交个朋友副总裁崔东升拆解AI在营销中的应用

作者：龚作仁文章来源：Laborer

2026-11-13 11:54

作者：龚作仁文章来源：Laborer

2026-11-13 11:54

邦小白快读

全文速览

总：AI营销落地不是拼模型强弱，而是拼企业信息数据化率和全流程数据打通，利润与增长都从提效降本和私域运营中来。

1. 利润端：交个朋友累计GMV突破650亿、年增速20%—30%，在增量不增利背景下净利润略增，原因是前几年AI投入见效，运营效率提升、成本下降；关键是把业务全流程经验、动作、结果结构化沉淀，让AI直接调用。

2. 内容端：人群细分后一类人群要一套素材，内容生命周期缩短；用AI调取商品、用户画像、历史素材与投放数据，批量生成脚本、分镜、图片、视频多版本，数量质量迭代提升而成本下降。

3. 私域端：用订单转积分、满减满赠沉淀高价值粉丝，结合AI人像侧写挖掘标签，匹配会员权益、种草内容和会员活动；再用直播预约、订单返现引导回流，私域粉丝超300万、回购率提升179%、粉丝价值提升51%、回流GMV超8亿。

总1：品牌营销正从泛创意转向人群细分与内容精准匹配。

1. 内容生产：人群越精准对素材匹配度要求越高，一类人群就要一套素材；交个朋友用AI调取商品、用户画像、历史素材和投放数据，批量生成脚本、分镜、图片、视频多版本，过去量小成本高的定制化内容也能做，内容数量、质量、迭代速度提升而成本下降。

总2：品牌渠道要公私联动，沉淀数据资产与人格资产。

1. 私域沉淀：通过订单转积分、满减满赠把全域高价值粉丝引入私域，私域粉丝超300万；消费记录和行为记录形成数据资产，颗粒度越细人群分得越清、触达越精准。

2. 人格资产：基于人格认同建立信任关系，靠粉丝口碑内容形成差异化；用KOC种草带动矩阵账号加粉、推高爆款声量、UGC扩散，并用直播预约、订单返现引导回流直播间。数据上回购率提升179%、粉丝价值提升51%、私域回流GMV超8亿、反哺全域业绩占比超20%。

总1：存量竞争下，卖家要抓AI提效和私域全域增长两条主线。

1. 机会与正面影响：交个朋友累计GMV突破650亿、年增速20%—30%，在增量不增利中净利润略增，靠AI投入带来运营效率提升和成本下降；AI降低精细化运营成本门槛，把原本投入产出比不划算的事变得值得做。

2. 风险提示：人群持续细分，泛泛创意打不动人；内容生命周期缩短，迭代速度跟不上流量就让给同行。

总2：可学习点是先提高信息数据化率，再做公私联动。

1. 让业务全流程经验、动作、结果结构化沉淀，直播全流程数据打通，提升选品精准度、上架速度和决策响应。

2. 私域用订单转积分、满减满赠沉淀高价值粉丝，结合AI人像侧写做精准标签和千人千面运营，匹配会员权益、种草内容、会员活动；再用直播预约、订单返现回流。

3. 合作方式上可引入私域陪跑服务商，交个朋友2023年4月引入星云有客共同搭建公私联动智能化私域增长体系；结果是私域粉丝超300万、回购率提升179%、粉丝价值提升51%、回流GMV超8亿、反哺全域超20%。

总1：工厂做产品与内容设计，要先让商品、用户和投放信息可被AI调用。

1. 生产设计需求：人群细分后一类人群要一套素材，AI可调取商品、用户画像、历史素材与投放数据，先判断给哪类人群做什么内容，再批量生成脚本、分镜、图片、视频多版本；过去因量小、成本高覆盖不了的定制化内容现在能做，数量、质量、迭代速度提升而成本下降。

