vizmaya desk · 8 October 2026

Founders blame the money. The research blames something else.

Money you have already spent is gone. Yet it still pulls on your next decision. We start from the very beginning: why that happens, why founders are caught by it, and what to do about it.

1. The idea

Money you have already spent is gone, so only the future should count

Say you paid $100 for a gym plan you now hate. That $100 is gone whether you keep going or quit.

So the only fair question is: will next month be worth it? Economists call money like this a sunk cost, and say it should not change your choice.

Already gone

$100

Nothing you do now changes it.

Still open

Next month

This is the only part you can still decide.

Source: Arkes and Blumer, 1985

1. The idea

People let the past decide, even when it makes little sense

Evidence A

Researchers told people a plane project had already cost $10 million. 41 said finish it. 7 said stop.

Then they asked whether to start the same project fresh with $1 million. Only 10 said yes. 50 said no.

The difference was the money already spent.

Source: Arkes and Blumer, 1985

1. The idea

The pull is real, but it fades with time

Evidence A

Ohio University sold theater season tickets at random prices: $15, $13 or $8. In the first half of the season, people who paid full price used 4.11 tickets on average. The discounted groups used about 3.3.

Paying more made people turn up. By the second half, the gap was gone.

Source: Arkes and Blumer, 1985

2. The surprise

The money already spent is the weakest pull on the list

Evidence A

Researchers pooled 166 studies of why people keep going when they should stop. A score of 0 means no link. A score of 1 means a perfect link.

Money already spent scored .24, the lowest. Pride at stake scored .47, nearly double.

Scores are approximate. Source: Sleesman et al., 2012

2. The surprise

Three pulls beat money, and none of them is about money

Pride .47

Stopping feels like admitting you were wrong. Researchers think this is the biggest pull of all.

Time put in .43

The more of your life you have given to something, the harder it is to walk away.

Feeling close to done .39

The nearer the finish feels, the harder it is to stop.

Source: Sleesman et al., 2012

3. Founders

A founder feels all three pulls at once

PrideThe company is part of who you are

Quitting can feel like admitting you failed as a person, not just as a business.

Time put inYears go in before you know

You give a lot of your life before the evidence is clear.

Feeling close to doneOne more feature, one more month

The finish line always looks near.

Owning the decisionYou made the first call

Earlier research found that people who made the first choice themselves commit the most to it.

This is our reading. The research covered students and professionals, and we found no study of founders alone.

Sources: Sleesman et al., 2012, Staw, 1976 (overview)

3. Founders

66

teams of three ran a startup in a lab game

Close friends are quicker to keep funding a failing company

Over 569 decisions, teams who were closer friends were more likely to keep investing in a startup that was going badly.

Hope pushed them on more than fear pushed them to stop. A co-founder who is a close friend is both a strength and a risk.

Evidence A. Source: Huang, Souitaris and Barsade, 2019

3. Founders

Running out of cash is usually the last cause, not the first

Evidence C

CB Insights studied 385 venture-backed startups that shut down. 70% ran out of money. The report calls that the final blow, not the root cause.

Many had also built something too few people wanted. Companies gave several reasons, so the bars add up to more than 100%.

We could read four of the nine reasons. Source: CB Insights

4. The catch

20 to 30%

more likely to finish after a costly start

Staying is not always a mistake: past spending can be a clue

Imagine you have forgotten why a project began. The size of the bet hints that someone once saw real promise in it.

In a business-school test, people who could not recall the original reasons were 20 to 30% more likely to finish after a costly start. The fix is simple: write down why you started.

Evidence A. Source: Baliga and Ely, "Mnemonomics"

4. The catch

Dyson built 5,126 prototypes. That was not the trap.

Each prototype changed one thing, so each one taught him something. The trap is repeating the same bet and hoping.

Per London Business School, it took 14 years from his first experiments to the 1993 DC01. His fortunes turned when he took control of the whole process, so he changed his plan too.

The test

Is each new try teaching you something, or are you repeating and hoping?

Evidence C. We have no evidence of his own decision rules, so this is a contrast, not proof. Source: London Business School

5. Who got out

Intel left memory chips by asking one question

In 1985 Intel had become a small player in memory chips, a business it helped create. Andy Grove asked Gordon Moore what a new boss would do if the board replaced them. Moore said: leave memory.

So they did. They kept the chips where Intel led, and kept spending on research. Quitting was a choice about where to fight.

The new-owner question

What would someone with no history here do today?

Evidence C, the usual retelling. Layoffs: about 7,000 (NPR) or 7,200 jobs and eight plants (1986 letter).

5. Who got out

Buffett kept a failing mill too long, and says so

Berkshire Hathaway began as a textile company. Each new investment in the mills looked sensible on its own. Added up, they earned too little.

He decided to close it in July 1985, after 21 years, and wrote that he should have quit sooner.

~$13Mwhat the machinery cost new

$866Kits value on the books

$163Kwhat it sold for in 1986

Evidence B. After costs, the 1986 sale lost money. Source: Buffett, 1985 shareholder letter

5. Who got out

Slack and Instagram quit their first idea and kept the useful part

Slack began as Glitch, a game that never found an audience. The team kept the chat tool they had built for themselves.

Instagram began as Burbn, a check-in app with too many features. Kevin Systrom kept the photo sharing, the one thing people really used.

Facebook later bought Instagram for $1 billion, when it had 13 employees.

Evidence C and D. Source: Inc.

6. What to do

Decide on the future, and set the rules in advance

Ask the new-owner question

What would someone with no history here do today?Grove and Moore

Write your stop signs now

Decide today which future results mean stop, and name one person allowed to tell you.Annie Duke

Write down why you started

So the past is not the only record you have.Baliga and Ely

Book a check-up

About 3 months into an idea, then about once a year for a live product.Dalton Caldwell, via Lenny's Newsletter

Count what else you could do

It helps, but it can backfire if you feel personally responsible or nearly done.Sleesman et al.

Reward early stops

At X, a research lab, teams get recognition for ending projects early.Astro Teller

Sources: Duke summary, Lenny's Newsletter, Stanford

Sources & method

How sure are we?

Strongest: the plane and theater experiments and the 166-study review (both read in full), Buffett's own letter, and the Baliga and Ely paper. Middle: news and company reports. Weakest: Wikipedia, newsletters and book summaries. We found none of these leaders using the words "sunk cost", so the lessons are our reading. We found no study of founders alone, and no evidence of Dyson's own decision rules.