The arithmetic that kills ideas
Divide the $5k target by a price to get a customer count, then run the churn multiplier to see the real, recurring recruitment quota it hides — most of the seven projects fail before a line of code gets written, which is exactly the point.
Every project on your shelf now has a verdict, and at least one of them carries the rest of the record too: a measured maintenance number and a candidate revenue shape, plus a named payer wherever buyer and user could come apart. This lesson is the one that takes some of them away. It is one line of division, it takes about ninety seconds per project, and it is the cheapest test in this entire course — which is exactly why it should run before anything else gets built, not after.
Target ÷ price = customers
The whole thing is that. Pick the revenue you are aiming at, divide by what one customer pays, and the quotient is how many customers have to exist. No conversion rates yet, no funnel, no growth curve. Just the division.
The most famous statement of it is Kevin Kelly’s 1,000 True Fans, published on kk.org in March 2008. A thousand true fans — his definition is “a fan that will buy anything you produce” — each spending a hundred dollars a year gets you a hundred thousand dollars a year. One multiplication, run backwards from a living wage.
Kelly attaches two conditions to that arithmetic in the same essay, and both are load-bearing rather than decorative. You have to keep the full hundred dollars, meaning you sell direct with no intermediary taking a cut. And you have to hold a direct relationship with each fan — which is not a precondition you satisfy once, it is the ongoing work the model quietly assumes you are doing forever.
Watch what the price term does, because it does almost all of the work. At $5 a month, a $5,000 monthly target needs a thousand paying customers. At $50 a month, it needs a hundred. Same target, same product, and the required customer base moved by a factor of ten because one number on a pricing page moved. Nothing else in this course has that kind of leverage.
Where people get burned
Kelly himself found the price term is where the model breaks. In The Reality of Depending on True Fans (April 2008) the ambient musician Robert Rich disclosed his actual finances: at the five-to-ten dollars a download or CD actually commands, a thousand fans caps out nearer ten thousand dollars a year — an order of magnitude under the model. Touring added roughly the same again for three to four months of work, and the total landed around a local garbage collector’s income. The arithmetic was never wrong. The price substituted into it was.
The churn multiplier
The customer count you just computed is not a finish line you cross once. For anything recurring it is a standing number — a level you have to hold — and holding it costs money every month forever, because some of those customers leave.
Monthly churn is the share of your standing base that cancels in a given month. At five percent monthly churn, a base of a thousand customers loses fifty of them by the end of the month. So before you have grown by a single customer, you must acquire fifty new ones just to be exactly where you started.
That is the treadmill, and it is why this section exists. The division hands you what looks like a one-time number. Churn converts it into an acquisition rate you must sustain indefinitely. Those are completely different commitments, and a solo project sized against the first one will be crushed by the second.
Compounding makes it worse than the monthly figure sounds. Five percent a month, applied twelve times over, leaves a little over half of a cohort standing at the end of a year. A base you spent a year building is a base you spend the following year rebuilding.
The calculator below runs both at once. Put your own target in, then break it: raise the price and watch the customer count collapse, raise churn a couple of points and watch the monthly replacement figure make the whole thing look like a job.
625
customers needed
32
new customers needed every month to stand still
That first figure is the size of the customer base the target demands, and the second is how many replacements churn quietly takes back out of it every single month, forever, just to hold that base steady.
The same division against each revenue shape
The lesson on choosing a revenue shape gave you seven to pick from: subscription, one-time purchase, advertising, marketplace, services, physical goods and licensing. The division runs against all seven, but price and customer do not mean the same thing in each one — and substituting the wrong quantity is how a project passes an arithmetic test it should have failed. Several of the seven share an arithmetic, so they are grouped here by what actually changes.
Subscription — the straight case
The one the calculator models. Price is per customer per month, the customer count is a standing base, and churn is the treadmill you just met. Its own row in the shape lesson says the same thing in words: hundreds of customers who don’t leave. This is what the arithmetic was designed for, which is part of why it is so easy to be optimistic in.
One-time purchase and physical goods — a quota, not a base
Both are paid once per buyer, so the honest reading is brutal: there is no standing base at all. Every month starts from zero and the quotient is a number of new strangers per month, every month, forever — the shape lesson’s phrase for it is a fresh crop of new buyers. In churn terms these shapes churn completely, by construction, every single period. Read the answer as a monthly recruitment quota rather than a customer base, because that is what it is.