总2：商业机会在AI降低精细化运营门槛与公私联动。

1. AI让原本投入产出比不划算的事变得值得做；企业应用AI上限取决于信息数据化率，差别在业务全流程经验、动作、结果能否结构化沉淀。

2. 推进数字化和电商可借鉴：直播全流程数据打通后选品精准度、上架速度、决策响应提升；用订单转积分、满减满赠沉淀私域粉丝，结合AI人像侧写做精准标签和千人千面运营，KOC种草放大口碑，直播预约、订单返现引导回流。案例数据：私域粉丝超300万、回购率提升179%、粉丝价值提升51%、回流GMV超8亿。

总1：行业趋势是AI从工具走向营销全流程，未来五年内容生产靠AI、流量分发靠模型协同、成交靠数据资产与人格资产。

1. 客户痛点：电商毛利越来越薄、直播行业增量不增利、流量见顶；人群细分使泛创意失效，内容生命周期缩短，迭代跟不上流量就让给同行。

总2：解决方案要围绕信息数据化率、AI内容生产和私域陪跑展开。

1. 先帮客户把业务全流程经验、动作、结果结构化沉淀，让直播全流程数据可被AI直接调用，提升选品精准度、上架速度、决策响应。

2. 内容端用AI调取商品、用户画像、历史素材与投放数据，批量生成脚本、分镜、图片、视频多版本，提升数量、质量与迭代速度并降低成本。

3. 私域端通过订单转积分、满减满赠沉淀高价值粉丝，结合AI人像侧写挖掘精准标签，匹配会员权益、种草内容和会员活动；用KOC种草、直播预约、订单返现做公私联动。案例：交个朋友2023年4月引入星云有客私域陪跑，私域粉丝超300万、回购率提升179%、粉丝价值提升51%、回流GMV超8亿、反哺全域业绩超20%。

总1：平台要回应商家在毛利薄、流量见顶下的增长需求，把公私联动和AI智驱增长做成可落地的运营能力。

1. 商家需求与问题：直播行业增量不增利，电商毛利越来越薄，流量见顶；人群细分后一类人群要一套素材，内容生命周期缩短，迭代跟不上流量就让给同行。

总2：平台最新做法可参考峰会与交个朋友案例。

1. 运营管理：推动商家提高信息数据化率，把业务全流程经验、动作、结果结构化沉淀，让直播全流程数据可被AI调用，提升选品精准度、上架速度和决策响应。

2. 工具与招商：用订单转积分、满减满赠沉淀私域高价值粉丝，结合AI人像侧写做精准标签、千人千面权益与内容匹配；用KOC种草、直播预约、订单返现引导回流，形成公私联动。

3. 风向规避：警惕泛创意失效和内容迭代滞后；可引入私域陪跑服务商合作，交个朋友2023年4月引入星云有客搭建智能化私域增长体系，私域粉丝超300万、回购率提升179%、粉丝价值提升51%、回流GMV超8亿、反哺全域超20%。

总1：产业新动向是AI在营销中落地，直播电商从增量不增利转向效率与私域增长。

1. 交个朋友累计GMV突破650亿、年增速20%—30%，净利润略增，说明AI投入带来的运营效率提升和成本下降开始见效。

2. 未来五年判断：内容生产主要靠AI，流量分发主要靠模型协同，成交主要靠数据资产与人格资产，私域是沉淀这两大资产的最佳场所。

总2：新问题与商业模式启示。

1. 新问题：企业应用AI上限不取决于AI本身强弱，而取决于信息数据化率；各家模型技术相差无几，真正差别在业务全流程经验、动作、结果能否结构化沉淀并被AI调用。

2. 商业模式：公私联动AI智驱增长，私域陪跑服务商合作；订单转积分、满减满赠沉淀粉丝，AI人像侧写做精准标签和千人千面运营，KOC种草建立人格资产，直播预约、订单返现回流。

3. 启示：AI革命本质是技术革命带来效率大幅提升，生产关系与生意形态随之改变；企业应把过去成本太高、不值得做的事用AI重做。数据：私域粉丝超300万、回购率提升179%、粉丝价值提升51%、回流GMV超8亿、反哺全域超20%。

声明：快读内容全程由AI生成，请注意甄别信息。如您发现问题，请发送邮件至 run@ebrun.com 。

我是品牌商卖家工厂服务商平台商研究者帮我再读一遍。

Quick Summary

Overall: AI marketing success depends not on the strength of the model but on the company's business-information datafication rate and end-to-end data integration. Profit and growth come from efficiency gains, cost reduction, and private-domain operations.

