Skip to content

Crowd9/giveaway-skills / skills/giveaway-promotion-plan

giveaway-promotion-plan

Verified

Promote or rescue a giveaway. Use for 'how do I promote my giveaway', 'launch posts', 'giveaway email sequence', 'partner brief', 'should I boost the post', 'nobody is entering', 'entries look fake', 'it is rigged', 'someone is impersonating us', 'we got taken down', 'can I change the Prize or end date', or 'should I extend'. Write schedules, copy and next steps.

FileKindSize
SKILL.mdskill md15.3 KB
evals/cases.mdmarkdown6.3 KB
evals/evals.jsonconfig32.0 KB
references/channel-playbook.mdmarkdown39.2 KB
references/email-sequence.mdmarkdown7.8 KB
references/evidence-detail.mdmarkdown3.2 KB
references/house-style.mdmarkdown4.7 KB
references/live-campaign-rescue.mdmarkdown16.1 KB
scripts/style_check.pyexecutablescript40.3 KB

references/channel-playbook.md

View on GitHub
# Channel playbook

Advice. Formats change, so treat lengths and features as current practice, and check any platform rule in the entry-method planner's compliance notes before using tag-to-enter or share-to-enter.

<!-- generated:pp_quintile_lead -->
Extracted: the top fifth of campaigns by Entrants saw a typical 14,070 Impressions against 552 for the bottom fifth, at Conversion Rates two points apart (27.6% against 25.4%). Reach is what separates them, so promotion is the lever this playbook is for. [`analysis/output/context_checks.json`, `top_vs_bottom_quintile`, 23,256 campaigns each side.]
<!-- /generated -->

## Measured traffic mix

Direct traffic supplies nearly half of all Impressions, well ahead of any single channel. The rest splits across smaller traffic sources, none above 10%. This cut is a share of Impressions and carries no campaign count of its own, so never borrow one from another table on this page: the pair that used to sit here matched no run of the source.

| Traffic source | Share of Impressions |
|---|---|
| Direct | 47.9% |
| The business's own site | 9.6% |
| Meta | 9.0% |
| YouTube | 6.1% |
| Giveaway directories | 4.8% |
| Search | 3.9% |
| X | 3.7% |
| Gaming communities | 3.1% |
| Gleam's own pages | 3.1% |
| Deals forums | 1.2% |
| TikTok | 0.9% |

Direct is the largest single line and means unattributed traffic. DMs, texts, QR links and untagged email can all leave visits without a recorded source. Direct alone does not establish word of mouth or prove that referrals worked. When attribution matters, use tagged links for each placement and recorded referral measures to check which activity brought people in.

Embeds carried 51% of Impressions and hosted pages 49%. Traffic splits almost evenly between the Gleam-hosted page and an embed on the organizer's own site, so the profile-prep and per-channel steps below apply whichever way the campaign is set up.

Source: analysis/output/field_cuts.json (referrer_channel_share, landing_kind_share).

## Where to list a giveaway

Extracted: the giveaway directories, deals forums and gaming communities that reached the most organizers. These are public third-party sites, so the host names are published.

What this table is, and what it is not. It counts the traffic these sites sent, which says they are worth submitting to and nothing more. How a site takes a submission, what it charges, how long it sits in a queue and whether it accepts a campaign at all are not in this data and not on any page read here. Tell the reader to check the site's own submission page, and never put a turnaround time in an answer. Price is the same: "free" is a claim about the site, and this data holds none. Both rounds of measured answers broke this rule the same way, one writing "free, and it's live within a day" and the other "submit free", so the sentence naming a directory carries "check the submission page for what they need" in the same breath. Directories carry 4.8% of Impressions in the traffic mix on this page, behind the business's own site at 9.6%, so never call them the biggest source of anything. A measured answer called them "the largest free reach on the table" in one round and "the largest free reach you have not already counted" in the next.

| Giveaway directories | Businesses reached | Impressions |
|---|---|---|
| sweepsadvantage.com | 4,808 | 4,364,572 |
| freestufftimes.com | 4,784 | 1,242,211 |
| contestgirl.com | 4,314 | 18,415,684 |
| giveawaybase.com | 3,111 | 4,450,711 |
| ozbargain.com.au | 2,314 | 2,973,405 |

| Deals forums | Businesses reached | Impressions |
|---|---|---|
| hardprize.ru | 994 | 26,430 |
| slickdeals.net | 671 | 80,294 |
| lottos.com.au | 653 | 324,699 |
| loquax.co.uk | 640 | 2,273,354 |
| ofertar.pt | 590 | 26,290 |

| Gaming communities | Businesses reached | Impressions |
|---|---|---|
| steamcommunity.com | 1,000 | 1,589,880 |
| store.steampowered.com | 202 | 47,759 |
| gaming.lenovo.com | 96 | 235,967 |
| ledemondujeu.com | 88 | 78,631 |
| gununiversity.com | 82 | 213,774 |

