如何读懂指标
«观看»在不同平台上是不同的事件：有的地方是看了半秒，有的地方是播放十秒，有的地方是每天一个独立用户。在平台之间比较数字时，很容易比较的其实是不同的量。
本页是一部指标词典，用平台自己的话写成，每条定义都附有指向文件的链接。
什么算作观看
Facebook. 主页视频观看（page_video_views）原文：«The number of times your Page's videos played for at least 3 seconds, or for nearly their total length if they're shorter than 3 seconds»——阈值为 3 秒，短于 3 秒的视频则须播放接近全长。内容展示（page_impressions）原文：«The number of times any content from your Page or about your Page entered a person's screen»——没有时间阈值 来源, 已核查 2026-08-07
Instagram. views 指标原文：«The number of times your content was played or displayed. Content includes reels, posts, stories»——即观看没有时长门槛：播放或在屏幕上显示即可。原有的 impressions 指标原文：«The number of times your posts, stories, reels, videos and live videos were on screen, including in ads»——自 v22.0 起弃用，21.04.2025 之后的所有版本也弃用 来源, 已核查 2026-08-07
LinkedIn. 视频观看（videoViews）原文：«A video ad playing for at least 2 continuous seconds 50% in-view, or a click on the CTA, whichever comes first. An interaction with the video (like going to fullscreen mode) doesn't count as a view»——视频 50% 处于可视区域时连续播放 2 秒，或点击 CTA；切换到全屏不算作观看。完整观看（videoCompletions）原文：«The count of video ads that… 来源, 已核查 2026-08-07
PeerTube. 播放 10 秒后计为一次播放；若视频短于 10 秒，则在播放到 50 % 时长后计入 来源, 已核查 2026-07-28
Pinterest. 视频观看，原文：«Video views — The amount of views for at least 2 seconds with 50% of video in view»（视频 50% 处于可视区域时不少于 2 秒；在 API 中该指标名为 VIDEO_MRC_VIEW）。展示，原文：«Impressions — The number of times your Pins or ads were on screen» 来源, 已核查 2026-08-07
Roblox. 视频展示计数条件：注视不少于 0.5 秒、占屏不少于 1.5%、角度不超过 55 度、像素可见度不低于 50%；图片的注视门槛为 1 秒；click-to-play 视频须观看不少于 15 秒 来源, 已核查 2026-07-28
RUTUBE. RUTUBE 未公布观看门槛；在合作伙伴计划要约的术语表中，目标行为被定义为（原文）««PLAYERSTART»（用户与播放器交互时在播放器中开始播放内容）»，即记录播放开始；各项指标«依据 RUTUBE 统计系统的数据»记录 来源, 已核查 2026-08-07
Snapchat. 自然流量：Spotlight 的观看以开始播放为准，没有门槛，原文：«Views — The number of times viewers started watching your Spotlight»；公开故事的原文：«Snap Views — The sum of views across each individual Snap in your Story»。广告：2-Second Video Views，原文：«The total number of impressions that meet the qualifying video view criteria of at least two seconds of… 来源, 已核查 2026-08-07
Telegram. 频道的每条帖子都有观看计数器；计数中包含转发副本的观看，原文：«Views from forwarded copies of your messages are also included in the total count»。该计数器不按人去重：大约 4 天后观看记录会被遗忘，再次打开帖子会重新计数。数字是近似值，原文：«Please note that these numbers are approximate — we don't want to keep a record for everything you've ever viewed» 来源, 已核查 2026-08-07
Threads. 帖子观看（views）原文：«The number of times your post was played or displayed»——播放或展示，没有时间门槛；该指标标注为«in development»。主页浏览是单独的用户级 views 指标，原文：«The number of times your profile was viewed» 来源, 已核查 2026-08-07
TikTok. Video Views 原文：«Number of times your video started to play. For each video impression, plays are counted separately and replays are excluded»——观看从播放开始起计，没有最短时长要求，重播不计。另有 2-Second Video Views，原文：«Number of times your video starts playing for at least 2 seconds in an impression session. Replays are excluded»；6-Second… 来源, 已核查 2026-08-07
