Sunk cost fallacy in business decisions: when to stop
The essentials in 30 seconds
- The sunk cost fallacy in business decisions is the habit of letting money already spent decide whether to continue.
- A go or stop decision uses forward costs and forward benefits only.
- In the illustrative case, continuing nets minus 100 while stopping and switching nets plus 450 (US$ thousands).
- Staged funding in 4 tranches of 150 gives natural review points.
- The EQOS Asia Audit pillar measures what exists on 9 structural dimensions (EQOS Asia method page, as of October 2026).
The sunk cost fallacy lets money already spent steer the next choice. This page focuses on the moment of stopping: which numbers a go or stop decision should use, how the test works on an illustrative platform project, and which routines make stopping a normal step. It sits within the guide to cognitive biases, and all figures are illustrative as of October 2026.
On this page
Compare the EQOS Asia offers and request a 30-minute pre-diagnostic.
What is the sunk cost fallacy, and how did Concorde give it a second name?
The sunk cost fallacy is the habit of letting past spending steer a future choice. A sunk cost is money, time or effort that no option can recover. The Concorde supersonic jet gave the pattern a second name: according to Wikipedia, the British and French governments kept funding it after its economic case had weakened.
The pattern grows with visible investment. Large budgets, public announcements and personal ownership all raise the pull. A team can feel the pull as loyalty to earlier work. Naming the feeling as a bias lets the team separate respect for the effort from the choice about the next dollar.
Which numbers should a go or stop decision use?
A go or stop decision uses 3 forward numbers: the cost still to be spent, the benefit still to be earned and the value that stopping would release, such as equipment sold or staff redeployed. History enters only as evidence about how reliable the estimates are. The total already spent appears in the file for the record and sits outside the comparison.
Write the comparison as a table with one column per option: continue, stop, or switch to the best alternative. Each column holds the forward cost, forward benefit and net value. The table takes 10 minutes to build and gives the committee a single page on which the sunk amount has no row.
What does a go or stop decision look like for a billing platform?
Take an illustrative billing platform for a Hong Kong subsidiary, with 1,200 already spent (US$ thousands). Finishing needs another 600 and is expected to earn 500. The net of continuing is minus 100. The platform is 80 percent built, and that percentage tempts the team to finish, yet the remaining 20 percent costs more than it returns.
Stopping releases 150 from an equipment sale and redeployed staff. A smaller project B costs 400 and is expected to earn 700, a net of 300. Stopping and switching nets 150 plus 300, or 450, against minus 100 for continuing. The gap of 550 is the cost of honouring the past, and it is visible on one page.
Which signals show sunk cost thinking is escalating a project?
Escalation of commitment shows in language and in numbers. Phrases such as we have come too far or we are almost there appear in meetings. The estimate of remaining cost rises review after review: in the example it moves from 400 to 500 to 600 over 3 reviews while the finish date slides. A rising remaining cost is a measurable signal.
Set a trigger in advance. If the remaining cost estimate rises at 2 consecutive reviews, an independent review starts within 2 weeks. The trigger is a number, so no individual has to raise the alarm. It arrives earlier than a loud debate would, while more options remain open.
How do stage gates limit sunk cost on a project?
Stage gates release money in tranches and turn one big commitment into several small choices. The remaining 600 splits into 4 tranches of 150. Each tranche requires a fresh estimate of cost to complete, a refreshed benefit and a written go, stop or switch verdict. Every gate is a natural moment to stop with dignity.
Keep the gate criteria short: forward net value above zero, remaining cost within 10 percent of the last estimate, and the best alternative still ranked second. A project that clears 4 gates has earned its budget on forward numbers each time. A project that fails a gate leaves a record from which the next project learns.
When should a fresh owner review a project hit by sunk cost?
People who started a project find it hardest to judge it. Ask a fresh reviewer from outside the team the inheritance question: if this project landed on your desk today with no history, would you fund the remaining 600? The reviewer reads only the forward table and gives a verdict in writing.
Protect the original sponsor in the process. Record that the original decision rested on the information of its day and that stopping is a result of new information. Pairing the stop with a lessons note and a redeployment plan for the team keeps morale high and makes the next project easier to start.
How does the EQOS Asia decision engine support a stop decision?
The method page describes the Audit pillar as a measure of what actually exists in a system, computed on 9 structural dimensions by a deterministic core (EQOS Asia method page, as of October 2026). The same inputs return the same score, so a stop verdict can be replayed and explained to a board.
The prediction pillar projects 3 scenarios, optimistic, median and adverse, and identifies tipping thresholds (EQOS Asia method page, as of October 2026), which suits a review that asks whether to continue. A reader can try a first score in the simulator, and the 30-minute pre-diagnostic, scheduled in Singapore time (SGT), tests a real project.
See also in this guide
Frequently asked questions
What is the sunk cost fallacy?
The sunk cost fallacy is the habit of letting money, time or effort already spent decide whether to continue. Past spending cannot be recovered under any option, so it carries no weight in the comparison. A go or stop decision uses the cost still to be spent, the benefit still to be earned and the value that stopping releases.
How do I know when to stop a project?
Stop when the forward net value of continuing falls below the forward net value of the best alternative. In the illustrative billing platform, continuing nets minus 100 and stopping with a switch nets 450. A trigger such as a remaining cost estimate that rises at 2 consecutive reviews brings the question to the table early.
Why is stopping hard for the team that started the project?
Ownership, visible budgets and public announcements raise the pull of past spending, and stopping can feel like a verdict on the people. A fresh reviewer, a written inheritance question and a lessons note turn stopping into a normal, respected step in the project cycle.
How can the EQOS Asia pre-diagnostic help with a stop decision?
A free 30-minute pre-diagnostic, scheduled in Singapore time (SGT) and held under NDA, applies the EQOS Asia equation to a real project (EQOS Asia contact page, as of October 2026). It shows where the structure of the decision is strong and where it carries risk. The simulator offers a first score for comparison.
Test your knowledge
1. What does the Concorde example illustrate?
- Continuing to fund a project after its economic case weakened
- A rapid product launch
- A merger of two airlines
Show the answer
Continuing to fund a project after its economic case weakened.
2. What is the forward net of continuing in the worked example?
- Plus 450
- Minus 100
- Plus 300
Show the answer
Minus 100.
3. What is the net of stopping and switching to project B?
- 300
- 150
- 450
Show the answer
450.
4. How many tranches does the remaining 600 split into?
- 4
- 3
- 6
Show the answer
4.
5. When does the independent review start in the example?
- After a loss
- Within 2 weeks of a trigger
- After the final tranche
Show the answer
Within 2 weeks of a trigger.
About the author
Stéphane Jambu
President, EQOS Forecast Asia
Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile
Sources: EQOS Asia method page (eqos-asia.com/method.html), simulator page (eqos-asia.com/simulator.html), and contact page (eqos-asia.com/contact.html), and en.wikipedia.org/wiki/Sunk_cost, consulted on 22 October 2026. Page updated on 22 October 2026.