Extracted: outcomes for campaigns where one traffic source supplies over 10% of a campaign's Impressions.

| Traffic source over 10% of Impressions | Conversion Rate | Entrants | Entries each | Campaigns | Businesses |
|---|---|---|---|---|---|
| YouTube | 41.4% | 548 | 3.50 | 13,447 | 2,890 |
| Deals forums | 34.7% | 479 | 4.91 | 6,767 | 514 |
| Giveaway directories | 31.4% | 682 | 4.81 | 17,968 | 4,388 |
| X | 30.3% | 393 | 4.89 | 18,917 | 4,349 |
| The business's own site | 27.9% | 771 | 4.31 | 10,914 | 1,832 |
| Direct | 26.0% | 490 | 4.56 | 106,096 | 16,876 |
| Meta | 24.8% | 465 | 3.40 | 21,251 | 6,030 |
| The Gleam directory | 23.9% | 622 | 4.78 | 7,443 | 2,436 |
| Search | 16.9% | 354 | 4.44 | 3,769 | 1,315 |

Campaigns receiving over 10% of their Impressions from YouTube or deals forums have the highest whole-campaign Conversion Rates in this table, while campaigns meeting that threshold for search have the lowest. These are conditional campaign medians, not source-attributed conversion or referral rates. The comparison does not show which channel converts its own visitors best or what adding a channel would do. [Extracted from `analysis/output/field_cuts.json`, `outcomes_when_channel_over_10pct`.]
Campaigns meeting the deals-forum threshold recorded a 34.7% whole-campaign Conversion Rate, behind the YouTube group's 41.4%. Their whole-campaign referral rate is lowest among the sources compared here. Size-band comparisons also describe whole campaigns. These outcomes alone do not establish that deals-forum visitors convert well or rarely refer a friend. Subscriber quality is unknown, and email action completions do not establish a count of distinct subscribers.

| Measure | Deals forums | For comparison |
|---|---|---|
| Conversion Rate | 34.7% | direct 26.0%, Meta 24.8% |
| Entries per Entrant | 4.91 | direct 4.56, Meta 3.40 |
| Referrals per 100 Entrants | 5.3, the lowest of the well-populated sources | direct 11.6, X 13.7 |
| Entries per Entrant, campaigns of 100 to 250 Entrants | 7.13 [1,312 campaigns, 200 businesses] | - |
| Entries per Entrant, campaigns of 1,000 to 2,500 Entrants | 1.76 [933 campaigns, 142 businesses] | - |
| Entries per Entrant, campaigns of 2,500 to 10,000 Entrants | 2.31 [230 campaigns, 59 businesses] | - |

Campaigns meeting the deals-forum threshold recorded 5.3 referrals per 100 Entrants against 11.6 for campaigns meeting the direct-traffic threshold. Their Entries per Entrant median is 4.91, above the direct and Meta groups. These measures cover all Entrants in each campaign. [Extracted from `analysis/output/field_cuts.json`, `channel_quantity_vs_quality`, `outcomes_when_channel_over_10pct` and `channel_quality_by_size_band`.]

Campaigns meeting the Meta threshold recorded a whole-campaign Conversion Rate five and a half points under the X group, 24.8% against 30.3%. Meta supplies a larger share of all Impressions. Whole-campaign Entries per Entrant are lower in the Meta group, including within the size bands shown. These comparisons do not identify which visitors completed Actions or establish that adding a referral or bonus-entry push would close the gap.

| Measure | Meta | X |
|---|---|---|
| Conversion Rate | 24.8% | 30.3% |
| Share of all Impressions | 9.0% | 3.7% |
| Entries per Entrant, overall | 3.40 | 4.89 |
| Entries per Entrant, 1,000-2,500 Entrants | 3.39 [3,707 campaigns, 1,683 businesses] | 4.69 [2,699 campaigns, 979 businesses] |
| Entries per Entrant, 2,500-10,000 Entrants | 3.34 [2,432 campaigns, 1,070 businesses] | 5.79 [1,430 campaigns, 498 businesses] |
| Entries per Entrant, 10,000+ Entrants | 2.84 [472 campaigns, 238 businesses] | 6.43 [325 campaigns, 115 businesses] |

Source: analysis/output/field_cuts.json (promotion_sites, outcomes_when_channel_over_10pct, channel_quantity_vs_quality, channel_quality_by_size_band).

## Tagging with UTM

Extracted, same campaign set: only 15% of Impressions carry a utm_source tag. Most traffic arrives untagged, which is exactly why the UTM line under Paid below matters on organic posts too. Tag every link you post, or the results review is left reading the traffic source and nothing about which post, story or email sent the visit.

| Of the tagged slice | Share |
|---|---|
| Email newsletters | 32% |
| Meta | 23% |
| Ad networks | 4.4% |
| TikTok | 3.4% |

Source: analysis/output/field_cuts.json (utm_source_share).

