宣传策略制品必知系列：新品item冷起步策略
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">在<span style="color: black;">宣传</span>搜推和自然搜推中，策略<span style="color: black;">制品</span>可能会时常遇到“冷<span style="color: black;">起步</span>”问题，而<span style="color: black;">相关</span>冷<span style="color: black;">起步</span>的<span style="color: black;">专题</span>，又<span style="color: black;">能够</span>分为物品item冷<span style="color: black;">起步</span>、用户冷<span style="color: black;">起步</span>与系统冷<span style="color: black;">起步</span>等几种类型，此时策略<span style="color: black;">制品</span>可能就需要依据<span style="color: black;">区别</span>的类型，拿出<span style="color: black;">区别</span>的应对策略。本文作者介绍了新品item冷<span style="color: black;">起步</span>的<span style="color: black;">关联</span>策略，<span style="color: black;">一块</span>来看。</p><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-tjoges91tu/TGjqS4RInsfn7k~noop.image?_iz=58558&from=article.pc_detail&lk3s=953192f4&x-expires=1728112522&x-signature=IgboWj%2FVQ9VwlxhyHQQCvYXvRP4%3D" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">今天来聊一个<span style="color: black;">大众</span>比较感兴趣，<span style="color: black;">亦</span>是策略<span style="color: black;">制品</span>必须要<span style="color: black;">认识</span>的一个<span style="color: black;">专题</span>叫做「冷<span style="color: black;">起步</span>」。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">冷<span style="color: black;">起步</span>这个词是一个在自然搜推和<span style="color: black;">宣传</span>搜推都会面临的一个问题，<strong style="color: blue;">这个问题的</strong><strong style="color: blue;">核心实质是<span style="color: black;">由于</span>系统/物品item/用户user<span style="color: black;">由于</span><span style="color: black;">累积</span>的数据量过少，<span style="color: black;">没法</span>做个性化<span style="color: black;">举荐</span>的问题，结合前面的<span style="color: black;">文案</span>内容简单说<span style="color: black;">便是</span>在召回<span style="color: black;">其中</span><span style="color: black;">关联</span>性不足<span style="color: black;">没法</span>被召回，在排序<span style="color: black;">其中</span>预估不准确<span style="color: black;">引起</span>的排序展示问题</strong>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">在<span style="color: black;">全部</span>冷<span style="color: black;">起步</span><span style="color: black;">其中</span>的类型就<span style="color: black;">能够</span>被拆分为三大类：</p><strong style="color: blue;">物品item冷<span style="color: black;">起步</span>问题；</strong><strong style="color: blue;">用户user冷<span style="color: black;">起步</span>问题；</strong><strong style="color: blue;">系统冷<span style="color: black;">起步</span>问题。</strong>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">这三类问题我讲拆分成三篇<span style="color: black;">文案</span>来给<span style="color: black;">大众</span><span style="color: black;">仔细</span>介绍策略，今天<span style="color: black;">咱们</span>先由物品Item<span style="color: black;">做为</span>切入来介绍。</p>物品item冷<span style="color: black;">起步</span>问题背景；<span style="color: black;">宣传</span>策略中<span style="color: black;">怎样</span><span style="color: black;">处理</span>新品冷<span style="color: black;">起步</span>问题的策略；个人总结与思考。<h1 style="color: black; text-align: left; margin-bottom: 10px;"><span style="color: black;">1、</span>物品item冷<span style="color: black;">起步</span>问题背景</h1>
<h2 style="color: black; text-align: left; margin-bottom: 10px;">1. 新品冷<span style="color: black;">起步</span>定义</h2>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">物品冷<span style="color: black;">起步</span><span style="color: black;">重点</span><span style="color: black;">处理</span><span style="color: black;">怎样</span>将新的物品，在<span style="color: black;">无</span>样本数据的<span style="color: black;">状况</span><span style="color: black;">举荐</span>给可能对它感兴趣的用户这一问题。</p>