1. Profit: Be Friends' cumulative GMV exceeded RMB 65 billion, with annual growth of 20%–30%. Amid growth without profit, net profit rose slightly because AI investments from earlier years began to pay off, improving operational efficiency and lowering costs. The key is to structure and preserve end-to-end business experience, actions, and outcomes so AI can call on them directly.

2. Content: After audience segmentation, each segment requires its own creative set, and content lifecycles are shorter. AI retrieves product data, user profiles, historical creative assets, and ad delivery data to generate multiple versions of scripts, storyboards, images, and videos in batches, improving volume, quality, and iteration speed while reducing cost.

3. Private domain: Use order-to-points conversion and threshold discounts and gifts to accumulate high-value fans, combine AI persona profiling to mine tags, and match member benefits, seeding content, and member events. Then use livestream reservations and order cashback to drive return traffic. Private-domain fans exceed 3 million, repurchase rate is up 179%, fan value is up 51%, and return-flow GMV exceeds RMB 800 million.

Overall 1: Brand marketing is shifting from broad creative campaigns to audience segmentation and precise content matching.

1. Content production: The more precise the audience, the higher the requirement for asset matching; each audience segment needs its own creative set. Be Friends uses AI to retrieve product, user-profile, historical creative, and delivery data, then generate multiple versions of scripts, storyboards, images, and videos in batches. Customized content that was previously too low-volume and costly is now feasible, with higher volume, quality, and iteration speed at lower cost.

Overall 2: Brand channels need public-private integration to build data assets and persona assets.

1. Private-domain accumulation: Convert orders into points and use threshold discounts and gifts to bring high-value fans from across channels into private domains; private-domain fans exceed 3 million. Purchase and behavioral records form data assets; the finer the granularity, the clearer the segmentation and the more precise the outreach.

2. Persona assets: Build trust through persona identification and differentiate via fan word-of-mouth content. Use KOC seeding to drive matrix-account follower growth, amplify hit-product buzz, and spread UGC; use livestream reservations and order cashback to guide return visits to the livestream. Data: repurchase rate up 179%, fan value up 51%, private-domain return-flow GMV above RMB 800 million, and contribution back to overall performance over 20%.

Overall 1: In mature-market competition, sellers should focus on two main lines: AI-driven efficiency and private-domain and full-channel growth.

1. Opportunities and positive impact: Be Friends' cumulative GMV exceeded RMB 65 billion, with annual growth of 20%–30%. Amid growth without profit, net profit rose slightly, driven by AI investment improving operational efficiency and reducing costs. AI lowers the cost threshold for refined operations, making previously unattractive ROI projects worth doing.

2. Risk notes: Audiences continue to fragment, and generic creative no longer resonates; content lifecycles are shortening, and if iteration speed cannot keep up with traffic, competitors will take the traffic.

Overall 2: Key takeaway: first raise the business-information datafication rate, then pursue public-private integration.

1. Structure and preserve end-to-end business experience, actions, and outcomes; integrate end-to-end livestream data to improve product-selection accuracy, listing speed, and decision response.

2. In private domains, convert orders into points and use threshold discounts and gifts to accumulate high-value fans; combine AI persona profiling for precise tags and personalized operations, matching member benefits, seeding content, and member events; then use livestream reservations and order cashback to drive return traffic.

3. On cooperation, sellers can bring in private-domain coaching service providers. In April 2023, Be Friends brought in Xingyun Youke to jointly build an intelligent private-domain growth system integrating public and private channels. Results: private-domain fans over 3 million, repurchase rate up 179%, fan value up 51%, return-flow GMV over RMB 800 million, and contribution back to overall channels over 20%.

Overall 1: For product and content design, factories must first make product, user, and delivery information callable by AI.