## Mass mail into the campaign

Extracted: among Impressions that carry an email UTM tag, share by provider named in utm_source. This is the organizer's own mass mail arriving at the campaign page, none of it sent by Gleam, and it is a share of the 15% of Impressions that carry any UTM tag at all, not a share of all traffic.

| Provider | Share of tagged email Impressions | Campaigns | Businesses |
|---|---|---|---|
| Unnamed or untagged | 35% | 14,110 | 2,818 |
| Listrak | 18% | 596 | 78 |
| Named newsletter source | 16% | 6,096 | 1,037 |
| Klaviyo | 14% | 3,818 | 1,185 |
| HubSpot | 3.9% | 1,172 | 313 |
| Attentive or SMS | 2.6% | 694 | 232 |
| Salesforce or ExactTarget | 1.9% | 331 | 74 |
| ActiveCampaign | 1.9% | 575 | 114 |
| ConvertKit | 1.8% | 236 | 57 |
| Sailthru | 1.4% | 728 | 36 |
| Omnisend | 1.1% | 654 | 212 |
| Iterable | 0.8% | 183 | 43 |
| MailChimp | 0.7% | 176 | 67 |
| Brevo or Sendinblue | 0.7% | 479 | 129 |
| Braze | 0.4% | 58 | 22 |

Campaigns that draw a tenth or more of their Impressions from their own tagged mass mail behave differently from campaigns with no tagged email traffic at all, though the two groups are not alike: the businesses that tag their email traffic are larger and run longer campaigns, so the gap describes who mails, not what mailing does.

<!-- generated:pp_mail_share -->
| Group | Conversion Rate | Entries per Entrant | Entrants (typical) | Campaigns | Businesses |
|---|---|---|---|---|---|
| Under 2% | 25.8% | 5.03 | 765 | 98,367 | 17,539 |
| No email traffic at all | 34.9% | 3.98 | 240 | 31,679 | 7,996 |
| 2 to 10% | 25.0% | 4.26 | 744 | 14,594 | 3,276 |
| 10 to 25% | 24.9% | 4.36 | 516 | 1,565 | 429 |
| A quarter or more | 27.4% | 4.86 | 438 | 313 | 53 |
<!-- /generated -->

An organizer with a list should tag the campaign link with an email UTM before sending, so the send shows up in this report. Gleam's export cannot see a send with no tag on the link.

## Mail clients: the wider view

Extracted: clicks whose traffic source is a mail client or a webmail page, which catches a send whether or not the link was tagged. 13,527,852 Impressions, 0.83% of all Impressions in the dataset, across 114,965 campaigns and 18,733 businesses. A mail-client source says the click came out of an email, never which email, so a link a friend forwarded counts the same as a newsletter.

| Client | Share of mail-client Impressions | Campaigns | Businesses |
|---|---|---|---|
| Gmail app | 82.7% | 104,961 | 17,468 |
| Gmail web | 7.6% | 66,268 | 12,886 |
| ISP or carrier webmail | 5.7% | 7,130 | 1,938 |
| AOL, GMX, Mail.ru and other free webmail | 1.6% | 11,583 | 2,616 |
| Outlook | 1.0% | 20,277 | 4,948 |
| provider hosted campaign page | 0.9% | 4,476 | 935 |
| Yahoo Mail | 0.4% | 6,715 | 2,215 |

Gmail's Android app alone carries 83% of it. Campaign and business counts are the largest single host in each group, never a sum, because one campaign reaches several.

Mail arrives later than everything else. On the start day and the six days after it, mail's share of its own total sits below traffic's share of its own (13.7% against 14.2% on day one, 6.3% against 7.9% on day three). From day eight it sits above, and stays above for the rest of the run. A list is what keeps a campaign moving after the launch push stops.

| Days a campaign received mail clicks | Campaigns | Businesses | Entrants | Conversion Rate |
|---|---|---|---|---|
| 1 day | 22,679 | 6,749 | 378 | 32% |
| 2 to 3 days | 24,817 | 7,487 | 515 | 29% |
| 4 to 7 days | 27,768 | 7,867 | 782 | 27% |
| 8 days or more | 39,445 | 8,682 | 1,489 | 20% |

Read that as who mails, never as what mailing does. Campaigns collecting mail clicks on eight days or more are four times the size of one-day campaigns and see a lower Conversion Rate, which is what size does to it everywhere in this data.

<!-- generated:pp2_mail_industry -->
| Industry | Campaigns | Businesses | Share of campaigns where mail is a tenth of traffic |
|---|---|---|---|
| Local services | 910 | 181 | 3.4% |
<!-- /generated -->

The published tier comparison is limited to Business, Pro and Hobby. Other tiers are withheld to protect small residual groups. By organizer country, Italy leads the share of campaigns drawing a tenth of traffic from mail, ahead of Romania, Germany and the United States.

