The startup finance terms that cost you money are not the ones you have never heard. They are the ones you think you already understand: ARR, valuation, pre-money, preference, retention. Each of those has a second, stricter meaning that only shows up in a document, and the gap between the two meanings is where founder money goes.
So this is a glossary ordered by cost rather than by alphabet. Nine terms, each with what it actually means, the arithmetic that proves it, and the one question to ask in the room. Every source is named and dated, because two of the most repeated rules in this subject turn out, when you trace them, to be somebody's blog post.
The film runs all seventy-nine terms in seven chapters, and the full annotated list lives at the finance terms sheet if you want the dictionary rather than the argument. This page is the argument.
Which startup finance terms actually cost founders money?
The ones with two meanings. Every term below has a casual meaning used in conversation and a precise meaning used in a document, and every dollar that surprises a founder lives in that gap.
You can sort the whole subject by where the gap opens. Some open in your own reporting, where you are the one being misled by your own numbers. Some open during the raise, where the gap is negotiated and priced into a document you sign. And some do not open until the exit, where the gap becomes a wire transfer to somebody else. Nobody is hurt by not knowing what EBITDA stands for. Founders are hurt by not knowing that the option pool comes out pre-money.
Bookings, billings, ARR, run rate: why four numbers all get called revenue
Because they are four different events at four different times, and only one of them is revenue.
A booking is a contract signed. Somebody has committed to pay you, has not paid you, and you have delivered nothing. A billing is the invoice, so you have asked for the money. Cash is the money arriving. Revenue is the money you earned by delivering something, and that word earned is the whole distinction. Revenue recognition, the standard that governs it, says you recognise revenue as you transfer what you promised. Not at signature, not at invoice, not when the money lands.
Run rate is a fifth thing again: your best recent month multiplied out to a year, which quietly assumes your best month was normal.
So when somebody says "we are at five million," the honest follow-up is five million of what. It is not a gotcha. All four numbers are legitimate and all four get reported, and they diverge most in exactly the situations where a founder most wants a big number.
ARR is where this gets specific enough to check. Annual recurring revenue is the revenue that recurs, which excludes one-time fees, setup and installation, and consulting and professional services. Andreessen Horowitz, in 16 Startup Metrics, names the two mistakes outright: counting non-recurring fees, and counting bookings. Both inflate the figure, and both are extremely common.
The question to ask: of that ARR number, how much is contracted and recurring, and what is it with services stripped out?
Where did the LTV to CAC ratio of 3:1 come from?
One blog post, in December 2009, by David Skok, on forEntrepreneurs. The sentence is: "It appears that LTV should be about three times CAC for a viable SaaS or other form of recurring revenue model."
Read the first two words again. It appears. No study, no dataset, no citation. An experienced investor writing down an impression, which the industry then turned into a law you will hear quoted in every accelerator and every board deck on earth.
Then the folklore did something worse than repeat it. It inverted it. In his follow-up piece, SaaS Metrics 2.0, Skok calls the original figures early guesses and writes that the best SaaS businesses have a ratio higher than three, sometimes as high as seven or eight. Three was his floor, the line for being viable at all. So the founder who hits exactly three to one and reports a healthy business is sitting at the bottom of a range its own author described as the minimum.
And the numerator is usually wrong before the ratio is even taken. Andreessen Horowitz spell out that LTV is the present value of the future net profit from a customer, and that the common error is calculating it on revenue, or on gross margin, instead. Revenue-based LTV can be three or four times the real figure.
Put both errors together and the most quoted rule in startup finance is an inflated number judged against a misread floor.
Where did the Rule of 40 come from?
A board meeting. Brad Feld published the Rule of 40 on 3 February 2015, and it is worth reading how he introduces it: he had heard something he had not heard before, from a late-stage investor at a board meeting, and that investor's firm called it the forty percent rule for a healthy software company.
The investor is not named. There is no study in the post, no dataset, no analysis of any kind.
Neither of these two rules is wrong. Growth plus margin adding to forty is a sensible instinct, and so is wanting a customer to be worth several times what they cost to acquire. The problem is the certainty they get quoted with. The two most repeated benchmarks in this entire subject are one sentence beginning "it appears" from 2009 and one unattributed remark from 2015, and tracing both took about ten minutes.
The habit worth building: when somebody quotes you a benchmark, ask where it came from. Most of the time the answer is a person rather than a study, and knowing that changes how hard you are willing to argue with it.
What is the option pool shuffle, and why does it cost 25% of a valuation?
Because of one word in a sentence you will actually see on a term sheet. Here is the clause, as it appears in a real one documented by Venture Hacks:
The $8 million pre-money valuation includes an option pool equal to 20% of the post-financing fully diluted capitalisation.
Includes. That word is the whole thing.
Do the naive arithmetic first. Eight million pre-money, six million existing shares, one dollar thirty-three a share. Now do it properly. The pool has to be created before the money goes in, so two million new option shares join the denominator. Eight million divided by eight million shares is one dollar a share.
Your effective pre-money valuation is not eight million. It is six.
Then notice who paid for it. The pool was carved out pre-money, which means it dilutes common stock only. The incoming investor is not diluted by it at all. You negotiated eight, you received six, and the two million funded the hiring the investment was supposed to fund.
The fix is not refusing a pool. You need one, and ten to twenty percent is normal. The fix is knowing that the size of the pool is a price negotiation exactly like the valuation is, and arguing about it with the same energy.
SAFE vs convertible note: what actually changed in 2018?
Y Combinator shipped the original pre-money SAFE in 2013 and replaced it with a post-money SAFE in 2018, and their stated reason is the most useful sentence a founder can read on this subject. They say the advantage of the post-money version is the ability to calculate immediately and precisely how much ownership of the company has been sold.