<h2 style="color: black; text-align: left; margin-bottom: 10px;">2. 问题<span style="color: black;">引起</span>的<span style="color: black;">原由</span></h2>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">无论是电商平台淘宝/京东、还是抖音流<span style="color: black;">媒介</span>视频的平台，都会存在物料item上新的<span style="color: black;">状况</span>（<span style="color: black;">针对</span>电商平台<span style="color: black;">便是</span>商品，<span style="color: black;">针对</span>抖音<span style="color: black;">便是</span>用户上传的视频）。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">抖音<span style="color: black;">每日</span>上传新视频达到千万量级，<span style="color: black;">每一个</span>新的物料<span style="color: black;">由于</span>缺少用户的<span style="color: black;">行径</span>数据/Item缺少特征标签等问题<span style="color: black;">原由</span>，会直接致使召回中<span style="color: black;">关联</span>性信息缺失不足（传统<span style="color: black;">根据</span>协同过滤<span style="color: black;">或</span>向量emedding召回方式直接失效）、预估排序中商品/ad特征不足带来的pCTR/pCVR预估不准，<span style="color: black;">没法</span>进行eCPM的准确排序问题。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">最后</span>带来的直接现象<span style="color: black;">便是</span>新品<span style="color: black;">没法</span>正常进行<span style="color: black;">举荐</span>展现PV获取，加重了新品<span style="color: black;">累积</span><span style="color: black;">行径</span>的困难，<span style="color: black;">导致</span>平台存量物料的马太效应，降低<span style="color: black;">举荐</span>系统的新颖性与多样性，<span style="color: black;">同期</span><span style="color: black;">有害</span>于供给端的物料生产。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">例如</span>我写了一篇小红书笔记，<span style="color: black;">长时间</span><span style="color: black;">没法</span>获PV量，<span style="color: black;">或</span>是<span style="color: black;">长时间</span><span style="color: black;">无</span>点赞、评论与关注，其实是<span style="color: black;">有害</span>于<span style="color: black;">全部</span>生态的健康塑造。</p>
<h2 style="color: black; text-align: left; margin-bottom: 10px;">3. 问题的<span style="color: black;">处理</span>与各家发展<span style="color: black;">状况</span></h2>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">日前</span>新品冷<span style="color: black;">起步</span>都是各大互联网<span style="color: black;">机构</span>都在积极并<span style="color: black;">注重</span>的<span style="color: black;">处理</span>对应的问题，新品冷<span style="color: black;">起步</span>问题是<span style="color: black;">举荐</span>系统<span style="color: black;">其中</span>赖以<span style="color: black;">长时间</span>可<span style="color: black;">连续</span>发展的「新颖性」、「多样性」与「准确性」并存健康发展的核心。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">1）阿里妈妈 新品冷<span style="color: black;">起步</span><span style="color: black;">处理</span><span style="color: black;">方法</span>NPA</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">在<span style="color: black;">宣传</span><span style="color: black;">其中</span>，新品Item冷<span style="color: black;">起步</span>是非常重要和关键的一个点。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">为何</span>这么说呢，<span style="color: black;">由于</span>电商平台<span style="color: black;">其中</span>，用户<span style="color: black;">倘若</span><span style="color: black;">期盼</span>自然<span style="color: black;">举荐</span>系统<span style="color: black;">其中</span>，能够<span style="color: black;">帮忙</span>新品度过冷<span style="color: black;">起步</span>状态，稳定获量更甚<span style="color: black;">得到</span>转化数据，就需要<span style="color: black;">经过</span><span style="color: black;">宣传</span>平台直通车/引力魔方来快速实现冷<span style="color: black;">起步</span>，<span style="color: black;">累积</span>数据并后期撬动自然<span style="color: black;">举荐</span>的展示效果（这个属于机制联动，需要<span style="color: black;">累积</span>全平台样本，带来冷<span style="color: black;">起步</span>效果）。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">22年9月中下旬阿里妈妈经营可续就在m峰会上专门针对新品加速冷<span style="color: black;">起步</span>提出了<strong style="color: blue;">NPA（New Product Accerlaration）新品冷<span style="color: black;">起步</span>加速<span style="color: black;">处理</span><span style="color: black;">方法</span></strong>，专门对搜索和<span style="color: black;">举荐</span><span style="color: black;">宣传</span>中新品前期拿量困难，预估不准确等问题提出<span style="color: black;">认识</span>决<span style="color: black;">方法</span>；<span style="color: black;">经过</span>神经网络&新品特性学习&专属新品召回排序通道等方式来<span style="color: black;">处理</span>。</p><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-tjoges91tu/TLiDDPE8bh6jPr~noop.image?