Blog
When Not to Use a Random Picker
Seven situations where reaching for a draw is the wrong instinct, from someone who builds them.
Seven situations where reaching for a draw is the wrong instinct, written by someone who builds them.
Two of these are cases where a random pick hides a decision somebody should be making, and one is a case where it is quietly unlawful.
One: when the options are not equivalent
A draw treats every option as interchangeable. If they are not — if one is meaningfully better, cheaper, safer or more likely to work — then randomising is a way of avoiding the comparison rather than resolving it.
This is the commonest misuse and it is usually a symptom of not wanting to argue. The draw feels neutral and it is actually a decision to ignore the information you already have.
Two: when somebody is accountable for the outcome
If a decision goes wrong and someone will be answerable for it, a random pick does not transfer the accountability — it only removes the reasoning. "We drew it" is not a defence for a choice that was yours to make.
Worse, it destroys the record. A decision made for reasons can be explained and revisited; a decision made by draw cannot, because there is nothing to explain.
Three: where consistency is the point
Structured interviews, standardised assessments and any process whose value comes from treating everyone identically are made worse by randomisation, not better. Varying the questions between candidates breaks the comparison the process exists to support.
The randomness belongs before the process — choosing which questions to use, once, for everyone — rather than during it.
Four: allocation with a legal or regulatory dimension
This is the case where a draw can be genuinely unlawful, and it is general information rather than advice. Combining payment, chance and a prize meets the definition of a lottery in many jurisdictions, and running one without authorisation is a regulatory matter rather than a design choice.
The same care applies to allocating anything where anti-discrimination obligations attach. A draw is neutral and the frame it draws from may not be, and a randomised process over a biased list produces a biased outcome with a fairness story attached.
Five: when the group will not accept it
A draw only works if the group agreed to be bound by it. Running one in a group that has not agreed produces an outcome that gets re-litigated immediately, and the draw has cost time without settling anything.
Agreement has to precede the draw. Asking afterwards whether people will accept the result is asking after they know what it is.
Six: safety-adjacent choices
Anything where an outcome could cause harm should not be randomised, however trivial the framing. A dare generator, a workout selector, a challenge picker — the tool has no idea who is using it or what they can do.
Every such tool on this site is framed as a suggestion that anyone may decline, and that framing is the mitigation rather than a disclaimer.
Seven: when it is a coin flip about something that matters
For a genuinely balanced decision with real stakes, the useful thing about a coin flip is not the outcome — it is watching your own reaction to it. Disappointment at the result is information you did not have a minute earlier.
That is a legitimate use of a draw and it is not the use most people think they are making. If you would have been relieved either way, take the answer. If you find yourself wanting to re-flip, you have learned something and should stop flipping.
The meta-failure: randomising the wrong step
Beyond the seven, there is a recurring pattern worth naming separately, because it looks like good practice and is not. It is randomising a step that does not matter while leaving the step that does entirely to judgement.
A hiring process that randomises interview order and leaves shortlisting to a manager's impression has randomised the cheap part. A giveaway that draws the winner impeccably from a list somebody curated by hand has done the same. In both cases the randomisation is real and it is decorative, because the decision was made upstream.
The diagnostic question is where the actual discretion sits. If you can point to a step where a human made a consequential choice with no constraint on it, that is the step to address — and randomising a later step is at best neutral and at worst a fairness story attached to a process that does not have one.
Frequently asked questions
When is a random pick the wrong tool?
When the options are not equivalent, when somebody is accountable for the outcome, when consistency is the point, or when the group has not agreed to be bound.
Why does accountability matter?
Because a draw removes the reasoning without transferring the responsibility. 'We drew it' is not a defence for a choice that was yours.
Can a draw be unlawful?
Combining payment, chance and a prize meets the lottery definition in many jurisdictions. This is general information, not legal advice.
What about a biased list?
A neutral draw over a biased frame produces a biased outcome with a fairness story attached. The frame is where to look.
Should agreement come before or after the draw?
Before, always. Asking afterwards is asking after people know what the result is.
Is a coin flip useful for a serious decision?
For noticing your own reaction to the outcome, yes. If you want to re-flip, you have learned something and should stop flipping.