Physical goods then change the numerator’s partner. Divide by margin per unit, never by price. Material, machine time, packaging, postage and the returns you did not budget for come out of every object, and the shape lesson is explicit that this cost does not fall towards nothing as volume rises the way software’s does. Dividing a target by the sticker price here understates the required unit count by whatever fraction of that price you never get to keep, which is usually most of it.
Services and licensing — per deal, not per customer
Neither prices per customer. A client pays per engagement or per hour; a licensee pays per contract. The division still runs, but the quotient is a number of deals, and it lands small enough to count on your fingers. That changes the question the arithmetic asks. It stops being can I attract enough people? and becomes can I name them? — and if you cannot list the specific humans or companies, you have failed the test at a much smaller number than anyone fails a subscription at.
Each has a second term the division alone will not show you. For services the binding constraint is usually not the deal count but your own available hours, and the revenue stops the day you do, so the arithmetic has a ceiling written into it that no price increase removes. For licensing it is the sales cycle: the shape lesson describes very few payers, each found and closed by hand over months, so a quotient of a handful is not the easy answer it looks like — it is a handful of multi-month hunts running in your evenings.
Advertising and marketplace — the price term is not a price
In both, nobody hands you the number you were about to divide by.
Advertising pays per thousand views, and the payer is an advertiser rather than the user. So the same division hands you a monthly traffic number instead of a customer count. That conversion is the useful part — it turns a revenue target into a specific quantity of visitors you must attract every month, which is far harder to be vague about than “more traffic”. The shape lesson does this arithmetic properly, with researched rates, and the answer it reaches is millions of pageviews a month. For most finished-but-quiet sites, that is the moment the idea dies.
Marketplace pays you a cut rather than a price, so the divisor is your take rate multiplied by the size of a typical transaction, and the quotient is transactions per month rather than customers. Then comes the part the division cannot see: that number assumes both sides of the market already exist. The shape lesson names liquidity as the real gate, and a cold start is hand-built sales work that happens entirely before the arithmetic starts producing anything at all.
Retrieval check
Two projects both need a thousand customers to hit target. One sells a monthly subscription with five percent churn; the other sells a one-time download. Which one has the harder acquisition job, and why is the intuitive answer wrong?
Check your answer
The one-time download, by a wide margin. The intuitive answer is the subscription, because churn is the thing this lesson made frightening — but five percent monthly churn means the subscription needs fifty new customers a month to hold its base steady.
The one-time product has no base to hold. It needs a thousand new customers every month, permanently, because last month’s buyers pay nothing this month. It is the hundred-percent-churn case. Subscription churn feels like the villain because it has a name and a percentage attached; the one-time shape hides the same cost by never quantifying it.
When the number comes back impossible
It usually does. That is the lesson’s title and it is not pessimism — finding out here costs you ninety seconds, and finding out after six months of evenings costs you six months of evenings. There are exactly three legitimate responses, and one very popular illegitimate one.
Raise the price
The first move, because it is the term with the leverage. Li Jin argued precisely this in 1,000 True Fans? Try 100 (February 2020): invert Kelly’s arithmetic to a hundred fans paying a thousand dollars a year, and the same living wage needs an audience an order of magnitude smaller. She backs it with Patreon data showing higher-tier pledges growing. Worth knowing who published it — Andreessen Horowitz is a venture firm invested in creator-economy platforms, including ones the piece cites, so this is thought leadership that happens to validate a category its author’s employer has money in. The mechanism still holds; the enthusiasm is not neutral.
Change the buyer
If the price cannot move with the buyer you have in mind, the buyer is the thing that is wrong, not the price. This is the distinction between who pays and who uses, arriving with a number attached: a monthly price that is absurd to a hobbyist is a rounding error inside a company’s tooling budget. Changing the buyer changes the price ceiling, which changes the division, which is often the only thing that rescues a project.
