ProvenMart · 先测,再放量ProvenMart · Prove it, then scale

先验证,再放量。Prove it.
Then scale.

先用小批量内容在海外社媒跑出数据,通过有效内容获取全球流量——面向出海的 AI 与科技团队。We test small batches on overseas social, then turn winning content into global traffic — for AI and tech teams going abroad.

PROVEN已验证才放量Scale only what works
内容测试报告Content test report07.31 – 08.18

某 AI Agent工具 · TikTok · 19 天 · 3,518 条An AI agent tool · TikTok · 19 days · 3,518 posts

短视频Short video条数Posts播放Plays进站Visits注册Sign-ups结论Verdict
真人出镜On-camera1,1733,238,66049.6%31,4820.97%5,16416.4%放量Scale
产品介绍Product intro1,172585,3709.0%4,3070.74%54312.6%优化后放量Refine, then scale
UGC内容UGC1,1732,704,38041.4%38,3971.42%8,19521.3%放量Scale
合计Total3,5186,528,410100%74,1861.14%13,90218.7%
80+ 增长客户80+ growth clients AI Infra · AI 应用 · SaaS · B2BAI Infra · AI apps · SaaS · B2B
发布平台Where we publish Instagram · TikTok · Facebook · X · YouTube · LinkedIn · Reddit
团队Team 华人团队 · 全球交付Based in China · delivers worldwide
自然曝光Organic views45万–300万450K–3M1,500 条 × 300–2,000 次观看1,500 posts × 300–2,000 views
进站Site visits5,400–36,000自然曝光 × 1.2%(基准)Views × 1.2% (baseline)
注册&下载&询盘Sign-ups · Downloads · Inquiries1,080–7,200自然曝光 × 1.2% × 20%(基准)Views × 1.2% × 20% (baseline)

30 个账号 · 25 天 · 1,500 条内容30 accounts · 25 days · 1,500 pieces of content

方法论Methodology

先小批量测,用数据说话,过线才解锁下一档。Test small, read the data, unlock the next tier only when it passes.

01

定需求(Brief)Brief

产品、目标市场、内容类型与量级、验收指标。字段固定,brief 即工单。Product, market, content type, volume, and success metrics — fixed fields, so the brief doubles as the job ticket.

时间Timing提交后 24 小时内回复方案We’ll reply with a plan within 24 hours
你只需填:产品 · 市场 · 指标You only fill in: product · market · metrics
02

测试批次Test batch

先出样片确认风格,再把小批量内容发到真实平台,多变量并行对比。Pilot samples first, then a small batch on real feeds, testing several variables in parallel.

时间Timing样片 2 天交付 · 测试批次 2 周出数Pilot samples in 2 days · test-batch data in 2 weeks
用你的账号发,没有则用我们的We post from your accounts — or ours, if you don’t have any
03

读数据Read the data

回收曝光、进站成本、互动与转化,写成内容测试报告:哪条有效、为什么。Impressions, cost per visit, engagement and conversion — turned into a content test report on what worked and why.

时间Timing测试期内持续回收,结束后交付报告Data collected throughout; report delivered at the end
UTM 逐条埋点 · 报告交付Per-post UTM tracking · report delivered
04

分阶梯放量Scale in tiers

只放大数据过线的内容,没过线的换脚本或换打法再测。放量走矩阵,付费投放和 KOL 可以一起配上。We scale what clears the bar, and retest what doesn't with a new script or approach. Scaling runs through our account network; paid ads and influencer partnerships can be added when you want them.

时间Timing报告确认后排期放量Scale scheduled once the report is confirmed
过线才解锁下一档Next tier unlocks only once the bar is cleared

内容制作Content production

样片Samples视频、图文与长文的示例画面Sample frames for video, posts and long-form
切片Clips
混剪Remix
AI 创作AI creation
真人 + AIHuman + AI
真人拍摄Live-action
图文Posts
长文Long-form
制作档位Production tier你需要带来What you bring对你素材的依赖Reliance on your material制作投入Production effort
切片Clips
长视频、直播、访谈等原始素材 → 截取精华,配乐加字幕Long-form footage → cut, music, captions
依赖高High reliance
投入低Low effort
混剪Remix
多段自有或已授权素材 → 重新拼接编排Multiple owned / licensed clips → re-sequenced
依赖高High reliance
投入较低Lower effort
AI 创作AI creation
产品资料与参考 → 基于已有素材重做表达,或 AI 从零生成画面与脚本Product info & references → re-expressed, or generated from zero
依赖中Medium reliance
投入中Medium effort
真人 + AIHuman + AI
产品与脚本方向 → 真人口播或展示,搭配 AI 背景、特效与产品画面Product & script direction → on-camera talent + AI visuals
依赖较低Lower reliance
投入较高Higher effort
真人拍摄Full live-action
产品实物与场景 → 全流程真人实拍,无素材复用、无 AIProduct & setting → fully shot; no reuse, no AI
依赖低Low reliance
投入高High effort
图文Posts
产品资料与卖点 → AI 生成图文与信息图,适配 X、Facebook 等平台Product info & points → AI posts and infographics for each platform
依赖中Medium reliance
投入较低Lower effort
长文Long-form
研究资料、采访对象或可实测产品 → 撰写长文章,一篇起订Research, interview access or a testable product → long-form, from one piece
依赖中Medium reliance
投入高High effort
← 素材现成,更轻← Material ready, lighter从零制作,更重 →Built from scratch, heavier →

发布Publishing

Q 你有可用于发布的账号吗?Do you have accounts we can publish from?
有Yes

用你自己的账号发布Post from your accounts

内容按你的账号定位与受众制作,沿用你已有的粉丝与历史,数据直接沉淀在你自己的账号里。Content is built for your own audience and account positioning; reach and history stay with you.