_iz=58558&from=article.pc_detail&lk3s=953192f4&x-expires=1728112522&x-signature=UahaJPjrzm1VNNmkYyXJi9woCMI%3D" style="width: 50%; margin-bottom: 20px;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;">2）抖音 倒三角流量池</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">抖音短视频视频冷启有个交互传闻<span style="color: black;">其中</span>比较有名的「8次<span style="color: black;">揭发</span>倒三角流量」的分发机制，<span style="color: black;">亦</span>标志着新视频从冷<span style="color: black;">起步</span>到稳定分发的状态。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">核心的思想<span style="color: black;">便是</span><span style="color: black;">经过</span>每一轮免费单列feeds流量的<span style="color: black;">揭发</span>分发，观察视频的互动与播放<span style="color: black;">状况</span>，决定冷<span style="color: black;">起步</span>视频<span style="color: black;">是不是</span><span style="color: black;">能够</span>进入到下一个流量池<span style="color: black;">其中</span>；播放与互动<span style="color: black;">状况</span><span style="color: black;">包括</span>视频播放过半的比例、完播比例，暂停与结束比例，互动<span style="color: black;">状况</span><span style="color: black;">包括</span>点赞次数、<span style="color: black;">保藏</span>次数以及评论、转发次数。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">除了PV视角还有UV维度的视角，这里举个例子，我拍了一个运动健身视频，在首次<span style="color: black;">揭发</span>300~500播放量的<span style="color: black;">状况</span>下，系统<span style="color: black;">针对</span>进入二次<span style="color: black;">揭发</span>的门槛需要满足点赞次数达到了150/评论达到30/<span style="color: black;">保藏</span>达到80，并且完播次数需要达到50次，才<span style="color: black;">能够</span>进入二次<span style="color: black;">揭发</span>的流量池，<span style="color: black;">倘若</span>不满足则视频<span style="color: black;">再也不</span>进入二次<span style="color: black;">揭发</span>流量池<span style="color: black;">其中</span>，后面的每次<span style="color: black;">揭发</span>池子<span style="color: black;">亦</span>是同一道理。</p><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-tjoges91tu/TLiDDQCAP25sjM~noop.image?_iz=58558&from=article.pc_detail&lk3s=953192f4&x-expires=1728112522&x-signature=5cy0N%2Fz9KTc7pBh0j4Guo6mSSMw%3D" style="width: 50%; margin-bottom: 20px;">
<h1 style="color: black; text-align: left; margin-bottom: 10px;"><span style="color: black;">2、</span><span style="color: black;">宣传</span>策略中<span style="color: black;">怎样</span><span style="color: black;">处理</span>新品冷<span style="color: black;">起步</span>问题的策略</h1>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">基于上面的核心问题<span style="color: black;">原由</span>和定义，<span style="color: black;">咱们</span>从<span style="color: black;">宣传</span><span style="color: black;">举荐</span>（自然<span style="color: black;">举荐</span>在<span style="color: black;">非常多</span>方面类似）的<span style="color: black;">方向</span>出发，来思考<span style="color: black;">能够</span>从<span style="color: black;">那些</span>方面来<span style="color: black;">处理</span>新品冷<span style="color: black;">起步</span>的问题。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">回顾<span style="color: black;">以上</span><span style="color: black;">其中</span>在召回/排序<span style="color: black;">其中</span>的冷<span style="color: black;">起步</span>的问题，<span style="color: black;">咱们</span>逐一拆解，来分析<span style="color: black;">处理</span>思路；<span style="color: black;">宣传</span><span style="color: black;">举荐</span>在召回<span style="color: black;">其中</span>要<span style="color: black;">处理</span>的<span style="color: black;">便是</span>快速<span style="color: black;">创立</span>物品和用户<span style="color: black;">关联</span>性准确度问题；在eCPM排序<span style="color: black;">其中</span>要<span style="color: black;">处理</span>的<span style="color: black;">便是</span>预估准确和出价准确的问题。</p>