1. Production and design demand: After audience segmentation, each segment needs its own creative set. AI can retrieve product data, user profiles, historical creative assets, and delivery data; first determine which type of audience a piece of content targets, then batch-generate multiple versions of scripts, storyboards, images, and videos. Customized content that was previously unservable due to low volume and high cost is now feasible, with higher volume, quality, and iteration speed at lower cost.

Overall 2: Business opportunities lie in AI lowering the threshold for refined operations and in public-private integration.

1. AI makes previously unattractive ROI projects worth doing; the ceiling for enterprise AI adoption depends on the business-information datafication rate. The difference lies in whether end-to-end business experience, actions, and outcomes can be structured and preserved.

2. For digitalization and e-commerce, the lessons are: integrating end-to-end livestream data improves product-selection accuracy, listing speed, and decision response; using order-to-points conversion and threshold discounts and gifts to accumulate private-domain fans, combined with AI persona profiling for precise tags and personalized operations; using KOC seeding to amplify word of mouth; and using livestream reservations and order cashback to guide return traffic. Case data: private-domain fans over 3 million, repurchase rate up 179%, fan value up 51%, return-flow GMV over RMB 800 million.

Overall 1: The industry trend is that AI is moving from a tool to the entire marketing workflow. In the next five years, content production will rely on AI, traffic distribution on model collaboration, and transactions on data assets and persona assets.

1. Client pain points: E-commerce margins are thinning, the livestream industry faces growth without profit, and traffic has peaked. Audience segmentation makes generic creative ineffective, while content lifecycles shorten; if iteration cannot keep up with traffic, competitors will take it.

Overall 2: Solutions should center on business-information datafication rates, AI content production, and private-domain coaching.

1. First help clients structure and preserve end-to-end business experience, actions, and outcomes, making end-to-end livestream data directly callable by AI to improve product-selection accuracy, listing speed, and decision response.

2. On content, use AI to retrieve product, user-profile, historical creative, and delivery data; batch-generate multiple versions of scripts, storyboards, images, and videos; improve volume, quality, and iteration speed while reducing cost.

3. On private domains, use order-to-points conversion and threshold discounts and gifts to accumulate high-value fans; combine AI persona profiling to mine precise tags; match member benefits, seeding content, and member events; and use KOC seeding, livestream reservations, and order cashback for public-private integration. Case: In April 2023, Be Friends brought in Xingyun Youke for private-domain coaching; private-domain fans exceeded 3 million, repurchase rate rose 179%, fan value rose 51%, return-flow GMV exceeded RMB 800 million, and contribution back to overall performance exceeded 20%.

Overall 1: Platforms must respond to merchants' growth needs amid thin margins and peak traffic by making public-private integration and AI-driven growth actionable operational capabilities.

1. Merchant needs and problems: The livestream industry faces growth without profit, e-commerce margins are thinning, and traffic has peaked. After audience segmentation, each segment needs its own creative set; content lifecycles are shortening, and if iteration cannot keep up with traffic, competitors will take it.

Overall 2: Platforms' latest practices can reference the summit and the Be Friends case.

1. Operations management: Encourage merchants to raise their business-information datafication rates and structure end-to-end business experience, actions, and outcomes, so end-to-end livestream data can be called by AI to improve product-selection accuracy, listing speed, and decision response.

2. Tools and merchant acquisition: Use order-to-points conversion and threshold discounts and gifts to accumulate high-value private-domain fans; combine AI persona profiling for precise tags and personalized benefit and content matching; use KOC seeding, livestream reservations, and order cashback to guide return traffic and form public-private integration.

3. Risk avoidance: Guard against the failure of generic creative and lagging content iteration. Introduce private-domain coaching service providers; in April 2023, Be Friends brought in Xingyun Youke to build an intelligent private-domain growth system. Private-domain fans exceeded 3 million, repurchase rate rose 179%, fan value rose 51%, return-flow GMV exceeded RMB 800 million, and contribution back to overall channels exceeded 20%.

Overall 1: A new industry development is AI adoption in marketing, with livestream e-commerce shifting from growth without profit to efficiency and private-domain growth.

1. Be Friends' cumulative GMV exceeded RMB 65 billion, with annual growth of 20%–30%, and net profit rose slightly, indicating that AI investment is beginning to deliver operational efficiency gains and cost reductions.