Twitch. 在游戏分析中，观看不去重，原文：«Live Views — Number of times any stream of this game was viewed on Twitch, live. This includes live streams and Premieres. If someone watches a game 5 separate times in a day, that counts as 5 views»；对于扩展，观看指屏幕上部分内容的可见性，原文：«Number of times that 75% or more of the extension iframe was visible in any… 来源, 已核查 2026-08-07
vc.ru. 广告块不少于 50% 出现在屏幕上持续一秒及以上即计为展示；默认同一用户每天看到同一广告不超过 5 次 来源, 已核查 2026-07-28
X (Twitter). 广告分析中的视频观看采用 MRC 标准，原文：«The video_total_views metric within the VIDEO metrics group will report on any views which are at least 50% in-view for 2 seconds, as per the MRC standard»（播放器不少于 50% 处于可视区域时不少于 2 秒）；原有定义保留为单独的指标 video_3s100pct_views（播放器 100% 可见不少于 3 秒）。对于自然的 view_count，视频门槛未公布，原文：«Aggregated across all posts… 来源, 已核查 2026-08-07
YouTube. 普通视频的观看门槛未公布。对于 Shorts，自 31.03.2025 起原文为：«Views will count the number of times a Short starts to play or replay, with no minimum watch time requirement»；原有计算方式保留为«Engaged views»指标（原文：显示«how many viewers chose to continue watching your video»），YPP 门槛和广告分成正是使用这一指标。对于广告，观看的原文为：«watches at least 30 seconds of an ad that's… 来源, 已核查 2026-08-07
VKontakte. VK 广告 (VK Reklama)（视频）：«Старт просмотра — количество стартов просмотра видео»（原文；观看开始——视频观看的开始次数）；«Просм. 3 сек — просмотр рекламного видеоролика до 3 секунд»（原文；3 秒观看——广告视频观看至 3 秒）；«Просм. 10 сек — просмотр рекламного видеоролика до 10 секунд»（原文；10 秒观看——广告视频观看至 10 秒）；«Просмотрено 100% — засчитывается при полном… 来源, 已核查 2026-08-07
Dzen. Dzen 区分四个概念，原文：«展示——出版物在 Dzen 信息流中的显示次数»；帖子的«观看»是«阅读了文本或查看了所附图片的用户数量»；视频观看是«观看了视频的用户数量»；«读完——将出版物浏览到底的用户数量»。观看按用户计算（去重），展示按显示次数计算；秒数门槛未公布 来源, 已核查 2026-08-07
Odnoklassniki. Odnoklassniki (OK) 原则上不把展示称为展示，原文：««渲染次数»——内容加载到信息流的次数»和«由于无法断定用户是否看到了帖子，在 OK 的统计中该指标正是称为«渲染次数»，而不是«展示»» 来源, 已核查 2026-08-07
TamTam. 一次浏览量计为一名唯一用户当天对一条消息的一次查看；总计数由每日数值累加而成 来源, 已核查 2026-07-28
什么被称为覆盖人数
Facebook. 主页触达（page_impressions_unique）原文：«The number of people who had any content from your Page or about your Page enter their screen»——按独立人数计，该指标被标注为估算值（estimated）；与展示的区别：展示按每次进入屏幕计数，触达按人只计一次 来源, 已核查 2026-08-07
Instagram. 原文：«The number of unique accounts that have seen your content, at least once, including in ads»——独立账号，包括广告带来的覆盖；该指标为估算值（estimated），支持 time_series 聚合（日/周/28 天） 来源, 已核查 2026-08-07
LinkedIn. approximateMemberReach 原文：«The estimated number of unique member accounts with at least one impression»——至少有一次展示的独立会员账号的估算数；仅提供不超过 92 天的周期。对于页面的自然数据是 uniqueImpressionsCount，原文：«Number of unique impressions»，数据只提供滚动的 12 个月 来源, 已核查 2026-08-07
Pinterest. 覆盖的对应指标是 Total audience，原文：«The total number of people who have seen or engaged with your Pins»；月度覆盖，原文：«Monthly total audience — The total number of people who have seen or engaged with your Pins in a 30 day window up to and including each day»（30 天滚动窗口） 来源, 已核查 2026-08-07
Snapchat. 自然流量：观众是去重的，原文：«Viewers — The number of unique viewers for your Story»（Spotlight 同理）；对于镜头，覆盖有明确定义，原文：«Reach — The total number of users that have played with your Lens(es)»。广告：Paid Reach，原文：«The number of unique paid impressions» 来源, 已核查 2026-08-07