<!-- generated:pp2_mail_tier -->
| Tier | Typical % of Impressions from mail clients |
|---|---|
| Business | 0.37% |
| Pro | 0.31% |
| Hobby | 0% |
<!-- /generated -->

<!-- generated:pp2_mail_country -->
| Country | Share of campaigns drawing a tenth of traffic from mail | Campaigns |
|---|---|---|
| Italy | 3.1% | 414 |
| Romania | 2.8% | 316 |
| Germany | 2.3% | 3,007 |
| United States | 1.7% | 54,517 |
<!-- /generated -->

Source: `analysis/output/email_traffic.json`, keys `totals`, `by_client`, `day_curve`, `days_mail_landed`, `email_share_by_industry`, `email_share_by_plan_tier`, `email_share_by_country`, built by the analysis behind it.


Source: analysis/output/field_cuts.json (email_traffic_by_provider, outcomes_when_email_over_10pct).

## Share clicks by reward size

Offering a Viral Share action is most of the gap in share clicks, before the reward size is even set.

| Group | Share clicks / Impressions | Share clicks per Entrant | Campaigns | Businesses |
|---|---|---|---|---|
| Offers Viral Share | 20% | 0.93 | 43,131 | 6,948 |
| No share action | 1.0% | 0.07 | 73,052 | 12,616 |

Among campaigns that offer it, the entries attached to a share move the figures much less, and the climb stops short of the top band:

| Share worth | Campaigns | Businesses | Share clicks / Impressions | Share clicks % of Entrants |
|---|---|---|---|---|
| 1 entry | 13,579 | 2,729 | 19% | 84 (0.84) |
| 2 to 4 entries | 8,616 | 1,866 | 21% | 89 (0.89) |
| 5 to 9 entries | 7,181 | 1,895 | 23% | 106 (1.06) |
| 10 or more entries | 13,755 | 2,184 | 19% | 97 (0.97) |

Source: analysis/output/field_cuts.json (share_clicks).

Campaigns offering a referral action recorded more Entries per Entrant at every campaign size, but that gap did not grow with campaign size. Comparing what campaigns with and without a referral action actually produce against what size and referral status predict on their own, the result lands close to that prediction at every campaign size tested, and the table carries how close. A referral action is worth building into the mix, but do not plan on the payoff scaling up just because the base to refer from is bigger.

<!-- generated:pp2_referral_size -->
| Campaign size (Entrants) | Predicted-actual ratio, referral action offered | Predicted-actual ratio, none offered |
|---|---|---|
| 100 to 250 Entrants | 1.07 [8,198 campaigns, 2,446 businesses] | 1.04 [24,823 campaigns, 7,629 businesses] |
| 250 to 500 | 1.14 [9,654 campaigns, 2,340 businesses] | 0.95 [15,971 campaigns, 4,602 businesses] |
| 500 to 1,000 | 0.98 [9,194 campaigns, 2,494 businesses] | 0.98 [12,595 campaigns, 3,457 businesses] |
| 1,000 to 2,500 | 0.89 [9,110 campaigns, 2,664 businesses] | 1.01 [10,864 campaigns, 2,858 businesses] |
| 2,500 to 10,000 | 0.88 [5,774 campaigns, 1,649 businesses] | 1.03 [6,929 campaigns, 1,657 businesses] |
| 10,000 or more | 0.92 [1,229 campaigns, 395 businesses] | 1.08 [1,942 campaigns, 436 businesses] |
<!-- /generated -->

Source: analysis/output/success_profiles.json (interactions.referral_action_by_campaign_size).

## What a shared link is worth

Extracted, viral_click_conversion_excluding_crypto: with finance and crypto organizers removed, median share-click conversion is 13.9%, on 42,512 campaigns from 6,919 businesses. This measures the share of Viral Share clicks that end in a referred entry within each campaign. The same campaigns have a median overall Conversion Rate of 21.8%, including referral traffic. Non-referral conversion is not available in this cut, so these figures cannot establish whether referral visitors convert worse than other visitors.

Crypto organizers are now out of the population before any cut runs, so `viral_click_conversion` and `viral_click_conversion_excluding_crypto` hold the same campaigns and report the same 13.9%. The gap those two keys used to describe has gone, and either one gives the figure to plan an ordinary business against.

| Referral-rate figure | Value | Campaigns | Businesses |
|---|---|---|---|
| Whole population | 13.9% | 42,512 | 6,919 |
| Spread on that figure | lower quarter 7.5%, upper quarter 22.9% | - | - |

| By industry | Referrals per click | Campaigns | Businesses |
|---|---|---|---|
| Creator or influencer | 28.9% | 1,008 | 133 |
| Travel and events | 16.4% | 1,755 | 326 |
| Gaming | 14.4% | 6,419 | 1,040 |
| Baby and kids | 14.4% | 642 | 78 |
| Software and SaaS | 14.0% | 682 | 262 |
| Marketing agency | 13.6% | 344 | 65 |
| Toys and collectibles | 13.2% | 1,936 | 283 |
| Automotive | 13.0% | 1,289 | 174 |
| Pets | 11.3% | 648 | 106 |

Creator and influencer campaigns sit well clear of every other industry on this table, because sharing a link is already what that audience does. Finance and crypto has dropped out of the cut altogether, since those organizers are excluded from the population. Pets, automotive and toys and collectibles sit at the bottom.

