Fast Error Exit Rule
Limit downside by detecting failed bets and exiting them quickly
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 98%
The Fast Error Exit Rule treats experimentation as necessary while controlling the damage caused by unsuccessful bets. Before committing substantial resources, define what progress should look like, how much capital or time may be lost, and when the evidence will be reviewed. If the initiative repeatedly misses those indicators, stop it, shrink it, or redesign it instead of defending the original conviction. The mechanism separates the frequency of mistakes from their magnitude: a company can survive many inexpensive errors but may not survive one error allowed to grow unchecked. Learning remains valuable only when the organization captures the lesson and prevents the same failure pattern from recurring.
Origin
Joe Ricketts developed the rule through TD Ameritrade's repeated technology experiments and reinforced it after investing roughly $100 million in the unsuccessful On Money financial-services platform.
Core principles
- 01Mistakes are unavoidable when building something new
- 02The magnitude of an error matters more than the error count
- 03A failed assumption should trigger action, not defensiveness
- 04Small experiments preserve capital for better opportunities
How to run it
- 1
Define the hypothesis
State what the initiative must prove about customer demand, technical feasibility, or economics. Convert confidence into a testable claim.
Pro tip Use an observable behavior, such as paid adoption, rather than favorable opinions.
Watch out Do not treat enthusiasm from interviews as proof of demand.
- 2
Cap the exposure
Set explicit limits for money, time, and organizational attention before starting. Choose a limit the business can afford to lose.
Pro tip Release funding in stages rather than approving the full budget immediately.
Watch out Sunk costs can make a predetermined limit emotionally difficult to enforce.
- 3
Test at small scale
Build the smallest credible version that can expose the core assumption. Collect evidence from real operating conditions.
Pro tip Test distribution and customer behavior as well as product functionality.
Watch out A technically successful prototype can still lack a viable market.
- 4
Review disconfirming evidence
Compare actual results with the predetermined indicators. Look specifically for evidence that the original thesis is wrong.
Pro tip Invite someone without ownership of the project to challenge the interpretation.
Watch out Do not move the success criteria merely to preserve the initiative.
- 5
Exit or correct quickly
Stop the initiative when the core assumption fails, or make a bounded correction and retest. Record the lesson so the organization does not repeat the same error.
Pro tip Preserve reusable technology or insight without preserving the failed project itself.
Watch out Continuing solely because of prior investment increases the magnitude of the mistake.
In the wild
Ricketts built On Money to combine brokerage, banking, and insurance information in one software package. Although prospective users said they wanted it, providers resisted sharing information and independent customers needed financial-planner support. He ultimately spent about $100 million but concluded that the evidence should have stopped the project around $20 million or $30 million.
→ The failed investment produced a durable rule to pull the plug once evidence shows that an initiative is not working.
TD Ameritrade repeatedly built software that failed to meet operational needs, discarded it, and tried again. By eliminating unusable implementations rather than protecting them, the company eventually created an end-to-end system that could execute and report trades within seconds.
→ Fast rejection of inadequate implementations helped produce a scalable online-trading platform.
Common mistakes
Scaling before validating
A large commitment made before the core assumptions are proven turns a manageable experiment into a major loss.
Confusing stated interest with adoption
People may praise an idea without changing their behavior or paying for it.
Defending sunk costs
Past investment cannot improve the future economics of an initiative that is demonstrably failing.
Is it for you?
Best for
It is best for founders and managers making uncertain product, technology, or market investments.
Not ideal for
It is not ideal for initiatives whose value inherently requires a long, evidence-free incubation period.
From the transcript
“It's the magnitude of the mistakes, not necessarily the total number.”
“Once you know that you've got something that's not working, pull the plug”
“It's good to make a mistake because you learn from that mistake and you won't repeat it.”
From the episode
Joe Ricketts Discusses Trade and Deregulation
Joe Ricketts