No spreadsheet ever had a happy ending

Somewhere right now, a good story is being killed by a man with a laptop. He isn’t a bad man. He owns a vest. He is about to open a spreadsheet, and the whole room is about to relax, because a spreadsheet is where arguments go to die and nobody has to feel responsible for the body.

I have watched this happen more times than I can — no. I could count. That’s the entire problem. I could put a number on it, and the second I did, you would stop listening to the story and start auditing the number, which is the exact crime I came here to complain about. Give me a minute. I’m working up to it.

A number is a story that stopped taking questions

A story is an honest animal. It has a shape, it has a shrug built into it, and you’re allowed to argue with it. You can tell me you don’t believe the part where the kids build a ship out of library books, and we can talk about it like people. A story knows it’s a story. It shows up in a t-shirt.

A number shows up in a suit. It doesn’t want to talk. It says: the adults have looked into this, the discussion is closed, here is 67 percent. And 67 percent has a crisp, decimal-pointed confidence that no story would ever have the nerve to fake, because a story would be embarrassed. That confidence reads as rigor. It isn’t rigor. It’s tailoring.

Here’s the move the suit is running. A story admits it might be wrong — that’s the shrug. A number hides the shrug in a footnote you’ll never open, then stands there radiating a certainty it didn’t earn. You were handed a guess in a lab coat and told to call it evidence. And you did, because you’re polite, and because the man who handed it to you seemed so sure.

The best thing I ever built failed the spreadsheet

Let me take the stand for a second, because I’ve been on both sides of this.

The first thing I built that mattered was a podcast client, back when the honest market size for podcast software was zero — not “small,” zero, because podcasts didn’t meaningfully exist yet, and a spreadsheet can’t model a verb that hasn’t been invented. Every number said no. There were no numbers. That was the number. I built it anyway, it became the first one of its kind, and years later there’s an award on a shelf for a different thing I made mostly to prove a point. The point was this one: the spreadsheet is a rearview mirror that keeps insisting it’s a windshield.

If I’d respected the numbers I was handed, I’d have respected my way into never doing anything. The numbers weren’t even wrong. They were just describing, in loving detail, the exact world the whole idea existed to leave.

No spreadsheet ever had a happy ending

Open any spreadsheet built to convince you of something and scroll right. Keep going. Out past the columns you were shown in the meeting, there’s always a cell where optimism quietly breaks the law. A growth rate that only works if nobody sleeps. A “conservative” assumption doing the cardio of an Olympic sprinter. The hockey stick — that beautiful curve that has never once bent that way in nature and bends that way in every deck.

Nobody reopens that tab. That’s the deal you make with a projection: you agree not to look at it again, and it agrees to make you feel responsible for having a plan. A spreadsheet isn’t a prediction. It’s a mood, formatted.

And moods don’t end well, which is the actual point, so let me say it flat: no spreadsheet ever had a happy ending. It can’t. A happy ending requires something to have actually happened, and a spreadsheet is a document specifically about things that haven’t happened yet. The only spreadsheets that end happy are the ones you fill in afterward, backward, so the story you already survived looks like a plan you definitely had. I’ve made those. They’re the only honest ones, and they’re honest entirely by accident.

The caviar is load-bearing

Now the part I’m actually mad about.

Have you noticed that the people who traffic in these numbers are, on average, doing great? The man in the vest is not eating what you’re eating. He’s at a dinner right now where the fish eggs cost more than your car payment, explaining to another vest that the market is “structurally attractive,” using numbers he will use tomorrow to explain the precise opposite, and everyone at the table will nod, because the nodding is the meal.

Meanwhile the rest of us — the schlubs, and I say that with love, I’m the mayor of us — are home eating frozen pizza, if we’re lucky, reading a guide that promises the one number that’ll change everything. It won’t. It was never for you. The number is the velvet rope. It exists to sort the room into people who get to sound certain and people who have to go measure something first.

That’s the con, and it’s an elegant one: confidence is expensive to fake and cheap to inherit. The smug guy doesn’t generate the certainty. He borrows it — from a research firm, from a founder who got lucky once, from a chart with a logo on it — and the borrowing is invisible, so it looks like he simply knows things. He doesn’t know things. He knows people who know how to say things. There’s caviar in this for exactly one of you, and it isn’t the one holding the pizza box like a steering wheel.

The number is how they keep the door

Here’s the thing about the establishment that took me too long to see: it doesn’t run on being right. It runs on being the one who gets to sound sure. The number is the door, they hold the key, and the key is a decimal point.

Ask who actually benefits when “the data says” ends a conversation. Not you — you have the idea and the pizza. It’s whoever owns the data, or rents it, or can afford to stand next to it in a good vest: the analyst, the consultant, the fund that also happens to own the report you’re being handed. They didn’t build the thing. They built the scoreboard, appointed themselves referee, and now every founder in the country is asking permission from a man who has never once been on the field.