By share worth, the referral rate steps up off the worth-1 band and then flattens, and clicks per Entrant barely move across the rest. Read a higher worth as a small lift at best, and remember the businesses that set a high worth differ from the ones that do not.

| Share worth | Referrals per click | Clicks per Entrant |
|---|---|---|
| 1 to 9 | 11.9% to 15.1% | 0.86 at worth 1 |
| 10 or more | 15.1% [13,665 campaigns, 2,183 businesses] | 0.98 |

Clicks are a subset of Impressions on 99% of campaigns. Where clicks exceed the campaign's own Impression count, on 288 of the 42,512 campaigns measured, treat it as a fraud tell worth checking before crediting a shared link's reach.

Source: analysis/output/field_cuts.json (viral_click_conversion, viral_click_conversion_excluding_crypto, viral_click_conversion_by_worth, viral_click_conversion_by_industry).

## Referral rate by organizer type and channel size

Referrals % of Entrants fall as a YouTube host channel grows. A smaller channel's audience sits closer to the creator and shares more per person, a large channel's audience treats one giveaway post as one post among many.

<!-- generated:pp2_youtube_referrals -->
| YouTube subscribers | Referrals % of Entrants | Campaigns | Businesses |
|---|---|---|---|
| Under 1,000 | 24 (0.24 per Entrant) | 709 | 203 |
| 1 million or more | 8 (0.08 per Entrant) | 817 | 96 |
<!-- /generated -->

Software businesses and B2B audiences sit well above consumer brands and retailers on referrals % of Entrants. That fits an audience used to forwarding a link at work.

<!-- generated:pp2_btype_referrals -->
| Business type | Referrals % of Entrants | Campaigns | Businesses |
|---|---|---|---|
| Software | 62 (0.62 per Entrant) | 30,426 | 5,133 |
| B2B | 44 (0.44 per Entrant) | 12,728 | 1,779 |
| Brand | 14 (0.14 per Entrant) | 31,193 | 5,708 |
| Retailer | 11 (0.11 per Entrant) | 21,733 | 2,266 |
<!-- /generated -->

Source: analysis/output/industries.json (by_youtube_subscribers, by_business_type, by_audience).

## Traffic mix by country

The available country row describes the United States. Other country comparisons are withheld to protect small residual groups. Choose promotion channels from your own audience evidence.

<!-- generated:pp2_country_traffic -->
| Country | YouTube share of Impressions | Gleam-hosted landing share | Campaigns | Businesses |
|---|---|---|---|---|
| United States | 2.5% | 43% | 57,193 | 8,708 |
<!-- /generated -->

Source: analysis/output/indicators.json (referrer_mix_by_country, landing_kind_by_country).

## Profile prep, the day before launch

- Bio link goes to the entry page for the whole run. For Instagram feed posts and TikTok captions that use this route, say "link in bio". For an Instagram Story with an entry-page link sticker, say "tap the link sticker". Match each call to action to the route available in its scheduled placement.
- Pin the launch post. On Instagram add a Giveaway story highlight and keep the link sticker story in it.
- Bio line names the giveaway and the close date while it runs, then goes back to normal the day after.
- Turn on comment and DM notifications for the accounts that will post. Questions land in the first hour.
- Post the eligibility line and the close time in the caption of every giveaway post, so a screenshot still carries them.

## Feed balance

Giveaway posts are promotional content. Keep the rest of the feed running through the run: the giveaway is one post per push per channel plus stories, and the educational and behind-the-scenes posts continue on their usual days. A feed that turns into giveaway reminders for two weeks loses the followers the giveaway was meant to win.

## Hashtags and who they bring

As practical advice, start with the brand's own tags and one or two for the product category. Assess generic giveaway tags (#giveaway, #win, #competition) and contest directories for audience fit before adding them. The available comparisons describe whole-campaign outcomes and do not establish subscriber quality or signup rates from those sources. Where attribution is available, measure signups from each source against the campaign's objective before deciding whether to keep using it.