TikTok. 原文：«The number of unique user accounts that were exposed to an ad impression at least once»——独立账号（不是人，也不是设备）；与展示的区别：展示计算«every exposure, even if that unique user account has already been exposed before»。基础帮助文档原文：«Number of unique users who saw ads at least once»。Reach 通常是估算值（因为抽样），确切数字只出现在 Reach & Frequency 广告活动的… 来源, 已核查 2026-08-07
Twitch. 独立观众按设备计算，而不是按人计算，原文：«(Unique Viewers) are collected based on device IDs. At a given moment, a device ID is assigned to a single hardware device. For example, if someone loads an extension or watches a game on 2 devices, that involves 2 device IDs and counts as 2 Unique Viewers» 来源, 已核查 2026-08-07
X (Twitter). X 未公布覆盖指标（reach）；取而代之的是展示（impressions），并明确定义为不去重，原文：«Count of times the post appeared on a user's screen. Not unique—the same user viewing twice counts as two impressions» 来源, 已核查 2026-08-07
YouTube. Reach 标签页建立在展示和独立观众之上。展示（impression）原文：«Impressions are counted if the thumbnail is shown for more than 1 second and at least 50% of the thumbnail is visible on the screen»（只在 YouTube 内部计算：首页、搜索、推荐；外部嵌入、YouTube Kids、Music、隐藏标签页不计）。Unique viewers 原文：«the estimated number of viewers who came to watch your videos over a… 来源, 已核查 2026-08-07
VKontakte. 原文：覆盖 (okhvat) 是«看到了您广告的独立用户数量»；«期末覆盖——从广告系列启动日期到所选周期结束期间看到了您广告的独立用户数量»；频次是展示与覆盖之比。展示：«所选周期内该广告的展示次数»（原文） 来源, 已核查 2026-08-07
Odnoklassniki. 原文：«Охват — число пользователей, которым хотя бы раз был показан контент из группы»；按独立用户计算，并分为两部分：«теперь охват делится на участников группы и людей вне сообщества»（原文） 来源, 已核查 2026-08-07
互动量包含什么
Facebook. page_post_engagements 原文：«The number of times people have engaged with your posts through reactions, comments, shares and more»——心情、评论、转发及其他，按次数计（非独立人数）；平台不定义互动率 来源, 已核查 2026-08-07
Instagram. accounts_engaged 原文：«The number of accounts that have interacted with your content, including in ads»（估算值；互动包括点赞、收藏、评论、分享、回复）。total_interactions 原文：«The total number of post interactions, story interactions, reels interactions, video interactions and live video interactions, including any interactions on boosted… 来源, 已核查 2026-08-07
LinkedIn. 页面的自然互动是相对于展示计算的比例，原文：«Number of organic clicks, likes, comments, and shares over impressions»（点击+点赞+评论+分享 / 展示；不是相对于关注者）。广告中的 totalEngagements 原文：«The count of all user interactions with the ad unit»。likeCount 可能变成负数：对推广帖子的点赞不计为自然互动，而取消点赞却计入（文档原文如此） 来源, 已核查 2026-08-07
Pinterest. 原文：«Engagements — The total number of engagements on your Pins. This includes saves, Pin clicks, outbound clicks, carousel card swipes, secondary creative (collections) clicks»；比率，原文：«Engagement rate — The total engagements with your Pins divided by the total number of times your Pins were seen. Engagements include… 来源, 已核查 2026-08-07
RUTUBE. 行为活跃度汇总为 IMPA，原文：««行为活跃度个体媒体指标»（IMPA/ IMBA）——由 RUTUBE 系统生成的标识符，是合作伙伴平台维度上行为活跃度各项指标与数据的集合»（要约术语表第 1.36 条）；其中也计入«内容单元的看完百分比»、在平台上停留的时间和互动 来源, 已核查 2026-08-07
Snapchat. 没有汇总的«engagement»指标——术语表分别定义各项行为，原文：«Screenshots — The number of times that a viewer has screenshotted your Snap»；«Clicks — The number of times a viewer has clicked or swiped-up to a destination from your Snap (such as driving to a web URL or viewing a Public Profile)»；«Replies — The number of times that a viewer has… 来源, 已核查 2026-08-07
