What 2.3 million giveaway draws reveal

The widest gap in the study, and it is not close.

December holds 24.60% of the year's giveaways. February, the next busiest, reaches 8.79%. August, the quietest, falls to 5.02%. December is nearly five times August.

And yet it is the month where a giveaway collects the fewest entries: a median of 106 comments, against 213 in November, 185 in March, and 144 across the year. A December giveaway gathers half what a November one does.

The explanation is crowding. In December your audience sees thirty giveaways in three weeks, and each of them captures less because they are dividing the same attention. The month when everybody posts is the month each of them harvests least.

Two honest caveats. December also concentrates advent calendars, which produce twenty-four separate posts, each necessarily smaller than a single giveaway — some of the gap comes from there, though the order of magnitude exceeds what that alone would explain. And the concentration is measured across the whole base: it is presumably driven by markets where Christmas structures the retail calendar, which is not everywhere.

What it changes: if you are after volume, November beats December and you will have it to yourself. If you are supporting the holidays, December is unavoidable — but expect half your usual participation, and size the prize accordingly.

Measured on 473,232 comments: the delay between the post going up and each comment arriving.

• 1 hour: 11.5% of comments are already in
• 3 hours: 23.5%
• 6 hours: 31.9%
• 12 hours: 39.0%
• 24 hours: 51.2%
• 48 hours: 61.6%
• 72 hours: 69.4%
• 7 days: 88.6%

The first day is worth as much as everything else combined, and the first three hours alone are worth nearly a quarter.

This corrects a piece of advice we had published ourselves. It is widely repeated — and we had written it — that a giveaway post "stops reaching anyone after 48 hours". It does not: 38.4% of comments arrive after the forty-eighth hour. What genuinely collapses is the week mark, beyond which only 11.4% is left to collect.

So seven days remains the right upper bound, for a different reason than the one usually given. It is not that nothing arrives; it is that the little still arriving no longer justifies the wait imposed on everyone who has already played. How long a giveaway should run covers the scheduling.

The result we did not expect to be this sharp.

We matched the configuration of 1,298 giveaways from 2024-2025 against the comments they actually received, and compared the median per group:

• No tag required — 124 giveaways, median of 203 comments
• One tag — 514 giveaways, median of 209 comments
• Two tags — 420 giveaways, median of 93 comments
• Three tags — 240 giveaways, median of 86 comments

Going from zero to one tag costs nothing: 203 against 209, not a meaningful gap. Meanwhile every person tagged receives a notification, which is a free contact with someone who does not follow you. The first mention is free in both senses.

Going from one to two divides participation by 2.2. The third takes almost nothing more: the damage is done at the second.

An honest reservation: these are correlations. Accounts requiring three tags are not necessarily the same accounts as those requiring one, and part of the gap may come from that rather than from the constraint. But the effect is monotonic and it covers 1,298 operations.

Across 3,969 giveaways that received at least twenty comments, we compared the number of comments to the number of distinct accounts.

In the median giveaway, 45.3% of comments come from someone who had already commented. More precisely: 61.5% of giveaways pass 30% duplicates, and 45.4% pass 50% — in nearly one giveaway in two, most comments are repeats.

Set against another figure, that becomes a problem: 23.3% of organisers do not deduplicate before drawing. So one giveaway in four picks its winner from a list where whoever commented ten times has ten times the chance — without anyone deciding it, and without entrants being told.

Of every figure in this study, that is the one with the most effect on whether a draw is fair. Picking a giveaway winner properly covers what the setting changes.

Across roughly a million recent giveaways:

• 90.9% require at least one tag. One mention 38.8%, two 29.5%, three 20.0%, none 9.1%. Most organisers therefore sit in the range where, by the measurement above, the constraint is costing them entrants.
• 23.3% do not deduplicate, as above.
• 82.3% name a single winner. Two or three winners account for 14.8%; beyond that it is negligible.
• 99.5% run on comments. Reposts and follows are residual in practice, whatever the guides say.

As for what to expect: across 2,362 posts, the median is 144 comments, the first quartile 49, the third 471, the ninth decile 1,260 and the 99th percentile 5,695. The largest observed was 37,997. The distribution is heavily skewed — if this is your first giveaway, the median is your number, not the ones you see going around.

Duration actually practised, between the post and the first draw across 89,747 giveaways: a median of 6.9 days, quartiles at 2.8 and 11.2. The three-to-seven-day band is the most common at 34.8%, and 61.2% are drawn within the week.

By day of posting, median comments collected: Sunday 171, Thursday 150, Wednesday 148, Tuesday and Friday 144, Monday 133, Saturday 120. Saturday is the worst day and Sunday the best — a 42% spread.

By time of day: 10am–2pm gives 164, 6pm–9pm gives 159, 2pm–6pm gives 156, then 9pm–midnight 129, 6am–10am 107, and overnight 100. Late morning edges out the evening, which contradicts the usual advice to post at 8pm. The gap between those two is small; the gap with early morning is not — posting before 10am costs a third of your entries.

Not to be confused with when draws are run, which follows the opposite logic: Monday (17.5%) and Sunday (16.2%), rarely Saturday (10.6%), peaking at 8pm.

Across 93,212 giveaways in 2024-2025: Instagram 64.0% (of which 3.9% in direct messages), Facebook 36.0%.

One giveaway in three still runs on Facebook, which is worth knowing for anyone who treats it as abandoned. Participation is not the same, though: a median of 88 comments on Facebook against 193 on Instagram, less than half. Read the two together — Facebook wants a more modest prize for a more local, older audience, and its rules differ. Running a Facebook giveaway covers what changes.

These are giveaways run with Lukky, not every giveaway in the world. They lean towards organisers who were looking for a draw tool — which is to say, presumably, towards those who take randomness seriously.

We do not know which countries they come from. A language is attached to only 30% of them, too few to publish a market breakdown. That matters for the seasonality above: the December peak belongs to the whole base, and the fraction we can attribute suggests it is far stronger among French-speaking organisers than Spanish-speaking ones — but that fraction is too partial to be a result.

The comparisons are correlations. We measure what was configured and what was obtained; we have not isolated the effect of a setting from the quality of the prize, the size of the account or the timing of the post.

• Ask for one tag, not two. The first is free; the second costs half your entrants.
• Spend your effort on the first three hours. They are worth a quarter of the total, and the first day is worth half.
• Close between three and seven days. Beyond that you buy 11% more entries at the price of everyone else's wait.
• Always deduplicate. The median giveaway is 45% duplicates, and one organiser in four draws without removing them.
• November over December, if volume is the goal.

The draws behind these figures were all run with Lukky, which generates the draw video at the end of each one.

Which month is best for a giveaway? November, if volume is the goal: it produces the year's highest median participation at 213 comments. December holds 24.6% of all giveaways and delivers the lowest at 106 — the month when everyone posts is the month each of them harvests least.

How long should a giveaway run? Three to seven days. Half the comments arrive within 24 hours, 69% within three days, 89% within a week. Beyond that only 11% is left.

Should I ask entrants to tag friends? One tag: it costs no participation and buys a notification to someone who does not follow you. Two drops the median from 209 comments to 93.

How many entries should I expect? The median is 144 comments per giveaway. A quarter get fewer than 49, and 1% pass 5,695.

Should duplicates be removed before drawing? Yes. In the median giveaway 45% of comments come from someone who had already commented, and nearly one giveaway in two passes 50% duplicates.

What is this study based on? 2,348,397 real draws and more than a billion entries processed with Lukky since 2019, with the detailed analyses covering the complete years 2024 and 2025.