## Per channel

| Channel | Launch format | Mid-run format | Last call | Notes |
|---|---|---|---|---|
| Instagram | Carousel: Prize hero, how to enter, eligibility. Link in bio and in a story with the link sticker | Reel of the Prize in use, story poll about the Prize | Countdown sticker story, feed post with "closes tonight" and the time zone | Pin the launch post. Stories every second day. Use a Collab post for the partner push so it sits on both feeds. Do not require tagging in comments |
| TikTok | Short video: Prize reveal in the first second, entry steps on screen, link in bio | Creator or staff video using the Prize | Video with the close time on screen | Comments are where questions land, answer them |
| X | Post with the Prize image and link. Pin it | Quote-post with a detail about the Prize | Post with the close time, reply thread with FAQ | Ask for a repost as the share action, one account per person |
| Facebook | Page post with link, share to relevant groups you admin | Photo post, live if the Prize suits it | Event-style reminder post | Do not require timeline shares or friend tags to enter |
| YouTube | Mention at the top of the next video plus a pinned comment with the link, community post | Short showing the Prize | Community post with the close time | State that YouTube is not a sponsor in the rules |
| LinkedIn (B2B) | Check current policy before recommending any announcement | Same policy check | Same policy check | Exclude LinkedIn entry mechanics, including comment, follow and repost requirements or bonus Entries. Check LinkedIn's current Professional Community Policies (https://www.linkedin.com/legal/professional-community-policies) before any announcement linking to a giveaway elsewhere, and do not assume such an announcement is permitted |
| Discord or Telegram | Announcement channel pin, role ping once | Reminder with a screenshot of entries so far | Final ping | One ping per push, no more |
| Email | Launch email | Mid email | Last call email, then Winners | See the email sequence reference |
| Website | Banner or bar with the deadline, entry page linked from the main nav for the run | Update the banner copy at the mid push | "Closes tonight" banner | Remove everything the day after the close |
| Store (Shopify and similar) | Announcement bar with the deadline, a giveaway page in the nav (the Gleam Shopify app creates one), a card on the collection the Prize came from | Order confirmation and thank-you page mention the giveaway, packing slip insert with the entry link | Announcement bar copy changes to the close time | Sync Entrants to the customer list with the campaign tag, so the store's own email flows can pick them up |

## Post template

Prize first, then the ask, then the deadline, then eligibility.

"Win [Prize] (worth [value]). To enter: [confirmed required action or actions]. Closes [date, time, time zone]. Open to [eligibility]. [Placement-specific entry route]."

Use "link in bio" for a feed caption using the profile link, "tap the link sticker" for a Story with that sticker, or the working entry link where the placement supports it. Apply the channel restrictions above before using this template.

Include only confirmed requirements in the entry sentence. For a campaign with one required action, name only that action. Put any optional action in a separate sentence, explicitly labelled optional, with its confirmed reward: "Optional: [action] for [confirmed additional Entries]." Omit unconfirmed rewards.

Add a referral line only when the referral action is configured and the reward and qualifying event are confirmed: "Earn [confirmed reward] when [exact qualifying event], using [working referral link]." Omit this line when those facts are absent or unconfirmed.

Swap the first line at each push: mid run leads with a detail ("The [Prize] arrives in [feature]"), last call leads with the time ("Closes in 24 hours").

## When almost nobody has entered (advice)

Common practice, nothing in the data covers it. Use each source's stated population. The email-traffic source
specifies location-row coverage and reports no Entrant floor. There is no base rate here for how often a quiet
campaign is broken and how often it is simply unseen. Work through it in this order, because the cheap checks
come first.

- Open the entry link yourself, signed out, on a phone. A campaign that will not load, or that asks for a login
  nobody has, records no Entrants and looks identical to one nobody wanted.
- Check the link in every place it was posted. A bio link pointing at last month's campaign is the commonest
  version of this.
- Complete one entry end to end and watch it appear in the report. That tells you the page, the Actions and the
  recording all work.
- Ask what audience the campaign has actually reached: followers, list size, the traffic the posts got. A
  small reach caps the result, so count how many people actually saw the campaign before judging its
  Entrants. No amount of promotion advice substitutes for reach that was never there.
- Then treat it as reach. The schedule above still applies over whatever days are left, with the last call
  brought forward so it lands before the close.
- Say which of these you checked and what it showed. A reader who is told to "promote harder" without a link
  test is being asked to spend money on a page that may be broken.

## Comments and DMs during the run

Answer in the first hour where you can, and pin one comment with the three questions everyone asks.

| Situation | Reply | Rule |
|---|---|---|
| "How do I enter" | "[Placement-specific entry route]. To enter: [confirmed required action or actions]. Closes [date, time zone]." | Match the available link route and confirmed entry requirements for that post |
| "Is this open in [country]" | "Open to [eligibility]. Sorry if that leaves you out this time." | Quote the eligibility line, never widen it in a comment |
| "Is this real" | "It is. Terms and the draw method are on the entry page: [link]." | Link, no argument |
| "I entered but got no confirmation" | "Check the address you used and the spam folder. If it is still missing, DM us the email and we will check the entry list." | Never post someone's email or entry details publicly |
| Fake account announcing Winners or DMing Entrants for a fee | Post and pin: "We contact Winners only from @[official handle] and [official email]. We never ask for payment, card details or a password. Report and block [fake handle]." Name the official handle, not "this account", the warning gets screenshotted and shared past the post it started on | Report the account to the platform, screenshot it, tell the Winner-communications step so the notification names the official account |
| "Rigged" or "same people always win" | "The draw is [method] from [N] valid entries and the record is at [link]. Happy to walk you through it." | Answer once with the record, then stop. The Winner-communications skill has the dispute reply |
| Abuse or spam | Hide or delete, block on repeat | Never delete questions or complaints, only abuse |
| Partner or creator asks to be tagged | Tag them in the next push and the Winner post | Keep to the brief |

Entrants who comment before they enter are the ones to answer first. A reply that lands while they are still on the post converts. Keep the thread going: a question answered publicly saves ten DMs.