Telegram. 没有统一的互动指标；频道统计中有«shares per post»，原文为«total_sharecount/postcount, for posts posted during the period in consideration»，以及 interactions 图表（按小时的观看和转发）——但没有对互动构成的定义 来源, 已核查 2026-08-07
Threads. 互动按各自独立的指标统计：likes、replies、reposts、quotes。Replies 原文：«The number of replies on the post. When the requested media is a root post, this number includes total replies. If the media is itself a reply, this number includes only direct replies»；在账号层级，replies 只包括«only top-level replies»（原文） 来源, 已核查 2026-08-07
TikTok. 公开定义中没有统一的«вовлечённость»指标。Clicks (all) 原文：«Number of clicks from ads, including social interactions»——点击中包含点赞、分享、评论、关注。Paid likes 原文：«Number of times an ad was liked during the ad impression»。另有«engaged view»——观看广告 6 秒以上（Engaged View-through Attribution） 来源, 已核查 2026-08-07
X (Twitter). 在公开文档中，互动只以公式给出：Engagement Rate = engagements/impressions；行为构成没有说明——engagements 字段的描述是同义反复的（«Total number of engagements»） 来源, 已核查 2026-08-07
YouTube. 原文：«engagement metrics (views, likes, dislikes, and subscriptions) reflect how many times your YouTube video or channel has been interacted with»。只计入经确认的真人行为，原文：«We want to make sure that your metrics are high quality and come from actual humans and not computer programs»；被剔除内容的例子，原文：«Using several devices to watch… 来源, 已核查 2026-08-07
Dzen. 互动只通过留存和时长来体现：留存曲线显示（原文）«вовлеченность аудитории в просмотр видео»——兴趣能保持到哪一段；另外还单独统计«вовлечённое время просмотра»（数据自 2023 年 8 月 1 日起计算）。行为清单和互动公式未公布 来源, 已核查 2026-08-07
Odnoklassniki. 原文：«互动度——进入过群组、打开过内容或留下过反馈的人数»；互动系数，原文：«按最近一周的日均值计算，公式为：(评论 + «классы!» + 分享) * 100 / 成员数量»；另有«反馈——至少有一次评论、点过«Класс!»（赞）或分享过内容的人数»（原文） 来源, 已核查 2026-08-07
为什么平台之间的数字无法比较
- 事件不同。 PeerTube 的十秒和 Roblox 的半秒都算«观看»，但同一段视频，后者的数量总会高出几倍。2. 唯一性不同。 按人计、按会话计和按«每天一人»计，会从同样的流量中得出不同的数字。3. 窗口不同。 一天的覆盖人数、28 天的覆盖人数和«所有时间»的覆盖人数是三个不同的数字，而窗口由平台自己选择。4. 方法学的沉默。 没有定义的地方，比较就是猜测。这样的数字我们照样展示，但不建议拿来比较。
平台不向任何人提供的东西
关于指标，平台沉默的时候多于开口的时候。平台不提供的数字，分析服务给不了，导出也给不了。
Instagram. 按粉丝数量的门槛：«Some metrics are not available on Instagram accounts with fewer than 100 followers»；«follower_count and online_followers metrics are not available on Instagram business or creator accounts with fewer than 100 followers» 来源, 已核查 2026-08-05
Instagram. «You cannot get insights for Facebook Pages»；API«cannot be used to get data for media owned by personal Instagram accounts»；«You can only get insights for a single user at a time»；«Live video Instagram Media can only be read while they are being broadcast» 来源, 已核查 2026-08-05
Instagram. impressions 指标已下线：«deprecated for v22.0 and will be deprecated for all versions on April 21, 2025»。对于2024年7月2日及以后创建的帖子，2025年4月21日之后请求 impressions 会返回错误；替代指标是 views 来源, 已核查 2026-08-05
Instagram. «If insights data you are requesting does not exist or is currently unavailable the API will return an empty data set instead of 0 for individual metrics» 来源, 已核查 2026-08-05