Stop
Sometimes no reachable price and no reachable buyer produce a customer count you can plausibly get to. That is a real answer and it is the answer this lesson exists to produce. Feed it back into the verdict: park it with a revisit date, or kill it. A project that fails the arithmetic and gets built anyway was never really a business decision — it was a build you had already decided on, looking for permission.
The move that is not allowed
Inventing a conversion rate that makes the number work. “If two percent of visitors convert, I only need…” is not arithmetic, it is a wish with a decimal point in it. A conversion rate is a measurement you do not have yet. Until you do, the honest form of the answer is a customer count plus one sentence naming where those specific humans would come from, and if that sentence cannot be written the project has failed the test.
Kelly is the cautionary tale here too, and it is to his credit that he published it himself. In The Case Against 1000 True Fans (April 2008) he collected hard financial data from seven creators supporting themselves this way plus partial data from roughly two dozen more, and reported he was unable to find much evidence that anyone was actually living on a thousand — or even five thousand — true fans. The ones partly succeeding leaned on high-priced goods rather than cheap ones, and on continuous labour finding and servicing fans. By The Stars of 1,000 True Fans (March 2011) real examples existed, and Kelly reported they typically needed nearer a hundred thousand fans, because digital prices and margins came in below what the original model assumed. Three revisits, all pointing at the same term: the price.
Now run it on your own two
Hands on
Fill the target arithmetic for every pursue project
Done when: Each “pursue” project in PORTFOLIO.md has a price, a customer count, and one sentence on whether a path to that count exists.
- Take the first of your pursue projects and write down the monthly revenue it would need to carry — its slice of the target, not the whole target. Add its maintenance cost to that slice: the measured number if this is the project you measured, the labelled estimate otherwise. A project has to clear its own upkeep before it earns anything.
- Write a price. Not a researched one yet — the positioning question comes later — but a specific number you could say out loud to a stranger without flinching. Vagueness here is what lets the rest of the exercise stay comfortable.
- Divide, and write the customer count into
PORTFOLIO.mdunder Target arithmetic. Use the calculator above rather than doing it in your head, and put your honest churn guess in while you are there. - Restate the count in whatever units its revenue shape actually deals in — the shape you chose for this project, or, if you have not chosen one for it yet, the shape that most plausibly fits it. One-time or physical goods: new buyers per month, and divide by margin rather than price. Services or licensing: a number of deals, plus the names. Advertising: a monthly traffic figure. Marketplace: transactions per month. The number that goes in the file should be the one that matches the shape, not the generic customer count.
- Write one sentence on whether a path to that count exists — naming where the people come from, not asserting that they will. “Roughly this many people search for this every month and the top result is weak” is a path. “It only needs a small percentage of the market” is not.
- Repeat for your second pursue project if you kept two, then compare the sentences side by side. One is usually noticeably weaker, and that comparison is worth more than either number on its own.
- Bring both rows into the chat. I’ll push hardest on the path sentences — an invented conversion rate wearing a percentage sign is the usual way this exercise gets quietly failed.
What this does not cover
You have just discovered that the price term does more work than every other variable combined, and you picked yours by instinct in about four seconds. That instinct is systematically wrong in one direction: engineers underprice, reliably, for reasons that have very little to do with the market and a lot to do with how it feels to attach a number to your own work. What comes next is where that number actually comes from — a price as a claim about who the product is for, rather than a calculation about what it cost you to make.
Read this next — primary source
1,000 True FansKevin Kelly, kk.org, March 4, 2008 (rewritten since) — free, about 15 minutes
The cleanest statement of target ÷ price = customers anyone has written: a thousand fans at a hundred dollars a year is a hundred-thousand-dollar living, and everything else follows from that one line. Read it with its own sequels, though — Kelly spent the years after publishing it arguing with it. Within two months he had run the real finances of a working musician, which came in an order of magnitude below the model, and published a piece titled “The Case Against 1000 True Fans” in which he went looking for people actually living this way and largely failed to find them. The version that loads at that URL today is his own rewrite, which quietly relaxes “1,000” to “thousands”. Both halves are the point: the arithmetic is sound and the inputs were optimistic, which is precisely the failure mode waiting in your own spreadsheet.
Stuck, curious, or think this lesson is wrong? Ask your teaching agent. The lessons are the scaffold; the conversation is where the learning gets unstuck.