## What the next channel is worth (extracted)

Reach pays, and it pays less each time. Splitting one size band into five equal groups by Impressions, the
campaigns with nearly six times the traffic drew about 45% more Entrants, and the pattern holds in every band
(5.5 to 10.7 times the Impressions for 1.22 to 3.17 times the Entrants). In the 10,000-plus band, the one with no
ceiling on Entrant counts, 10.7 times the reach goes with 3.17 times the crowd, close to the square root of the
reach multiple.

Plan against that curve. Doubling the traffic in front of the page goes with roughly 40% more Entrants, so the
second channel is worth less than the first and the fifth is worth less again. Where two pushes cost the same
effort, the one reaching a warmer audience wins on Conversion Rate, and the one reaching a bigger cold audience
wins on raw Entrants by less than its size suggests.

How many traffic sources a campaign records is not in this cut, so nothing here can say whether a wide channel
mix produced the reach or followed from it. Nothing here can price a channel a business has not tried.

Source: `analysis/output/reach_returns.json`.

## Where the traffic lands across the run

Extracted: typical share of a campaign's Impressions landing on each day since start, for campaigns run about 7, 14 and 30 days. Entries are not in this cut, so it tracks visits per day and never who entered that day. Write "traffic" or "visits" in the answer. A measured answer wrote "Entries flatten from day 10 to day 13" off this table, which promises the reader something the cut cannot see.

<!-- generated:pp2_curve -->
| Duration | Day one | Peak day | Trough | Close |
|---|---|---|---|---|
| 7 days | 20.6% (23,879 campaigns, 5,739 businesses) | Day 1 | none, declines straight through | Day 6, 7.8% (23,755 campaigns, 5,724 businesses) |
| 14 days | 10.7% (15,997 campaigns, 4,817 businesses) | Day 1 | Day 10, 4.4% (17,443 campaigns, 5,176 businesses) | Day 13, 5.1% (15,919 campaigns, 4,785 businesses) |
| 30 days | 4.0% (20,348 campaigns, 5,030 businesses) | Day 2, 4.9% (21,783 campaigns, 5,270 businesses) | Day 18, 2.2% (23,255 campaigns, 5,458 businesses) | Day 29, 3.0% (19,596 campaigns, 4,704 businesses) |
<!-- /generated -->

Day one carries a fifth of a week-long campaign's total traffic, a tenth of a two-week one's and a twenty-fifth of a month-long one's. Quote the row for the run the reader is actually planning. A week-long run falls every day straight through to the close, with no recovery. A two-week or month-long run falls hard through the first week or two, flattens into a long stretch at under half the peak day's share, and picks up a little in the final days. Neither close gets back to the launch share.

<!-- generated:pp2_curve_read -->
| Duration | Launch peak | Quiet-stretch share of launch | Late-run uptick |
|---|---|---|---|
| 7 days | Day 1 | none, declines straight through | none |
| 14 days | Day 1 | 41% of the peak day | day 13 0.8 points above day 10 |
| 30 days | Days 1 to 2 (peak on day 2) | 45% of the peak day | day 29 0.8 points above day 18 |
<!-- /generated -->

Source: `analysis/output/prize_timing_cuts.json` (`impression_curve_by_duration`).

Share clicks follow a tighter curve than Impressions: on a week-long run they peak on day 1 and day 2, then decay every day to the close. Impressions and shares both crest in the first two days, so the launch and second push carry both.

| Day | Share of total share clicks | Campaigns | Businesses |
|---|---|---|---|
| Day 1 | 22% | 6,807 | 1,774 |
| Day 2 | 17% | 7,739 | 1,967 |

Source: analysis/output/field_cuts.json (share_click_curve_by_duration).

## Turning the curve into pushes

- **Launch, day one.** The biggest traffic day on a week-long or two-week run, and within a point of the peak on a month-long one. Land the launch post and email inside the same two hours the campaign goes live.
- **Second push, while traffic is still falling.** A week-long run never settles into a flat stretch, so the second push slows a decline that is already underway.
- **The quiet middle.** Traffic sits at under half the peak day's share and barely moves day to day. A partner or creator post belongs here, to give the flat stretch something to move on.
- **The last days.** On a two-week or month-long run, traffic on the close reads a little above the quiet middle by itself. On a week-long run it does not, the decline runs straight to the end. In every case the close stays a fraction of the launch share, so the last-call post, email and countdown sticker go out on schedule regardless. The close is not going to fill itself in.