LinkedIn. «This endpoint returns organic statistics only. Sponsored activity is not counted in this endpoint.» 来源, 已核查 2026-08-05
LinkedIn. «Time-bound follower counts are aggregated and not segmented by facet»；«Time-bound follower statistics (queries that include a timeIntervals parameter) do not return demographic breakdown fields. The following fields are unsupported when a time range is… 来源, 已核查 2026-08-05
LinkedIn. «This endpoint no longer returns totalFollowerCounts» 来源, 已核查 2026-08-05
LinkedIn. «The results for any individual facet are limited to the top 100 results. For example, if a page has followers in more than 100 countries/regions, only the top 100 countries/regions will be returned.» 来源, 已核查 2026-08-05
LinkedIn. «Time-bound statistics is not supported for specific share queries»；«The /organizationalEntityShareStatistics endpoint doesn't support pagination»；«Shares with no actions or impressions are not included in the list of elements» 来源, 已核查 2026-08-05
LinkedIn. «The shareCount for the time-bound statistics won't match the lifetime statistics because the time-bound data won't include the count from instant reposts»；关于 likeCount：«This field can become negative when members who liked a sponsored share later unlike it.… 来源, 已核查 2026-08-05
Pinterest. «If Pin was created before 2023-03-20 lifetime metrics will only be available for Video and Idea Pin formats» 来源, 已核查 2026-08-05
Pinterest. «The reporting endpoints covered in this guide are only for organic Pins, any new performance a Pin receives once its been promoted will be reported on in our ad analytic endpoints» 来源, 已核查 2026-08-05
Pinterest. 1天、7天和30天的汇总指标«unique to this method of fetching and not available in other analytics endpoints» 来源, 已核查 2026-08-05
RUTUBE. “如何在 RUTUBE 工作室应用中查看某个视频的收入？目前应用中没有这个功能，但您可以在工作室的网页版中查看分析”；“能否在 RUTUBE 工作室应用的‘概览’标签页中调换或隐藏板块？目前没有这个功能” 来源, 已核查 2026-08-05
RUTUBE. 具有受限查看权限的用户“无法查看‘变现’板块” 来源, 已核查 2026-08-05
Telegram. 统计功能的准入门槛未予公布：«Administrators of channels of a certain size (the exact limit is a server-side config, returned in the can_view_stats flag of channelFull) can invoke stats.getBroadcastStats to get detailed channel statistics» 来源, 已核查 2026-08-05
Telegram. Bot API 手册中没有任何一个获取频道或群组统计的方法：«statistics»一词只出现在管理员权限的描述中——«access supergroup statistics»和«access channel statistics; for channels only» 来源, 已核查 2026-08-05
Telegram. 关于频道统计方法明确写道«Only users can use this method»；在超级群组上调用会得到错误 BROADCAST_REQUIRED——«This method can only be called on a channel»，而没有权限时得到 CHAT_ADMIN_REQUIRED«You must be an admin in this chat to do this» 来源, 已核查 2026-08-05
Threads. 没有2024年春季之前的数据：«The earliest Unix timestamp that can be used is 1712991600»（2024年4月13日）；«User insights are not guaranteed to work before June 1, 2024» 来源, 已核查 2026-08-05
Threads. «Returned metrics do not capture nested replies' metrics»；«An empty array will be returned for REPOST_FACADE posts»；对本身就是回复的帖子，replies 这个指标«includes only direct replies» 来源, 已核查 2026-08-05
Threads. followers_count 和 follower_demographics 这两个指标：«This metric does not support the since and until parameters» 来源, 已核查 2026-08-05