2. Five-year outlook: Content production will mainly rely on AI, traffic distribution mainly on model collaboration, and transactions mainly on data assets and persona assets. Private domains are the best place to accumulate these two assets.

Overall 2: New questions and business-model implications.

1. New question: The ceiling for enterprise AI adoption depends not on how strong AI itself is, but on the business-information datafication rate. Model technologies are broadly similar across vendors; the real difference lies in whether end-to-end business experience, actions, and outcomes can be structured, preserved, and called by AI.

2. Business model: Public-private integration and AI-driven growth through cooperation with private-domain coaching providers. Use order-to-points conversion and threshold discounts and gifts to accumulate fans; use AI persona profiling for precise tags and personalized operations; use KOC seeding to build persona assets; and use livestream reservations and order cashback to drive return traffic.

3. Implications: The AI revolution is essentially a technology revolution that sharply improves efficiency, with production relations and business forms changing accordingly. Enterprises should use AI to redo what was previously too costly or not worthwhile. Data: private-domain fans over 3 million, repurchase rate up 179%, fan value up 51%, return-flow GMV over RMB 800 million, and contribution back to overall channels over 20%.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

在星云有客2026私域未来峰会上，交个朋友副总裁崔东升围绕峰会主题「公私联动AI智驱增长」，带来《把不可能变成可能，AI在营销中的落地生根》分享。

这场分享，回答了电商企业普遍关心的三个问题：

电商毛利越来越薄，利润从哪里来？

AI人人都在谈，具体怎么落地？

流量见顶，私域如何撬动全域增长？

以下为崔东升的演讲精华整理。

01 如何用AI守住利润

交个朋友直播间累计GMV已突破650亿，GMV年增速维持在20%—30%。在整个直播行业都面临增量不增利的背景下，交个朋友的净利润反而略有增长。原因是前几年在AI上的投入开始见效：运营效率提升，成本下降。

企业应用AI能力的上限，不取决于AI本身有多强，而取决于企业的信息数据化率——经营信息能被AI直接调用的比例。各家模型的技术水平相差无几，企业之间真正的差别，在于业务全流程的经验、动作、结果能不能被结构化沉淀，被AI直接调用。

目前，交个朋友直播全流程的数据已经全部打通，可供AI直接调用。选品精准度、上架速度、决策响应速度，都因此有明显提升。

AI带来的改变，不只在直播间的运营环节。内容生产端的效率瓶颈，AI同样给出了解法。

过去，一条爆款素材通投全量人群、以概率博流量。现在，这套做法走不通了。人群持续细分，一类人群就要一套素材；人群越精准，对内容匹配度的要求就越高，泛泛的创意打不动人；内容生命周期在缩短，迭代速度一旦跟不上，流量就让给了同行。

在内容生产端，交个朋友用AI直接调取商品、用户画像、历史素材与投放数据，团队不必再靠试错去找方向，就知道该给哪一类人群做什么内容。判断清楚之后，脚本、分镜、图片、视频可以用AI批量生成多个版本。过去因为量小、成本高而覆盖不了的定制化内容，现在也能做。内容的数量、质量与迭代速度同步提升，生产成本反而下降。

本质上，AI降低了精细化运营的成本门槛，让原本投入产出比不划算的事变得值得做，让不可能变成可能。

02 如何用私域撬动全域增长

过去五年，电商的内容生产主要靠人，流量分发主要靠算法，成交主要靠货与信任；未来五年，内容生产主要靠AI，流量分发主要靠模型协同，成交主要靠数据资产与人格资产。而沉淀这两大资产最好的地方，就是私域。

交个朋友从2022年开始布局私域，2023年 4月引入星云有客作为私域陪跑服务商，共同搭建公私联动的智能化私域增长体系。通过订单转积分、满减满赠等方式，把全域的高价值粉丝沉淀到私域。目前，交个朋友的私域粉丝已超过300万。