Read the row for the run length the reader actually has, and quote its days as days since launch before converting
them to dates. A measured answer turned days 10 to 13 of a fourteen-day run into 6 to 12 October on a run that
started on 1 October, which is the four days on either side of the right answer.

| Duration | Second push | Quiet middle |
|---|---|---|
| 7 days | day 3 to 4 | none, declines straight through |
| 14 days | day 4 to 6 | days 10 to 13 |
| 30 days | day 5 to 8 | days 15 to 26 |

## Fourteen-day calendar

A worked example for a two-week run on Instagram plus email, with a partner on day 8. Shift the dates to the timing plan. The partner's own day wins: day 8 is where a partner post does the most good on the curve above, and the quiet-middle table puts that window at days 10 to 13, but a partner who can only post on day 6 posts on day 6.

| Day | Push | Feed | Story | Email | Other |
|---|---|---|---|---|---|
| -1 | Prep | Pin ready | Highlight created | Segment built, link tested | Bio link set, partner brief sent |
| 1 | Launch | Carousel: Prize, how to enter, eligibility | Link sticker, "just launched" | Launch email within two hours | Pin the post, answer comments for the first hour |
| 2 | - | Usual content | Poll about the Prize | - | - |
| 3 | - | Usual content | Link sticker reminder | - | - |
| 5 | Mid | Reel of the Prize in use | Entries so far, confirmed referral reward if configured | Mid email | - |
| 6 | - | Usual content | Question sticker, answer three | - | - |
| 8 | Partner | Collab post with the partner (appears on both feeds) | Repost the partner's story | - | Partner posts in their morning |
| 10 | - | Usual content | Link sticker reminder | - | - |
| 12 | - | Usual content | Countdown sticker set to the close | - | - |
| 13 | Last call | "Closes tomorrow" feed post with the time zone | Countdown reminder | Last call email | - |
| 14 | Close | - | "Closing in [hours]" story at the audience's peak hour | - | Take the banner down the day after |
| 15 to 21 | Draw | Winner post, once the Winner has agreed | Winner story, tag the partner | Winners email | Bio back to normal, highlight stays |

Three feed posts and one collab in fourteen days, with stories carrying the reminders. The feed stays a feed.

## Social figures to record

Write these down at launch, at close and 30 days after close, from the platform's own insights, and hand them to the results review:

- Follower count on each promoted channel. Report the difference as net follower change. New followers and unrelated unfollows affect the total, so it cannot show how many giveaway followers stayed. Measure retention only where records track that specific group over time.
- Reach and saves on the launch post against your typical post from the previous month.
- Link clicks from the bio link and from each story sticker, and the UTM totals from the entry page.
- Comments answered and the typical time to first reply on the launch post.

None of this is in the campaign dataset, which counts entries and Impressions on the entry page only. The comparison is against your own previous posts.

## Partner and creator brief

One page, sent a week before their date.

- What: one post and one story on [date], plus a repost of the Winner announcement.
- Copy and assets: attached, editable, keep the link and the eligibility line.
- Link: the entry page with their UTM or code, so they can see what they drove.
- What they get: named as the Prize partner on the entry page and in every email, aggregate campaign results, product. Share Entrant-level data only where the recorded notice and permissions cover sharing with that partner and its intended use, and the platform terms permit it.
- Timing: post between [hours] in their audience's time zone. Tell us when it is live.
- Rules: no purchase claims, no changes to eligibility, disclose the partnership where required.

## Paid

- Schedule organic first. Paid amplifies a plan, it does not replace one.
- Boost the launch post only, to people who look like the email list or the channel's engaged followers, and to the regions in the eligibility line.
- Cap the spend at the cost of one more Winner. If the campaign is producing leads, cost per Entrant against the value of a lead decides whether to add more.
- Never run paid to audiences outside eligibility. Wasted spend and angry Entrants.
- Add a UTM to every link you post, on every channel, paid and organic. Source is the network, medium is the format (post, story, email, partner), campaign is the giveaway name. Without it the results review has traffic sources and no way to separate your posts from a partner's.
- A store with Shopify Audiences or the equivalent can build the retargeting audience from the tagged customer segment without an export only where the recorded notice, permissions and platform terms cover that advertising use.
- After the close, use only Entrants whose recorded notice and permissions cover custom-audience advertising, and check the network's terms before uploading. Entry-only contact details do not establish that permission. Exclude the Winners and anyone already a customer, then run the launch offer to the rest. Practice, with no dataset figure behind it: the dataset holds no purchase data.
- Build a lookalike audience for the next campaign's launch push only from Entrants whose recorded notice and permissions cover that use, where the network's terms permit it. Giveaway entry alone does not establish permission. Judge that prospecting audience on cost per lead against your usual sources.

## Afterwards

- Announce Winners on every channel used to promote, within a week.
- Thank everyone and give them something small: a code, a guide, early access to the next drop.
- Ask Winners for a photo and use it only with permission.
- Save the creative that got the most entries for every 100 Impressions for the next campaign, and tag the new subscribers with the campaign name so their behaviour can be compared later.
- The non-Winner series and the list hygiene pass that follows it belong to giveaway-winner-communications. Point the plan there and leave the copy to that skill.
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 196kUpdated Claude CodeCodexVerifiedInstall