Threads. 文档中说 views 和 shares 这两个指标«in development» 来源, 已核查 2026-08-05
TikTok. Display API 的字段清单中既没有观众人口统计，也没有留存、展示来源和独立观众数——只有 like_count、comment_count、share_count、view_count 来源, 已核查 2026-08-05
TikTok. «Videos returned are for the given user's public TikTok video posts»；«It is possible that the API returns less videos than the max count due to reasons such as videos deleted/marked as private by users etc.» 来源, 已核查 2026-08-05
Twitch. Twitch API 参考手册中没有通过 API 获取自有频道统计（观看、观众、订阅、直播时长）的途径：Analytics 一节只有关于扩展和关于游戏的报告 来源, 已核查 2026-08-05
Twitch. 游戏报告的门槛以数字给出：«a report is available only if the game was broadcast for at least 300 minutes (5 hours) over the time period covered by the report» 来源, 已核查 2026-08-05
Twitch. «Clips data is available via the Get Videos API endpoint»——游戏报告中没有关于切片的数据；«New data fields may be added in new versions of the report» 来源, 已核查 2026-08-05
X (Twitter). «Non-public, organic, and promoted metrics are only available for posts created within the last 30 days.» 来源, 已核查 2026-08-05
X (Twitter). 广告接口：«We do not support multi-segmentation»——一次只能按一个维度细分，而且只能从四个中选：AGE、GENDER、METROS、PLATFORMS。部分 ENGAGEMENT 指标«are not available at the account and funding instrument level»。LIFE_TIME_VALUE_MOBILE_CONVERSION«available only to advertiser accounts enabled for… 来源, 已核查 2026-08-05
YouTube. 平台不公布隐藏数据的门槛：«The actual thresholds at which data is limited are not published and subject to change at YouTube's discretion»。受限的有：«Demographics data, such as age and gender»、«Geographic data, including metrics or dimensions related to the countries or… 来源, 已核查 2026-08-05
YouTube. «You won't see demographics data for private or unlisted videos, regardless of the amount of traffic»；«You won't see any geography data for the following, regardless of the amount of traffic: Realtime reports»，私密视频和通过链接访问的视频也是如此 来源, 已核查 2026-08-05
YouTube. «one-off search terms and URLs that drive little traffic may not show»；当所选的筛选项或维度达不到门槛时，«only the top results are available» 来源, 已核查 2026-08-05
YouTube. 通过 API：«YouTube Analytics API responses do not contain metric breakdowns for deleted items» 来源, 已核查 2026-08-05
YouTube. 已过时的注释类指标：«YouTube Analytics API reports only return data for the annotationClickThroughRate and annotationCloseRate metrics as of June 10, 2012. In addition, YouTube Analytics API reports only return data for the remaining annotation metrics as of July 16… 来源, 已核查 2026-08-05
YouTube. 后台导出的上限：«Downloaded reports are limited to 500 rows. Use the YouTube Reporting API to download more than 500 rows of data.» 来源, 已核查 2026-08-05
YouTube. 关于收入的保留说明：«Estimated revenue metrics are subject to month-end adjustment and don't include partner-sold and partner-served advertising»；«Estimated earnings data in YouTube Analytics doesn't reflect invalid activity adjustments in real time. Final revenue data… 来源, 已核查 2026-08-05
VKontakte. “帖子统计目前仅对成员数不少于5000的社群开放。” 来源, 已核查 2026-08-05
VKontakte. “该方法返回最近300条帖子的数据，且这些帖子的发布时间不早于2015年12月9日。” 来源, 已核查 2026-08-05
