Freeze eligibility before drawing
Record the closing time, eligibility rules, entry limit, disqualifications, and prize order before selection. Resolve invalid and late entries before any outcome is visible. A list changed after a result appears creates discretion that randomness cannot correct.
Assign every valid entry a unique numeric identifier. Duplicate names can represent different people, so keep a private mapping with the minimum personal data needed. Public evidence should use masked names or entry numbers rather than email addresses, phone numbers, addresses, or birth dates.
Handle duplicates according to the rules
A one-person-one-entry drawing should deduplicate people using a declared method and review ambiguous matches. An entry-based promotion may allow several chances per person; each valid chance then needs its own distinct identifier. Removing those entries would change the promised weighting.
NumUtility removes duplicate numeric values and treats each distinct number once. Assign separate numbers for authorized multiple entries and verify the parsed eligible count. Do not paste the same number repeatedly to create weighting because it will not do so.
Three complete draw examples
Input: 50 people numbered 1–50, one prize. Formula: choose uniformly from 50 identifiers. Calculation: each has probability 1/50=2%. Result: suppose 17 is drawn. Interpretation: identifier 17 is provisional until eligibility is verified.
Input: 12 entries and three equal prizes. Formula: sample three identifiers without replacement. Calculation: 12 choices, then 11, then 10. Result: three distinct identifiers. Interpretation: no entry wins twice.
Input: 100 entries, one person validly owns four. Formula: one-prize probability=valid entries/total entries. Calculation: 4/100=4%. Result: that person's chance is 4%. Interpretation: this follows entry weighting, not one-person-one-chance rules.
Plan prizes, alternates, and redraws
State whether prizes are equal, drawn in order, or limited to one per person. For unequal prizes, use the published order or a separate documented assignment. Equivalent prizes can be selected together without replacement.
Draw alternates only under a predeclared condition such as confirmed ineligibility or failure to respond by a stated deadline. Log contact attempts and the redraw reason. Never redraw merely because an organizer dislikes a valid outcome.
Keep useful evidence while protecting privacy
Retain the final identifier list, rules version, timestamp, tool, requested quantity, raw result, eligibility verification, and any replacement decision. A witness, list hash, or screen recording can improve confidence for an important community draw.
Follow promotion, tax, consumer, platform, and privacy rules in the relevant jurisdiction. Software selection does not supply official rules or consent. Publish enough method detail for understanding without exposing private entrant data.
Separate fairness from appearance
A cryptographically strong random source supports unpredictable selection, but cannot correct hidden entries, undisclosed weighting, or selective redraws. Describe concrete controls instead of claiming a draw is guaranteed fair or independently certified without evidence.
Practice with clearly separate sample data. Reset before the binding draw and announce when it begins. Never label a valid live result a test after seeing it. For weighted selection, model each authorized chance explicitly or use a reviewed weighted method.
When winners must be unique people, remove all other identifiers belonging to a selected person only if the rules impose a one-prize-per-person limit. Document that transformation. Otherwise, post-selection discretion can change probabilities for remaining entrants.
For unequal weighting, ordinary duplicate-free selection is not enough. Model every authorized chance with its own identifier or use a reviewed weighted method. Show entrants how chances are assigned without publishing private account data.
Before announcing, verify the selected identifier against the frozen private mapping, eligibility evidence, and prize restrictions. Call the result provisional until that check is complete. Verification should confirm facts, not introduce a subjective preference.
Build the list from a stable source export rather than copying names from a scrolling interface where entries can be missed. Count source records, excluded records, and final eligible records, and reconcile the totals. Keep exclusion reasons such as duplicate, late, or ineligible in a private review file so the organizer can answer a legitimate question consistently.
If an entrant withdraws after the list is frozen but before selection, apply the published withdrawal rule and create a new timestamped version. Do not erase the earlier file. Version history shows that the change preceded the draw and was not prompted by an unwanted result. The same rule should apply to every similarly situated entrant.
For several prize tiers, probability depends on order and eligibility after each award. A one-prize-per-person rule removes all of the winner's remaining entries before the next tier. If repeated wins are allowed, those entries remain. State the rule in advance because both designs can be mathematically valid yet produce different chances.
A public livestream can improve visibility but does not replace data controls. Viewers cannot verify unseen exclusions or private mappings from the screen alone. Pair visible selection with a concise methodology, final count, masked result, and contact timeline. Avoid showing browser notifications, private tabs, or entrant details during screen sharing.
If the tool or browser closes before a result is saved, follow a predetermined incident procedure. Do not reconstruct a missing value from memory when evidence is uncertain. Record the failure, confirm no usable result was captured, restore the same frozen population, and rerun with a witness when appropriate.
Use the Random Number Picker
Paste distinct numeric identifiers, choose the quantity, check the eligible count, and draw. Results are selected without replacement, so returned values do not repeat.
The picker does not validate people, contact winners, identify one person behind several identifiers, or establish legal compliance. The organizer owns those steps and should preserve the frozen input alongside the result.
Reserve entries deserve the same controls as primary winners. Either select a declared number of ordered alternates during the original session or run a documented later draw from the correct remaining population. Do not quietly choose a convenient substitute. Record whether alternates were ranked and when each became eligible to receive the prize.
If an entry list contains decimals or negative identifiers, the picker can accept them, but simple positive whole-number IDs are easier to communicate and less prone to formatting changes. Spreadsheet exports may turn long numbers into scientific notation or remove leading zeros. Validate the exact parsed values before selection and preserve the original-to-draw mapping.
After the promotion closes, retain records only as long as required for administration, disputes, tax, or legal obligations. Then dispose of personal data securely according to the published privacy notice. An audit trail does not justify indefinite retention of every entrant's contact details, and a public winner announcement should use consented information.
For team or classroom use, explain the process before displaying the outcome. Participants should know the final eligible count, number of selections, replacement rule, and how conflicts will be handled. Afterward, report the selected identifiers and verification status using privacy-safe labels. If someone challenges the result, compare the frozen list and recorded procedure rather than generating another number informally. A repeatable governance process builds more trust than visual animation. Animation may make a draw engaging, but it is not evidence of stronger randomness. Likewise, a plain result is not less fair merely because it appears immediately. Evaluate the source, mapping, rules, and records.