An unsourced number isn’t a bug in that system. It’s the product. It keeps the outsiders measuring while the insiders eat. The establishment doesn’t fear your terrible idea — it has survived a thousand of those. It fears the morning you stop needing its numbers to believe your own eyes.

Why we keep believing the guy in the vest

I want to be fair to us for a second, because we’re not stupid. We’re just built wrong for this particular room.

We’re wired to read the familiar as safe and the confident as competent. It kept our ancestors from eating the interesting berries and following the nervous man into the tall grass. Excellent instincts for berries and grass. Ruinous instincts in a conference room, where the confident man is confident precisely because he’s never once had to be right, and the nervous one has a business to lose.

So a number lands as safety. “67 percent, per a name you recognize” switches off the one question that matters: says who, and did they actually say it? I’ve started checking. It’s the most depressing hobby I’ve ever taken up. You pull the thread on the impressive statistic and there’s nothing on the other end — a firm that published the opposite, a sample size that quietly changes shape between two paragraphs, a citation that leads to another citation that leads to a man in a vest. The confidence was the whole product. There was never anything underneath.

And here’s what I can’t stop noticing: the people who genuinely know the terrain hedge. They say it depends. They tell you the honest answer varies by who and where. The certainty runs inversely proportional to the homework. Fake data and false confidence travel together, always — a guy and the friend who vouches for him.

What a story does that a number can’t

A good story is the opposite deal, and it’s a better one.

A story tells you it’s a guess. It has a narrator, and the narrator can be wrong out loud, in front of you, which is the single most trustworthy thing a piece of information can do. A number can’t apologize. A story can. A story also lets you in — you can test it against your own life, argue with the middle, keep the parts that fit and bin the rest. Nobody ever kept the useful 40 percent of a statistic. You take the number whole or not at all, and that’s a tell, because that’s how you sell things that can’t survive being examined in pieces.

And a story is yours. You can’t own a borrowed number; it belongs to the firm on the chart. But the story of what actually happened in your market, to your people, on your specific terrible Tuesday — that’s the one asset the vest can’t expense. It’s cheaper than caviar, and it has the enormous advantage of being true.

I’m not telling you to burn the spreadsheets

Because that would be a story with no shrug in it, and I’d be doing the exact thing I’ve spent a thousand words yelling about.

Numbers are fine. Numbers are good. A number you derived yourself, from your own market, that you can defend at 2 a.m. to a hostile stranger — that number is worth more than every deck in this city. The problem was never math. The problem is borrowed math worn as a costume, certainty rented by the hour, the specific 67 percent that came from nobody and was built to end the conversation before you noticed it hadn’t started.

So do the annoying thing. Ask who said it. Follow the number home. If it turns out to live nowhere, you’ve just saved yourself a quarter and a small piece of your soul. Mark your own guesses as guesses, out loud, like an adult. And when a man in a vest hands you a percentage with a decimal point and a steady face, thank him, and go measure the thing yourself.

Dreams don’t have numbers

That isn’t a weakness. It’s the definition. A number measures something that already exists, and a dream is the part that doesn’t exist yet — the one square of the map a spreadsheet physically cannot reach. You can’t run a regression on the moon before anyone’s stood on it. There’s no data up there. The absence of data is the entire point.

Dreams crossed oceans. Dreams built the pyramids. Dreams put twelve people on the moon while the accountants back at mission control chewed their pencils to splinters. Every one of them got the same memo first — the numbers don’t support this — because of course they didn’t. A number is a status report on the world as it is, handed to the one person in the room trying to make it something else.

So dream anyway. And when the accountant does his Soup Nazi routine, slides the ledger back across the desk, “no numbers for you” — let him keep the soup. You were never going to find the moon in a spreadsheet. Nobody worth a footnote ever asked the man with the calculator for permission to want something.

The happy ending

It was never in the spreadsheet. I checked. I scrolled all the way right and found a single cell that said, in effect, “and then a miracle occurs, at 40 percent month over month.”

The happy ending is somewhere dumber and better. It’s in the thing you built because you couldn’t stop thinking about it, back when every reasonable person with a good number told you not to. It’s the pizza, eaten without shame, on the night something you made finally worked. The caviar people can keep the caviar. I’ve had it. It’s fine. It tastes like the ocean filed a complaint.

So go build the terrible idea. Let the reasonable people keep their reasonable spreadsheets and their reasonable track record. And if you ever want the long version — how to derive the numbers that are actually yours, the ones with the shrug left in — I wrote a short book about it. It costs less than a bad lunch, and it admits, on the last page, the one number I got wrong. Which is one more admission than the vest will ever make you.

Author’s note: After writing this post, I tripled down on my commitment to real facts and found even more mistakes. Read about the errors.