这些粉丝的消费记录和行为记录，积累起来就是数据资产。数据颗粒度越精细，人群分得越清，触达越精准。在粉丝运营上，交个朋友结合星云有客的AI人像侧写能力，从多维数据中挖掘精准人群标签，再匹配对应的会员权益、种草内容和会员活动。在千人千面的精细化运营下，交个朋友的私域粉丝回购率较此前提升179%，粉丝价值提升51%。

人格资产，是基于人格认同建立起来的信任关系。这种关系靠粉丝口碑内容建立，形成品牌的差异化优势。交个朋友通过星云有客的KOC种草功能，带动矩阵账号加粉、推高爆款声量、推动UGC扩散，粉丝的真实口碑由此在全域持续放大。

同时借助直播预约、订单返现等运营动作，引导私域用户回流直播间，带动全域成交。目前私域回流GMV已超8 亿元，私域反哺全域业绩占比超20%，私域已成为全域持续增长的核心。

写在最后

AI革命的本质是技术革命带来的效率大幅提升，生产关系与生意形态随之改变，从而带来新的商业变化。

从直播间全流程打通，到短视频素材、明星素材业务的全面突破，再到深耕私域、沉淀数据资产与人格资产，交个朋友在做的事，本质上都是把过去成本太高、不值得做的事，用AI重新做了一遍，把更多不可能变成可能。

星云有客2026私域未来峰会「公私联动AI智驱增长」已经圆满落幕，更多嘉宾的精彩分享，将在后续陆续发布，请持续关注星云有客公众号，获取最新峰会资讯。

注：文/龚作仁，文章来源：Laborer，本文为作者独立观点，不代表亿邦动力立场。

【版权提示】亿邦动力倡导尊重与保护知识产权。未经许可，任何人不得复制、转载、或以其他方式使用本网站的内容。如发现本站文章存在版权问题，烦请提供版权疑问、身份证明、版权证明、联系方式等发邮件至run@ebrun.com，我们将及时沟通与处理。

已点赞

好文章,值得被更多人看到

FAQ回顾

交个朋友如何用AI提升直播电商运营效率？

交个朋友将直播全流程数据打通，让AI直接调用经营信息，提升选品精准度、上架速度和决策响应速度，运营效率提升、成本下降。其直播间累计GMV已突破650亿元，GMV年增速维持在20%—30%，净利润略有增长。

AI在电商内容生产中有哪些落地应用？

电商内容生产可用AI调取商品、用户画像、历史素材与投放数据，先判断目标人群需要的内容，再用AI批量生成脚本、分镜、图片和视频多版本。这样能覆盖过去因量小成本高做不了的定制化内容，提升内容数量、质量和迭代速度，同时降低生产成本。

电商企业如何通过私域撬动全域增长？

通过订单转积分、满减满赠等方式将全域高价值粉丝沉淀到私域，结合AI人像侧写挖掘精准人群标签，匹配会员权益、种草内容和会员活动；再用KOC种草、直播预约、订单返现等动作引导回流。交个朋友私域粉丝超300万，私域回流GMV超8亿元，反哺全域业绩占比超20%。

交个朋友私域运营效果如何？

交个朋友私域粉丝已超过300万。通过结合AI人像侧写的千人千面运营，其私域粉丝回购率较此前提升179%，粉丝价值提升51%；私域回流GMV超过8亿元，私域反哺全域业绩占比超过20%，成为全域持续增长的重要支撑。

企业应用AI营销的关键瓶颈是什么？

关键不在AI模型本身强弱，而在企业的信息数据化率，即经营信息能被AI直接调用的比例。各家模型技术水平相差无几，企业真正的差别是业务全流程的经验、动作和结果能否结构化沉淀，并被AI直接调用。

标签:: 业界社交电商 AI电商直播电商私域运营交个朋友星云有客内容电商

“点点”之后再推“搭搭” 小红书AI赛道又落一子

日前，小红书正内测一款名为“搭搭”的独立App，定位并非私人专属Agent工具，而是一个由普通人用“活人感”经验创造、分发领域Agent的社区平台——平台内的Agent被称为“搭子”。

未来零售电商平台内容电商种草

2026-11-13 11:43:58