据 AlphaMoat 统计,截至 2026-09,TikTok Ads 累计下载 4,465 次、获 13 个星标,在全部 5,078 个 AI 技能中排名第 2,165 位,在「商业运营」分类(297 个)中排名第 119 位。
| 平台 | 累计安装 | 累计下载 | 星标 | 评论 |
|---|---|---|---|---|
| ClawHub ↗ | 138 | 4,125 | 13 | - |
| SkillHub ↗ | 152 | 4,465 | 13 | — |
各平台的计数口径不同,统计时间不一致。
主文涵盖视频开场与字幕、Spark Ads、受众定位、出价与学习阶段、广告组结构、转化追踪、创意测试和扩量。文中提供了多项经验阈值与策略建议,应视为待验证的操作参考,结合当前平台规则、目标和账户数据测试,不保证投放效果。 这是指导,不是账户访问或投放执行。使用创作者帖文的 Spark Ads 需创作者授权;花费预算或改广告前核对当前规则与权限。
资料发布于 2026-10-27
以下是根据技能文档生成的示例,不代表实际运行结果。
2026-09 起统计方法调整:同名技能按作者分别统计,并覆盖全部技能。图中虚线为调整前的月份,与之后的数据不直接比较,也不计算跨越调整的增长。
| 月份 | 累计下载 | 环比 | 星标 |
|---|---|---|---|
| 2026-09 | 4,465 | — | 13 |
| 2026-08 | 4,144 | +6.1% | 13 |
| 2026-07 | 3,906 | +8.0% | 13 |
| 2026-06 | 3,618 | +5.0% | 13 |
| 2026-05 | 3,447 | — | 12 |
| 2026-03 | 2,599 | — | 12 |
以下是作者发布的原始文件,按纯文本显示,本站未做修改或审核;其中的链接不可点击,命令请核实后再运行。抓取于 2026-10-23。
--- name: TikTok Ads description: TikTok advertising strategy, creative best practices, Spark Ads, and platform-specific optimization metadata: category: marketing skills: ["tiktok", "ads", "social-ads", "video-ads", "short-form"] --- ## Creative Rules
- First 1-3 seconds determine everything — if viewers don't stop scrolling immediately, the rest doesn't matter - Native content outperforms ads — shoot vertical (9:16), use trending sounds, avoid polished brand look - Hook patterns that work: pattern interrupt, bold claim, direct question, unexpected visual, POV setup - Text overlays are mandatory — 80%+ of users watch with sound off, burned-in captions aren't optional - Refresh creative every 7-14 days — fatigue hits faster than any other platform, have pipeline ready ## Spark Ads - Spark Ads use organic posts as ads — keeps comments, likes, shares visible and accumulating - Organic engagement on Spark Ads compounds — viral potential unlike standard ads - Creator handles can be swapped — run ads from your brand handle or creator's handle - Require creator authorization code — valid for 7, 30, or 365 days, get longer duration upfront - Boosting already-performing organic posts is safer than untested creative — data before dollars ## Targeting Mistakes - Interest targeting is less accurate than Meta — TikTok's algorithm learns faster from creative signals - Start broad, let algorithm optimize — overtargeting restricts the machine learning that makes TikTok work - Custom Audiences need 1000+ users minimum — smaller lists won't deliver - Lookalike percentages are different scale — 1% on TikTok is broader than 1% on Meta - Behavior targeting (video completion, engagement) often outperforms interest targeting ## Bidding Strategy - Learning phase needs 50 conversions in 7 days — budget must support this or campaign never optimizes - Cost Cap is stabler than Bid Cap for beginners — limits volatility while algorithm learns - Lowest Cost burns budget fast on launch — switch to Cost Cap after gathering initial data - Minimum daily budget per ad group: $20-50 depending on objective — below this, delivery is inconsistent - Don't edit campaigns during learning phase — resets progress, wait until 50 conversions ## Campaign Structure - One campaign per objective — mixing awareness and conversion in same campaign confuses optimization - 3-5 ad groups per campaign is optimal — enough for testing, not so many that budget fragments - 3-5 creatives per ad group — algorithm needs options to test but too many dilutes learnings - Separate iOS and Android campaigns post-ATT — iOS tracking is limited, don't let it pollute Android data - Catalog ads for e-commerce — dynamic product ads outperform static for retargeting ## Platform Mechanics - Content is served via For You algorithm, not follower graph — follower count matters less than Meta - Hashtag challenges are paid products — organic hashtags have minimal targeting value in ads - Comments section is public and active — negative comments kill CTR, moderate or disable if problematic - Video completion rate is primary signal — optimize for watch time, not just clicks - TikTok pixel has same iOS limitations as Meta — use Events API for server-side tracking ## Creative Testing - Test hooks independently from body — first 3 seconds are separate variable from rest of video - A/B test in creative tools, not ad platform — TikTok's native split testing is limited - User-generated style beats studio quality — imperfect > polished on this platform - Test creator vs brand voice — same script performed differently converts at different rates - Winners emerge fast — if no signal in 3-5 days, creative likely won't work ## Conversion Tracking - Standard pixel misses 40%+ of iOS conversions — implement Events API for accuracy - View-through window defaults to 7 days — conservative window (1 day) shows truer performance - Lead gen forms have high drop-off — keep fields to absolute minimum (name, email, phone) - TikTok Shop has native tracking — if selling via Shop, conversion data is more complete - Attribution is last-click by default — view-through conversions are modeled, not deterministic ## Scaling Patterns - Increase budget 20-30% max per day — larger jumps reset learning phase - Horizontal scaling: duplicate ad groups with different targeting — avoids audience overlap issues - Creative is the scaling lever — same audience with new creative > same creative with forced reach - CPMs spike during holidays and elections — plan 2-3x budget for Q4 and avoid political seasons if possible - International markets have lower CPMs — test UK, Canada, Australia before saturating US ## Common Failures - Repurposing Instagram Reels directly — TikTok users detect non-native content instantly - Logo in first 3 seconds — viewers skip ads, not content. Logo at end, not beginning - Long intros or brand statements — no one waits, hook must be immediate - Ignoring comments — negative comments are social proof against you, moderate actively - Expecting Meta-level tracking accuracy — TikTok attribution is less mature, expect discrepancies
为 AI agent 提供电话号码并代办电话事务的技能。
通过 Trello REST API 管理看板、列表与卡片。
通过知了标讯接口查询和分析招投标相关信息。
产品经理工作台:从需求分析到 PRD 与迭代规划。
面向日历管理与会议排期的技能。
提供市场趋势、竞品与用户行为分析的市场分析服务。
按数据分析筛电商爆款,覆盖多个平台的品类。
通过 Maton 托管 OAuth 访问 Salesforce CRM 的技能。
把用户纠正与自我复盘沉淀成分层记忆、并按使用情况升降级的技能。
在保留公式、格式与工作簿结构的前提下创建、检查并修改 Excel 文件的技能。
处理 Word 与 DOCX 文档的作业规范,重点是改完之后样式、编号与版式不跑偏。
把数据库、表格与导出的原始数据变成可决策结论的数据分析技能。
改 PPT 之前先看清结构的一套规范,目标是改完还能干净渲染。
数据月份:2026-09 · 下载/星标/安装数聚合自公开技能注册库(ClawHub、SkillHub),按月更新。口径说明见数据方法。