<h2 style="color: black; text-align: left; margin-bottom: 10px;">1. 利用标的新品metadata信息做<span style="color: black;">举荐</span></h2>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">利用标的物跟用户<span style="color: black;">行径</span>的<span style="color: black;">类似</span>性<span style="color: black;">能够</span><span style="color: black;">经过</span>提取<strong style="color: blue;">新入库的标的物的特征（如新品的标签/类目<span style="color: black;">归类</span>/发货<span style="color: black;">位置</span>、采用TF-IDF算法</strong><strong style="color: blue;">提取的文本特征、基于深度学习提取的图像特征等）</strong>，<strong style="color: blue;">来把新品item label维度信息给Onehot Emebedding向量化</strong><strong style="color: blue;">，<span style="color: black;">经过</span>计算标的物特征跟用户<span style="color: black;">行径</span>特征（用户<span style="color: black;">行径</span>特征通他看过的标的物特征的叠加，如加权平均等）的<span style="color: black;">类似</span>性</strong>，从而将标的物<span style="color: black;">举荐</span>给与它最<span style="color: black;">类似</span>的用户。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">比方说类似的标的物被某些用户消费过，<span style="color: black;">能够</span>做<span style="color: black;">类似</span>物品的<span style="color: black;">举荐</span>，来<span style="color: black;">帮忙</span>物品快速<span style="color: black;">累积</span>数据。</p>
<h2 style="color: black; text-align: left; margin-bottom: 10px;">2. 利用<span style="color: black;">宣传</span>排序公式bid出价来进行E&E探索策略</h2>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">还记得之前给<span style="color: black;">大众</span>介绍的内容，<span style="color: black;">宣传</span>的排序公式<span style="color: black;">是由于</span>两个部分<span style="color: black;">构成</span>，<strong style="color: blue;">一部分是预估问题，另一部分<span style="color: black;">便是</span>竞价bid问题。</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">举荐</span><span style="color: black;">宣传</span><span style="color: black;">其中</span><span style="color: black;">通常</span>会<span style="color: black;">经过</span>bid（<span style="color: black;">由于</span>抬高出价平台无损，<span style="color: black;">宣传</span>承担新品探索成本，新品投放<span style="color: black;">宣传</span>主<span style="color: black;">通常</span>会默认需要付出更高的单价CPC成本）竞价来<span style="color: black;">提高</span>在排序<span style="color: black;">其中</span>的先后<span style="color: black;">次序</span>，<span style="color: black;">而后</span>借用<strong style="color: blue;">强化学习中的E&E（exploration-exploitation）思想</strong>，将新标的物<span style="color: black;">揭发</span>给随机一批用户，观察用户对标的物的反馈，找到对该标的物有正向反馈（购买，<span style="color: black;">保藏</span>，分享等）的用户，后续将该标的物<span style="color: black;">举荐</span>给有正向反馈的用户<span style="color: black;">或</span>与该用户<span style="color: black;">类似</span>的用户，<span style="color: black;">平常</span>的Exploration<span style="color: black;">办法</span>有，<strong style="color: blue;">朴素Bandit、 Epsilon-Greedy、UCB、Thompson Sampling,、 LinUCB、COFIBA等</strong>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">在<span style="color: black;">宣传</span>DSP平台<span style="color: black;">通常</span>会结合「一键起量工具」<span style="color: black;">或</span>是「冷<span style="color: black;">起步</span>工具」，核心思想都是在前期单独设置一笔固定预算/抬高<span style="color: black;">宣传</span>bid-CPC出价水平，来进行E&E的探索，<span style="color: black;">累积</span><span style="color: black;">更加多</span>的点击/转化样本，让pCTR和pCVR预估更准确，<span style="color: black;">然则</span>前期的CPC成本远高于稳定投放时期的投放成本。</p><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-tjoges91tu/TLiDDST1AGmM92~noop.image?_iz=58558&from=article.pc_detail&lk3s=953192f4&x-expires=1728112522&x-signature=qRlZE8i0JwuWVMFke%2FKgH47biTg%3D" style="width: 50%; margin-bottom: 20px;"><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-tjoges91tu/TLiDDTFCpCZDrL~noop.image?_iz=58558&from=article.pc_detail&lk3s=953192f4&x-expires=1728112522&x-signature=DEQSbIEA046IjrTYKLr%2F2TQNSrU%3D" style="width: 50%; margin-bottom: 20px;">
