What "transparent" can and cannot mean
Be precise about the claim you are making. No consumer tool can prove to a stranger that a draw was not manipulated — an organiser with something to hide can always run the draw twice off-camera and publish the result they preferred. What a good process can do is make manipulation visible if it happens: the entrant list is published before the draw, the draw is a single continuous recording, the result is written down immediately, and every claim you make is checkable against something. That is a much more defensible position than "trust us, it was random", and it is worth stating in those terms rather than implying an audit that did not happen.
Step 1 — Publish rules before entries open
Almost every dispute traces back to a rule invented after the fact. Write down, before the first entry:
- The closing date and time, with a time zone.
- What counts as an entry, and whether one person can have several.
- Exactly how bonus entries are earned and how many each action is worth.
- How many winners there will be, and whether runners-up will be drawn at the same time.
- How the winner will be contacted and what happens if they do not reply within a stated window.
- Any eligibility limits — region, age, shipping restrictions.
Platform-specific promotion rules apply on top of your own. Instagram, for example, requires that promotions include a release of the platform and acknowledge it is not sponsored by it; check the current promotion guidelines for the platform you are running on, since these change and vary by region. Prize draws are also regulated differently in different countries — if the prize is substantial, take local advice rather than copying another creator's terms.
Step 2 — Freeze the entrant list
At the closing time, export the entrant list once and treat that file as the record. Post it or a hash of it before you draw, so the list you spin can be compared with the list you promised. Then do your de-duplication and eligibility filtering as a visible step — explaining "247 comments became 231 entries after removing duplicates and ineligible regions" is far stronger than silently arriving at 231.
A CSV or TXT export loads straight into the giveaway wheel, which parses the file in your browser — nothing is uploaded — and lets you choose the name column, preview the rows and dedupe before the entries land on the wheel.
Step 3 — Handle bonus entries honestly
If your rules award extra entries, the mechanism must match the rules exactly. There are two correct ways to do it, and the difference is presentational rather than mathematical:
| Approach | How it looks | When to use it |
|---|---|---|
| Repeat the name on multiple lines | Each line is an independent slice on the wheel; three lines is three chances. | Small entrant lists, where viewers can count the lines and verify the odds themselves. |
| Enable weights | One row per entrant with a number; weight 3 against weight 1 is three times as likely. | Large or imported lists, where repeating names would make the list unreadable. |
Both give the same probabilities. Under weighting, an entrant's chance is their weight divided by the total weight of all entries — so in a draw with 100 single-entry participants and one participant on weight 3, that entrant's chance is 3/102, or about 2.9%. Publish the weights alongside the entrant list; a bonus-entry scheme nobody can audit is the most common source of "it was rigged" accusations.
Step 4 — Draw on camera, in one take
The recording is the evidence, so structure it deliberately:
- Start the recording with the full entrant list visible and scroll through it, so viewers can find their own name before anything is drawn.
- Show the settings: number of winners, whether repeat winners are allowed, whether weights are on.
- Draw without cutting. One continuous clip from list to winner is the entire point; an edited draw proves nothing.
- Post the unedited clip, not a screenshot of the result.
Draw runners-up in the same session rather than spinning again days later if the first winner does not reply. On the giveaway wheel, setting the number of winners above one runs a real spin per winner in sequence; with repeat winners off, each winner leaves the wheel before the next spin, so the remaining entrants' chances rise accordingly and every winner is distinct. That behaviour — sampling without replacement — should be stated in the rules, because it is what most people assume but few draws actually say.
Step 5 — Write the result down immediately
A result announced only in a video is hard to reference later. Capture a written record at the moment of the draw containing the timestamp, the entrant count, the winners in order, and the settings that were in force. The giveaway wheel's Draw Record does this: it downloads a file listing the draw time, entrant count, requested winners, whether repeats and weights were enabled, the randomness method, the winners in draw order, and a SHA-256 hash of the canonical draw inputs.
That hash is worth describing accurately, because it is easy to oversell. It is computed over the recorded inputs — the entries, their weights, the repeat setting and the requested winner count — so two records of the same configuration produce the same hash and any alteration to the recorded list produces a different one. It lets someone compare a published record against a published entrant list. It does not prove the draw was not re-run, and WheelOurNames does not describe it as a certification or an independent verification. Say the same when you publish it.
Step 6 — Publish and follow through
- Post the winners publicly, in the same place the giveaway was announced.
- Attach the Draw Record or its hash next to the recording.
- Contact winners through the channel you said you would, within the window you stated.
- If a winner does not respond, say publicly that you are moving to the runner-up drawn in the original session — not that you are running a fresh draw.
Common mistakes
- Changing the rules mid-draw. Adding an eligibility criterion after seeing the winner is indistinguishable from rigging, even when it is innocent.
- Redrawing quietly. If you must redraw, say why and record it.
- Claiming certification. "Verified fair" implies a third party. Describe your evidence instead: the published list, the unedited recording, the record file.
- Losing the entrant list. Keep the frozen export; without it, nothing else you published can be checked.
- Bonus entries nobody can count. If the weights are not published, the odds are not public.
For the mechanics of how the selection itself is computed — the random source, the removal of modulo bias and how weights become probabilities — the fairness page documents it, and its simulator lets a sceptical viewer run thousands of virtual spins and inspect the distribution.