VKontakte. 文档中没有披露 stats.get 返回值的构成：“成功执行后返回一个包含统计数据的对象”——没有字段清单 来源, 已核查 2026-08-05
VKontakte. date_from 和 date_to 参数“已弃用”，自5.86版起改用 timestamp_from 和 timestamp_to 来源, 已核查 2026-08-05
Dzen. “对于2019年6月1日之前发布的视频，‘观看后订阅’这一指标不可用。” 来源, 已核查 2026-08-05
Dzen. 显示门槛以数字给出：“读完分布图……当发布内容的读完次数超过100时才显示数值”；“观看深度……当发布内容的观看次数超过100时才显示数值” 来源, 已核查 2026-08-05
Dzen. 地理维度的门槛：“如果某个城市或国家的数据不足，它们会被归入‘其他’一栏。数值小于0.5%的城市和数值小于0.1%的国家都会落入这一栏” 来源, 已核查 2026-08-05
Dzen. 早于这些日期就没有指标：“您无法查看早于2022年1月25日的数据”（订阅者方面）；“您无法查看早于2021年6月21日的事件数据”；“参与观看时长的数据自2023年8月1日起计入。您无法查看更早日期的信息”；“性别和年龄数据自2021年6月21日起显示” 来源, 已核查 2026-08-05
Dzen. 这些数字注定与发布内容下方看到的对不上：“所选周期内创建的所有评论都会计入，包括已删除的，因此当前实际的评论数量可能与表中的数字不符”；点赞也一样——“即使用户后来取消了标记”。此外还有“只有未被施加限制的发布内容才计入观看时长” 来源, 已核查 2026-08-05
Dzen. 实时统计“不经过预处理，可能包含机器人流量和 SEO 流量”；该板块的数据“不影响变现” 来源, 已核查 2026-08-05
阅读时的陷阱
- 空的回答不是零。 如果字段是空的，就是没有数据；不能把它当作零（见 平台指标）。2. 小数值会被隐藏。 YouTube 明确表示，隐藏数据的阈值不会公布：小的数值会从报告中消失，但并不因此变成零。3. 时区。 分析数据有时按 UTC 计（例如 Vimeo 的导出）— 平台的«昨天»可能和你的不一致。4. 展示量的降低按规则来，不按传闻来。
Bluesky. 没有直接承诺降低展示量。社群规则中点出的是对账号的限制和删除，而不是降权：«Accounts that engage in extreme or persistent toxic or disruptive behavior may face restrictions or removal.»关于刷量的禁令：«Do not artificially manipulate features or social signals to gain unearned reach or… 来源, 已核查 2026-08-05
Instagram. 降低展示量是被直接承诺的，而且方式不止一种。对违反推荐规则的：«Content that goes against our Recommendation Guidelines can still be distributed normally to your followers, but we try not to recommend it to non-followers»；«if you repeatedly post content that goes against… 来源, 已核查 2026-08-05
LinkedIn. 降低展示量是被直接承诺的，但说得很笼统。原文，2026年3月12日：«Less generic content and engagement bait: We're reducing recycled and click-driven posts to make the Feed feel more useful. We want the Feed to be relevant to your interests, not a popularity… 来源, 已核查 2026-08-05
Pinterest. 降低展示量被列为一项单独的措施，而且定义是逐字给出的：«Limiting distribution: When we limit the distribution of a Pin, board or account, it will continue to be accessible on Pinterest to the user who posted it and to users with a direct link, but it won't be… 来源, 已核查 2026-08-05
Reddit. 可访问页面上没有因违规而降低展示的直接承诺；点名的是删除和被读者投票压低。原文照录：«Low-quality or manipulative content is quickly downvoted by users, reducing its visibility.»。关于自身的工作：«Our goal is simple: limit the impact of bad actors and exposure to their content. We don’t… 来源, 已核查 2026-08-05
Threads. 直接承诺不予推荐的只有一项——来自未关注账号的政治内容：«On Threads, we aim to not recommend political content from accounts you don’t follow.»在可访问的页面上没有其他关于降低展示量的承诺 来源, 已核查 2026-08-05
TikTok. 降低展示量用«ineligible for recommendation»这个说法来表述。原文：«Reviewed content found to depict things like graphic medical procedures or legal consumption of regulated goods, for example – which may be shocking if surfaced as a recommended video to a… 来源, 已核查 2026-08-05
Twitch. 平台没有把降低展示量作为一项单独措施来承诺；点名的是删除。原文：«Our recommendation systems are designed with focus on your safety, so we may remove content that we determine has violated Twitch's Community Guidelines or Terms of Service.»唯一像降权的做法是直播优先于录像：«live channels… 来源, 已核查 2026-08-05
关于方法。 每条定义都是平台文件的引文或转述，附有链接和日期。我们不计算派生指标，也不指定«正确的»指标。