<h2 style="color: black; text-align: left; margin-bottom: 10px;">3. 其他<span style="color: black;">宣传</span>平台的单独扶持策略</h2>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">平台发展到精细化运营<span style="color: black;">宣传</span>投放<span style="color: black;">周期</span>，为了<span style="color: black;">提高</span><span style="color: black;">宣传</span>item物料库的库存深度，会做<span style="color: black;">非常多</span>扶持平台新品的策略，来<span style="color: black;">加强</span>平台供给端-<span style="color: black;">宣传</span>主端的物料深度，<span style="color: black;">同期</span><span style="color: black;">亦</span><span style="color: black;">加强</span><span style="color: black;">宣传</span>主上传物料的积极性。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><span style="color: black;">通常</span>会从「召回扶持」、「排序预估加权」以及「单独<span style="color: black;">宣传</span><span style="color: black;">举荐</span>资源位」来<span style="color: black;">帮忙</span>新品获取前期展现；</strong></p>召回扶持<span style="color: black;">便是</span>给予<span style="color: black;">宣传</span>新品有单独的多路召回通道，<span style="color: black;">加强</span>多路召回混合归一化时候的比重系数，不让新品埋没在召回<span style="color: black;">周期</span>，<span style="color: black;">倘若</span>一<span style="color: black;">起始</span>就没了就更<span style="color: black;">不消</span>探讨排序问题了；排序预估加权是单独在排序<span style="color: black;">周期</span><span style="color: black;">守护</span>一张新品item表，每当新品进入到粗/精排序<span style="color: black;">周期</span>，在预估pCTR和pCVR的时候给予权重打分，让其预估的值更高，排序更加靠前（当然需要<span style="color: black;">保准</span>粗精排一致性原则，<span style="color: black;">不可</span>粗排加权，精排不做任何动作）；单独<span style="color: black;">宣传</span><span style="color: black;">举荐</span>资源位扶持，类似京东小魔方有新品约束<span style="color: black;">要求</span>的<span style="color: black;">宣传</span>位，是<span style="color: black;">不可</span>展现非新品类型，降低新品在eCPM排序中的竞争难度，<span style="color: black;">提高</span>新品的展现效果，前期快速<span style="color: black;">累积</span>样本量。<img src="https://p3-sign.toutiaoimg.com/tos-cn-i-tjoges91tu/TLiDEZ73sbsRYm~noop.image?_iz=58558&from=article.pc_detail&lk3s=953192f4&x-expires=1728112522&x-signature=gk2bC32DFnXou%2BbrQcWdk5Yj7oY%3D" style="width: 50%; margin-bottom: 20px;">
<h1 style="color: black; text-align: left; margin-bottom: 10px;"><span style="color: black;">3、</span>总结与思考</h1>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">新品item冷<span style="color: black;">起步</span>是冷<span style="color: black;">起步</span><span style="color: black;">其中</span>其中的一个分支，核心还是围绕E&E快速<span style="color: black;">累积</span>item数据样本，来<span style="color: black;">加强</span>物品和用户之间<span style="color: black;">关联</span>性的过程。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">任何的扶持类型（召回扶持、预估扶持）都是对<span style="color: black;">宣传</span>平台<span style="color: black;">来讲</span><span style="color: black;">长时间</span>是有收入损失的，<span style="color: black;">由于</span>pCTR高估，<span style="color: black;">或</span>新品展现机会更高，<span style="color: black;">然则</span>item<span style="color: black;">宣传</span>商品未被点击，C计费的场景是<span style="color: black;">没法</span>收取到对应的<span style="color: black;">宣传</span><span style="color: black;">花费</span>。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><span style="color: black;">因此呢</span>核心的最佳<span style="color: black;">办法</span>还是<span style="color: black;">经过</span>E&E的强化学习方式<span style="color: black;">帮忙</span>新品快速<span style="color: black;">累积</span>数据和样本，让系统在召回和预估<span style="color: black;">周期</span>都作出正确的「<span style="color: black;">选取</span>」，缩短与降低新品冷<span style="color: black;">起步</span>的时间，达到稳定投放的状态。</strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">item冷<span style="color: black;">起步</span>策略是<span style="color: black;">每一个</span><span style="color: black;">举荐</span>系统都会遇到的问题，之前在召回排序<span style="color: black;">其中</span>都是对理想态、存量物料进行说明的方式，策略<span style="color: black;">制品</span>应该和算法思考<span style="color: black;">怎样</span>积极<span style="color: black;">处理</span>Item冷<span style="color: black;">起步</span>的问题，<span style="color: black;">保准</span><span style="color: black;">举荐</span>系统的稳恒状态。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">本文由 @策略<span style="color: black;">制品</span>Arthur 原创发布于人人都是<span style="color: black;">制品</span>经理，未经许可，禁止转载</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">题图来自 Unsplash，基于 CC0 协议</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">该文观点仅<span style="color: black;">表率</span>作者<span style="color: black;">自己</span>，人人都是<span style="color: black;">制品</span>经理平台仅<span style="color: black;">供给</span>信息存储空间服务。</p>
楼主的文章深得我心，表示由衷的感谢！ 你的见解独到，让我受益匪浅，非常感谢。
页:
[1]