Sit with that. The reason they rebuilt the instrument is that with the old one, founders could not tell how much of their company they had given away. Pre-money SAFEs are still in circulation.
The other half of this is the valuation cap, and the sentence to remember is that a valuation cap is not a valuation. Nobody has said the company is worth that number. It is a ceiling on a future conversion, and founders quote it as a valuation constantly.
A convertible note does a similar job and is legally debt: it carries an interest rate and a maturity date, and if it does not convert, it is a debt the company owes. A SAFE has neither. And if you stack several of either across eighteen months, watch the most favoured nation clause, which quietly ratchets every earlier investor up to the best terms you ever gave anybody.
The question to ask: is this pre-money or post-money, and after it converts, what do I own fully diluted?
What does participating preferred cost on an exit?
Eight million dollars, in the standard worked example, and the founders and employees pay all of it.
Non-participating preferred means the investor chooses: take the liquidation preference, or convert to common and take their percentage, whichever is worth more. One or the other. Participating preferred means they do not choose. They take the preference off the top, and then they also share pro rata in whatever is left.
They get paid twice. Run it on ten million invested for twenty percent, on a fifty million dollar exit:
- Non-participating. Ten million preference, versus twenty percent of fifty million, which is also ten million. A wash. They take ten.
- Participating. Ten million off the top first, then twenty percent of the remaining forty, which is another eight. They take eighteen.
Every dollar of that eight million difference came out of the common stock. Now stack it with a two times preference multiple, where ten million invested takes twenty million off the top before anybody else is paid, and you have the fifty million dollar exit where the founders take home almost nothing. It is multiplication, and it is written into the document.
If you cannot negotiate participation away, negotiate a cap on it: the investor participates only until they have received a set multiple of their money, then stops, and the rest flows to common.
Why does 120% net revenue retention not mean what it sounds like?
Because it blends two opposite things on purpose. Gross revenue retention asks how much of the revenue you started the year with is still here, and it can never exceed 100%. Net revenue retention runs the same calculation but adds expansion revenue from customers who upgraded, which is why it can exceed 100%, and why every SaaS company in the world quotes it.
Andreessen Horowitz put the consequence plainly: net churn understates the losses, because it blends upsells with actual churn, while gross churn estimates the real loss to the business.
So a company can report 120% net revenue retention, which sounds superb, while losing a third of its customers, because a handful of large accounts expanded enough to paper over all of it. The same trick has a smaller cousin in logo churn versus revenue churn, which count customers and dollars respectively and are not interchangeable in either direction.
The question to ask: whenever net revenue retention is quoted on its own, ask for gross.
The waterfall: why the headline sale price tells you nothing
Because nobody gets a percentage on an exit. Everybody stands in a queue.
Debt is paid first. Then the preferred shareholders take their preferences in order, and the last money in is usually the first paid out. Then participation, if participation exists. Only what survives all of that reaches common stock, which is what founders and employees hold. Founders are at the bottom of the waterfall, always.
That is why two identical headline numbers can pay two founders wildly different amounts, and why the clauses in the middle of a term sheet matter more than the number on its front page. The valuation is the number you negotiate. The waterfall is the number that pays you, and they are not the same negotiation.
Two further reasons the price is not the price. An earnout makes part of your payment depend on hitting targets after the sale, inside a company you no longer control, on goals set by the people who now own it. Escrow holds a portion back for a period, to cover problems that surface later.
The 83(b) election: thirty days, no extension
This one is not a negotiation. It is a deadline, and it is the most expensive piece of paperwork in a startup.
If you hold shares that vest over time, you can be taxed as each portion vests, on its value at that moment. If the company grows, that value climbs, and you owe tax on paper gains for shares you cannot sell. The 83(b) election lets you choose to be taxed at the start instead, when the shares are worth almost nothing.
The window is thirty days from the transfer of the shares. That is in the tax code itself, and the code also says the election may not be revoked once made. Thirty days, no extension, no appeal.
Founders have lost life-changing sums to this, not because they made a bad decision but because nobody told them a decision existed. The neighbouring terms have the same shape. Incentive stock options can receive more favourable tax treatment only if you meet holding periods written into the code, and there is a hundred thousand dollar annual limit on how much can first become exercisable. A 409A valuation, which sets the strike price on employee options, has to be less than twelve months old to sit inside its safe harbour.
Nothing on this page is legal or tax advice, and this section is the clearest illustration of why. The point of knowing these words is not to handle them yourself. It is to know which questions to take to a real lawyer and a real accountant on the day you are granted shares, instead of finding out afterwards.
The three questions that cover most of it
You do not need seventy-nine definitions to stop being the least informed person in the room. You need three habits.
- Ask which number. Five million of bookings, billings, run rate or recognised revenue are four different companies. The same applies to GMV, where the honest follow-up is the take rate, because a hundred million in GMV at a ten percent take rate is a ten million dollar business.
- Ask where the benchmark came from. Two of the most quoted in the field are blog posts. Several are excellent instincts. None of them is an accounting standard, and the ten minutes it takes to check is the cheapest research you will ever do.
- Ask what it does at the exit. Preference, participation, multiple, anti-dilution and the option pool are one machine rather than five separate clauses, and the machine only runs once. Model the waterfall on a realistic sale price before you sign, not on the optimistic one.
The full seventy-nine terms, chapter by chapter with the sources, are at the finance terms sheet. The neighbouring glossaries, built the same way, are marketing terms explained through one business, the YouTube terms that decide if you get clients, and every AI term explained with its primary source.
If you are earlier than any of this and the live question is what to build rather than how to fund it, we have pieces on building a one-person company and making money vibe coding.
