Fair Ways to Pick a Giveaway Winner
The generator is not the problem
In practice, nobody argues about the random number. They argue about everything surrounding it: was my entry in the list, did that person really get five entries, did you draw twice because you did not like the first name, why did the winner turn out to be your cousin.
All of which means a fair draw is mostly a documentation exercise. The tool takes half a second; the credibility comes from what you did before and after.
Three commitments cover almost every objection.
Freeze the list. Close entries at a stated time, export the list, and publish it β a pasted comment thread, a spreadsheet link, a screenshot of the full set of names. Once the list is public and timestamped, nobody can claim they were left out, and you cannot be accused of adding anyone.
Draw once. One visible action, from a fixed list, with the result accepted. Rerunning is the single most damaging thing you can do, and the fact that nobody saw the earlier attempts is exactly why it damages you.
Show the mechanism. Say in one sentence how the winner was chosen. It costs nothing and it pre-empts the entire genre of complaint.
Wheel or numbered draw?
Both are equally random. They differ in what an audience can check.
| Entrants | Method | Why |
|---|---|---|
| Under 20 | Wheel spinner | Every name is a visible slice; nothing to cross-reference |
| 20β40 | Random name picker | Still readable, and remove-winner draws the runners-up |
| 40β200 | Numbered draw | Slices become unreadable; publish the numbered list instead |
| 200+ | Random number generator | Scales without limit, easy to film in one shot |
| Multiple prizes | Either, with no repeats | Remove-winner on the wheel, or unique on the number draw |
The wheel wins on persuasion at small sizes because the audience can literally count the slices. The numbered draw wins on scale, but it moves the burden of proof onto the list: viewers see the number 287 come up and have to trust your mapping of 287 to a person, so that mapping must be published before the draw, not after.
For multiple prizes, decide in advance whether one person can win twice. On a wheel, that is the remove-winner tick box; on a number draw, it is the unique option.
Recording it properly
A screen recording is the cheapest insurance available. What makes one convincing is that it happens in a single unbroken take:
- Show the frozen entry list, scrolled top to bottom.
- Show the list being pasted into the tool, or the range being set to the entry count.
- Draw once.
- Keep recording while you read out the result.
That sequence answers every reasonable question at once. A live stream is stronger still, because it cannot be re-taken β which is why larger promotions use them. A screenshot of the winning name is the weakest option, since it proves only that the tool can display a name.
Note that this site stores nothing: no entry list, no draw, no winner. Your recording is the record. The one thing that does persist is the page link, which carries your entries in the address bar β so the same URL you drew from can be shared afterwards and will show the same wheel.
Duplicates, bots and the awkward cases
Duplicate entries are where most informal giveaways come unstuck, because the rule is usually never stated. Settle it in advance and say it out loud:
- One entry per person. Deduplicate the list before you draw, and say you did.
- Extra entries are earned. Leave duplicates in, because each is a legitimate slice. Publish the count next to each name so the weighting is visible β this is exactly how a raffle with multiple tickets works.
Obvious throwaway accounts are worth removing, but remove them before the draw and before you know who wins. Filtering after the result is indistinguishable from choosing the winner, whatever your intentions were.
Finally, publish a claim window and draw your reserves in the same recorded action. A winner who never replies is common, and having the second and third names already drawn and public turns an awkward moment into an administrative one.
For the mechanics of building the wheel itself, see how to make a custom spinner wheel. For what makes the underlying draw trustworthy, how random is random has the detail.
Frequently asked questions
Wheel or random number β which is fairer?
Neither is fairer; they are equally random and differ only in what an audience can verify. A wheel shows every entrant as a visible slice, which is convincing up to about forty names. A numbered draw scales to any size but requires you to publish the numbered list separately, since viewers cannot see who number 287 is. Pick by list length.
How do I handle duplicate entries?
Decide the rule before you collect entries and state it publicly. If one person can only win once, deduplicate the list before drawing. If extra entries were genuinely earned β for a share, a referral, a second ticket β leave the duplicates in, because each one is a legitimate extra slice. What causes rows is deduplicating quietly after seeing who won.
Should I record the draw?
Yes, for anything with a prize worth arguing over. A screen recording that shows the entry list, the tool and the result in one unbroken take answers almost every question anyone can raise. Live-streaming is stronger still, since it cannot be re-taken. Both are far more persuasive than a screenshot of the winning name.
What if the winner never responds?
Announce the rule before you draw: how long they have to claim, and what happens next. The clean approach is to draw a reserve or two at the same time, in the same recorded action, and publish their positions. Quietly drawing again a week later looks identical to drawing until you get a result you prefer.
Do I need to say the draw was random?
Say how it was run, in one line: where the entries came from, whether duplicates were removed, and what tool drew the winner. That single sentence prevents most complaints. Promotions also carry legal requirements that vary by country and by platform, so check the rules of your jurisdiction and the platform you are running it on.
How the randomness works
Sources: