In October 2019 Daniel Vassallo opened pre-orders for a 173-page PDF about Amazon Web Services. Fifteen months later he published a line-by-line accounting of what that PDF and one follow-up video had earned: $237,207 in sales and $210,822 in profit between October 2019 and September 2020, an average of $24,802 a month. There was no company, no funding, no launch team. There was a Twitter account, a Gumroad page, and two files. What makes the story unusually useful is that Vassallo published the attribution too: which tweet made which amount, which price change worked, and which channel he gave up on.
Product one: 160 hours, priced at $24
The first product was The Good Parts of AWS, a 173-page PDF. In his own breakdown of the nine months, Vassallo says it took roughly 160 hours to produce. He opened pre-orders on 3 October 2019 at $24 and shipped the finished file on Christmas Day 2019.
The price then moved several times. It went to $28 at launch, up to $38 after about a month, and down to $15 in mid-March 2020. That last cut was not a small tweak. Vassallo's own conclusion from the experiment is that the lower price produced more volume, more word of mouth and more satisfied readers than the higher one did, and the momentum from the $15 period fed everything that came after.
For anyone whose instinct is to charge as much as possible on day one, that is worth sitting with. Builtplain has written before about the opposite failure mode, pricing a first product too low to defend. Vassallo's case is a reminder that the right price is an experiment with real evidence attached, not a personality trait.
Product two: 16 hours of work, released in April
The second product was a single 100-minute video called Everyone Can Build a Twitter Audience, released on 22 April 2020. Production time, by his account: about 16 hours. That is the ratio worth noticing. The AWS PDF took ten times the effort of the video, and the video went on to become the better-known product.
He launched it at $99, dropped it to $89 in week three, and kept lowering it after that. By January 2023, when the startup publication Failory reviewed the course, the reviewer paid $10 for the video alone and noted a $20 tier that added a spreadsheet of Vassallo's tweets and their stats. Failory also records that the course sold more than $6,000 in copies within four hours of the announcement tweet, and that the AWS ebook had passed $80,000 in five months by May 2020.
Where the money actually came from
This is the part almost nobody publishes. Vassallo listed the top revenue sources for each product. For the AWS PDF: the announcement tweet brought $16,659, a Reddit ad brought $13,734, and the link in his Twitter bio brought $12,303. For the video course: the announcement tweet brought $20,015, a tweet announcing a $10 price reduction brought $13,438, and a follow-up discount tweet brought $8,849.
Smaller channels filled in the rest. His personal website produced $9,326. A single email to 880 recipients produced $6,367. A Hacker News post produced $5,159.
Read those numbers as a ranking rather than a recipe. The single largest line item in the whole business was a tweet announcing that a product existed, sent to people who already followed him. In the earlier of his two public write-ups, covering the first four months and $104,754 in sales, he attributes 95% of the revenue to his Twitter audience directly or indirectly, and 5% to a paid campaign. He also gives the audience's scale: in his first 14 months on the platform, his tweets collected 35 million impressions and 1.7 million engagements.
His valuation of that asset is blunt. "If someone were to offer me $500K to take away my Twitter account, I definitely wouldn't take it," he wrote. That is the same point Builtplain keeps arriving at from other directions: distribution, not the product, is the scarce thing, and the list you own beats the platform you rent.
The stunt that was not really a stunt
One experiment stands out. Vassallo ran a pay-what-you-want promotion for one hour with a $1 minimum. It produced about $8,000 in revenue across 2,559 units, and the tweet announcing it passed 150,000 impressions.
Look at the second number rather than the first. Two and a half thousand people got the product in an hour. For a maker whose entire distribution depends on word of mouth, buying 2,559 new readers for an average of roughly three dollars each, and getting paid to do it, is a customer acquisition channel disguised as a giveaway.
What he says failed
Vassallo is specific about the mistakes, which is rarer than the wins.
- The $99 launch price. He believes the high initial price dampened the launch spike, and estimates the product would have done better at $35-45. His reasoning is about the sequence, not the number: "It's tough to get $100 from your most loyal fans, and then go promote the same product at $50 a few weeks later."
- Paid ads. Reddit worked briefly. He reports paying about $0.50 per click at the start and around $0.10 during the early COVID period when advertisers pulled back, then finding the rates uncompetitive by mid-June. He looked at other platforms, could not find reasonable rates, and gave up quickly.
The honest reading of the ad experiment: it worked while the auction was cheap, and stopped working when it was not. That is a market condition, not a skill.
The parts a reader can actually copy
Strip out the personality and a fairly plain sequence remains.
- Write the thing you already know cold. The AWS guide came out of professional experience, not research. It was a filter on a subject he understood, which is why 160 hours was enough.
- Sell before you finish. Pre-orders opened 12 weeks before the file was delivered. Money arrived before the product did, and so did the evidence that it should exist.
- Make the second product cheap to build. Sixteen hours of recording, sold to the same audience, on a topic that audience had watched him live through.
- Announce it more than once. Three of his highest earning events were tweets: the launch, a price cut, and a follow-up on the same price cut.
- Test the price downward. Both of his products ended up cheaper than they started, and both did better afterwards.
- Count where the money came from. Gumroad and most checkout tools will tag revenue by referrer. Very few sellers ever look.
The part that is not repeatable on demand
The audience came first, and it was not small. Thirty-five million impressions in 14 months is not a side effect of publishing a product; it is the reason the product sold. Vassallo did not run a campaign to build it and cannot hand anyone a schedule for reproducing it. Someone starting today would be building on a different platform, in a noisier market, and the four-hour, $6,000 launch is a function of the audience, not the file.
Timing helped too. The March price cut and the cheap Reddit clicks both landed in the specific weeks of spring 2020 when attention was abundant and ad budgets were not. He says so himself about the ad rates.
What survives the caveats is the shape of the thing. Two files. A checkout link. An audience built by writing in public first and selling second, which is where first customers usually come from for products this size. Vassallo's contribution was not the discovery that you can sell a PDF. It was showing his work afterwards, down to the dollar, so that the rest of us could see which parts were skill and which were weather.
Discussion
Sign in with Google or just a name. No email link, no password to remember.