how to pick a random winner

Picking the winner takes ten seconds. Everything that makes those ten seconds fair happens before and after them: rules written before anyone entered, an entry list cleaned the way you said it would be, and a draw that a disappointed entrant can inspect afterwards. This guide runs that order — rules, list, proof — with the actual behaviour of the real tools and the actual text of the official rulebooks, and it is a process guide, not legal advice; the law appears briefly, attributed, near the end. Publish the giveaway draw rules where entrants can find them without asking — secrecy reads as rigging. Screenshot the giveaway draw before, during and after; the album is your appeals court.

Write the rules before entries open

A fair draw’s first document is not the winner announcement; it is the rule sheet, published before the first entry arrives. The complete version contains: who may enter (age, country or region); how to enter, including a free alternative route if any entry action could count as a cost; the deadline, with its timezone; the prize and its value; how the winner will be chosen, naming the method rather than just “randomly”; how the winner will be contacted; how long they have to respond; and what happens then — the redraw policy, including how many runner-ups you will draw from the same list, in reserve, before announcing anyone.

Each line exists because its absence has burned someone. The response window is the clearest: the strongest process guidance found for this guide advises stating five to seven days up front, with one follow-up message at two to three days — invented after a silent winner, the same numbers look rigged. In the UK this “before” has regulatory teeth: the advertising code (CAP 8.17) expects significant conditions to be available before entry and does not permit the terms to be changed afterwards, closing dates included. Rules written first are not bureaucracy; they are the only version of you that cannot be accused of moving the goalposts. Every giveaway draw needs its runner-ups drawn in the same session, not invented after a winner goes quiet. The best giveaway draw is the boring one — announced early, run plainly, closed on time.

Clean the entry list — and be able to show it

“We removed the duplicates” is a claim, and claims need behaviour behind them. The market’s best-known comment picker documents its own behaviour precisely: duplicate-filtering is a setting you switch on, not a silent default; when on, entries are deduplicated by username and only each user’s first comment is kept; bonus entries bypass the filters entirely. A host who never found that setting has not run the draw they described. Whatever tool you use, learn its actual duplicate behaviour and say it in the rules.

Then screen what remains: entries that break your stated eligibility (wrong region, under-age, the same person behind five accounts — one platform’s fraud tooling treats duplicate-IP entries as first-valid, rest-invalid, while warning that shared households can look identical); and the quiet risk nobody mentions, completeness — comments hidden by the platform, held for review, or attached to ads can be missing from the pool your tool collects, so reconcile the list against the post before drawing. If you allow weighted entries (three entries for a share, say), that is legitimate — when disclosed. Its mirror image is the sharpest line in this guide: an unremoved duplicate is an undisclosed weighting. Publish the count before cleaning and after; the arithmetic is the audit’s beginning. Record the giveaway draw as it happens; the recording is your answer to every later argument.

Prove the draw: the proof ladder

Fairness that cannot be shown is a promise. Climb as high as your prize justifies.

Rung one — record it. Draw live, or record the whole draw uncut; as one draw service puts it, a cut destroys the proof. The recording shows the list, the tool, the settings and the result in one unbroken take.

Rung two — publish the list before the draw. Post the final entry list (or its cryptographic hash) before anyone wins. A hash is a fingerprint of the list: publish the fingerprint first, and afterwards anyone can check that the list drawn from is the list that existed — nothing swapped, nothing quietly added.

Rung three — let a third party hold the record. For prizes worth arguing about, RANDOM.ORG’s third-party draw service exists precisely for this: a single drawing for up to 500 entrants costs $4.95, and its record of the draw — organiser, time in UTC, every entrant, the winners — cannot be modified, is kept for at least five years, and can be checked by entrants themselves with their email address. Its guarantees are the three a fair draw needs: the organiser cannot influence the outcome, cannot predict it, and every entrant has an equal chance.

Underneath all three rungs sits the randomness itself. Prefer tools whose randomness is cryptographic: MDN’s documentation warns that the common Math.random() “does not provide cryptographically secure random numbers” and points developers to Crypto.getRandomValues() instead; Wheel of Names states in its own FAQ that its results are driven by that cryptographic source (one comparison site disputes the claim — the maker’s own documentation is the evidence weighed here). The balanced view from developers themselves: fine for a classroom spinner; for a raffle with real money attached, use the cryptographic kind. And one freedom the ladder gives you: position in the list does not matter — every serious picker source states each entry’s chance as 1-in-N regardless of order, so “I commented first” buys nothing, and neither does commenting last. The window matters too: a giveaway draw announced for five to seven days collects trust along with entries.

