When to stop a project

Training 1025 minutessunk costs

You grow a small service and test a separate product. $1,200 is already spent on the test and cannot be recovered. $1,800 is left in the account, and the current service runs on its own.

Training goal: Choose by future results and keep the option to stop

▶ Play this training

Situation

1. Finishing the product costs another $600, while confirmed future income is only $450. There is no other income, no benefit to the current service and no other cost; stopping costs nothing. Which option leaves more money from today on?

Finish it to win back at least $450
Add another $1,200 to the budget and relaunch
Stop the project

2. For a new decision, future income is unknown: with 50% probability it is $1,000, otherwise $300; finishing costs $600. A $50 check shows the exact scenario before you pay to finish, and you have enough money for any option. Which option gives the best expected money result from now?

Check first and finish only at $1,000
Finish right away without a check
Skip both the check and the finish

3. Next month a measurable test costs $200, and one new paid sale brings $25 of margin. There is no other benefit or new cost; judge the result by this month only. What minimum number of sales covers the test?

6 sales are enough
At least 8 sales are needed
At least 10 sales are needed

4. The test is over: it cost $200 and brought 6 sales at $25 margin. Another identical test costs $200; there is no new hypothesis, fix or sign of growth, and next period's goal is still money. What now?

Repeat the same test to win the loss back
Double the budget at the same return
Pause and find a new hypothesis

What you will learn

Money already spent and gone does not change which future option is better.
A check is valuable when its result can change the decision.
Set the stop rule before the test and keep it apart from a safety margin.

Your real-life step

For one current project, write down separately: past sunk costs, future payments, confirmed income and the stop condition.

Questions and answers

Should you ignore past spending completely?

When choosing a future action, a sunk amount is not added to just one option. But experience, equipment you bought and information gained can still affect future results.

How much can you pay to test a hypothesis?

No more than the expected gain from changing the decision, given the quality of the information and the time. In step 2 accurate information before paying is worth up to $150 compared with launching blind; the check costs $50.

Does a positive expectation mean you should continue?

No. Possible damage, available cash and other options matter too; in step 2 a sufficient cash cushion is stated on purpose.

▶ Play this training Track your budget in Hanimani

More trainings in this module

Numbers are illustrative. A game, not financial advice.