosophy, not the question. The way out of the delegation trap is explaining the reasoning and writing it down, so the next decision doesn’t need you.
- Delegate, but don’t abdicate. And trust, but verify. Handing over power isn’t handing over responsibility.
- Read what you sign. $3.3 million.
- Being, not having. Doing it yourself is slower and sometimes costs real money. That’s allowed if the doing is the point.
- Enough is a number you have to actually pick. Otherwise it’s always slightly more than you have.
Rating
4 stars.
It loses one because the confident advice about validation and growth is inseparable from circumstances that no longer exist, and because a book this deliberately anti-analytical has no way of telling you when its own advice stops applying. You need a second book to catch its mistakes.
It keeps four because half of it is failures described in enough detail to learn from, because the humanist argument underneath is the one most business books skip entirely, and because it takes an hour. Best value-per-hour of anything I’ve read this year, and I’d say that even though it’s free.