Draw day, and the winner who goes silent

Run the draw promptly — within forty-eight hours of the deadline is the strongest guidance’s figure — and announce it with the proof attached: the recording or record link, the entry count, and the winner named as your rules said they would be. (One UK rule worth knowing before you announce: CAP 8.28.5 expects the surname and county of major prize winners to be published or available.) Contact the winner by the stated method, start the stated clock, and send the stated follow-up.

Then the moment this whole structure exists for: silence. Because your rules were written first, nothing is improvised — the window expires, the forfeiture is documented (the dates and the attempts, in writing), and the pre-drawn first runner-up steps up. No second draw, no fresh list, no accusations with anywhere to land. The runner-up system only works because the reserve names came from the same pool, at the same moment, under the same recording. When the giveaway draw ends, announce once, clearly, with the proof attached — then stop touching it.

Platform rules, from the official pages

YouTube (from its official contest policies): your promotion is solely your responsibility; entry must be free; you may not imply YouTube’s involvement without written consent; and your official rules must state that YouTube is not a sponsor, require entrants to release YouTube, link the Community Guidelines, and include a legally compliant privacy notice about the data you collect.

TikTok: the promotion rules that could be read on TikTok’s own official pages for this guide cover TikTok Shop in the US — free entry, items under $500, the platform’s official LIVE Giveaway feature only — so treat that scope honestly: they are the rules of one programme, not of every TikTok giveaway everywhere.

Instagram: a necessary admission — for this guide, Instagram’s own official wording could not be read directly, so nothing here will be dressed up as its text. What secondary sources consistently quote from Meta’s promotion rules is a required statement that the promotion is in no way sponsored, endorsed or administered by Instagram, plus a complete release of the platform by each entrant; verify the current wording on Meta’s own help pages before you publish your rules, because platforms revise them. X’s official giveaway guidance likewise could not be verified for this guide. Treat your first giveaway draw as a rehearsal for a bigger one; the habits scale, the shortcuts do too.

The legal shape, briefly

General information, as the sources themselves frame it — not advice about your draw. In the United States, explainer law is consistent that an illegal lottery combines three things: a prize, chance, and consideration (something of value given to enter). A sweepstakes removes the consideration — hence the free-entry route — and a skill contest removes the chance. Two state regimes are quoted in every serious guide: in New York and Florida, if total prizes exceed $5,000, the promotion must be registered and bonded in advance (New York: at least 30 days ahead, a $100 fee; Florida: at least 7 days, $100), and Rhode Island requires registration once prizes pass $500 (a $150 fee). In the UK, a genuinely free draw or prize competition needs no gambling licence when it meets the Gambling Act 2005 exemptions — but the CAP Code still demands what this guide has been demanding throughout: evidence of a verifiably random process or a demonstrably independent observer watching the draw (CAP 8.24). Bigger prize, bigger jurisdiction, real money mechanics — that is the point where general information ends and a lawyer begins.

Frequently asked questions

How do I pick a random winner fairly?
In three parts: write the complete rules (including the redraw policy) before entries open; clean the list the way the rules said, and publish the counts; then draw with proof — an uncut recording at minimum, a published list or third-party record for anything valuable. The picker itself is the smallest part.

Should duplicate entries count once or more than once?
Either is defensible — undisclosed is not. The common default is one entry per person, but check your tool: the best-known comment picker’s duplicate filter is a setting that keeps only each user’s first comment, and bonus entries can bypass filters. Whatever behaviour you run, your rules should describe it.

Does it matter if I comment first or last?
No — this is the giveaway world’s favourite myth. Every serious picker source gives each entry a 1-in-N chance regardless of position. The order-related risk that is real is different: comments hidden or held by the platform may never reach your list at all.

Do I have to livestream my giveaway draw?
No. A livestream is the top-proof option most small hosts do not need; an uncut recording is rung one, a pre-published entry list (or its hash) is stronger still, and a third-party record outranks both. Match the rung to the prize.

What if the winner does not respond?
Whatever your pre-written rules say — which is why they must say it. The standard shape: a five-to-seven-day window stated at launch, one follow-up at two to three days, the forfeiture documented, and a runner-up drawn from the same pool at the same draw stepping up. Never re-draw from scratch.

Is my giveaway legal?
The question this guide can honestly answer is narrower: is it structured legally? Free entry is the foundation almost everywhere; the US adds registration duties at prize thresholds ($5,000 in New York and Florida; $500 in Rhode Island), and the UK demands a verifiably random process or an independent observer. Beyond that shape — your country, your prize, your platform — you need the official rules and, for anything substantial, actual advice.

Fairness is paperwork done early

Nobody trusts a draw because the host felt honest; they trust it because the rules predate the entries, the list survived inspection, and the proof still exists. Set the next one up that way, and the ten seconds of picking become the least interesting part — as they should be. When it is time to pick, the random name picker draws from your cleaned list with the duplicates handled the way your rules promised.