  • 发布排期、文案与编号由我们提供We provide the posting schedule, copy and IDs
  • 发布方式按平台与你的习惯双方约定Posting method agreed with you per platform and your habits
  • 数据回收后写入内容测试报告Data goes into the content test report once collected
无No

用我们的账号发布Post from our accounts

用我们的账号发布:按你的目标平台与行业配置合适的账号,排期发布、一稿多发。你不用开通、授权或交出任何社媒账号,也不用先搭号再开测。Posted from our accounts, matched to your platforms and industry. Nothing to open, authorize or hand over — and no need to build accounts before testing.

  • 号池:TikTok 300万+ · Reddit 2万+ · Instagram、X、Facebook 等 5万+Account pool: 3M+ TikTok · 20K+ Reddit · 50K+ on Instagram, X, Facebook and others
  • 覆盖 Instagram · TikTok · Facebook · X · YouTube · LinkedIn · Reddit · 小红书Covers Instagram · TikTok · Facebook · X · YouTube · LinkedIn · Reddit · Xiaohongshu (RedNote)
  • 发布记录与截图留档,按有效条数对账Posting records and screenshots kept; reconciled by valid post count
  • 发布失败、被删或账号异常,免费补齐Failed, removed or restricted posts are made up free of charge
  • 验证通过后,放量同样走这条通道Once validated, scaling runs through the same channel

交付Delivery

阶段 01Stage 01
创作Create

按 brief 交付成片。每条独立编号;AI 内容附上工具与 Prompt 记录。Finished pieces delivered to the brief. Every piece is numbered; AI work includes its tool and prompt record.

  • 视频 / 图文 / 长文Video / posts / long-form
  • 英文字幕与母语表达English subtitles and native-level wording
  • 每条含一次常规修改One standard revision per piece
验收点Milestone样片确认 → 批量交付Pilot samples approved → batch delivered
阶段 02Stage 02
审核Review

每一条成片出厂前,过这三道。Every piece clears these three checks before it ships.

  • 制作自查:画面、字幕、第三方 Logo 与幻觉Production check: visuals, captions, third-party logos, hallucinations
  • 品牌与母语二审:品牌合规与目标市场母语审查Brand and native-language review for the target market
  • 版权与法务兜底:版权严格把关Copyright and legal check, strictly enforced
验收点Milestone三道通过 → 才可发布All three passed → cleared to publish
阶段 03Stage 03
发布Publish

每条内容对应一条发布:文案、编号、追踪链接一致。发布和制作分开报价,都写进方案。Each piece has a matching post: copy, ID and tracking link line up. Production and publishing are both priced in the plan.

  • 有账号用你的,没有用我们的(见下节)Your accounts if you have them, ours if not (see Publishing)
  • Instagram · TikTok · Facebook · X · YouTube · LinkedIn · Reddit
  • 逐条差异化发布文案 · UTM 埋点Per-post differentiated copy · UTM tracking
验收点Milestone内容与发布一一对应Content and posts match one-to-one
阶段 04Stage 04
验证Validate

交付一份内容测试报告,写明每条内容的数据与结论。A content test report with the data and verdict for each piece.

  • 曝光 · 进站成本 · 互动 · 转化Impressions · cost per visit · engagement · conversion
  • 胜出内容与原因拆解Winning content and why it won
  • 逐条结论,可对照编号核对A conclusion per piece, checkable against its ID
验收点Milestone报告交付Report delivered

内容测试报告Content test report

某 AI Agent工具 · 19 天 · 3,518 条 · 07.31–08.18An AI agent tool · 19 days · 3,518 posts · Jul 31–Aug 18

短视频Short video渠道Channel条数Posts播放Plays进站Visits注册Sign-ups结论Verdict
真人出镜On-cameraTikTok1,1733,238,66049.6%31,4820.97%5,16416.4%放量Scale
产品介绍Product introTikTok1,172585,3709.0%4,3070.74%54312.6%优化后放量Refine, then scale
UGC内容UGCTikTok1,1732,704,38041.4%38,3971.42%8,19521.3%放量Scale
合计Total3,5186,528,410100%74,1861.14%13,90218.7%

播放占比Share of plays

  • 真人出镜On-camera49.6%
  • 产品介绍Product intro9.0%
  • UGC内容UGC41.4%

进站率Visit rate

  • 真人出镜On-camera0.97%
  • 产品介绍Product intro0.74%
  • UGC内容UGC1.42%

案例Case studies

案例 ACase A10→100 · 存量破局10→100 · Grow the base

开发者内容增长Developer content growth

AI 基础设施 · 全球开发者AI infrastructure · global developers

只追「新模型上线」会疲劳,留存不足 5%。我们先搭一条可复制的内容生产线:同一提示词对比、接入教程、功能 GIF、长文帖串,把心智从「想试一下」推到「工程默认入口」。Launch-day hype alone went stale — retention under 5%. We built a repeatable content engine first: same-prompt comparisons, integration tutorials, feature GIFs, and long-form threads — shifting the story from “worth a try” to “the default choice for engineers.”

做成了什么内容What we made

  • 对比测评Comparisons同一 Prompt 跑多家模型,出截图与适用场景,让开发者一眼看到差异The same prompt across models, with screenshots and use-cases so the difference is obvious
  • 接入教程TutorialsCursor / LangChain / SDK 接入步骤,跟着就能发出第一次成功请求Cursor / LangChain / SDK walkthroughs that get to a first successful request
  • 功能短片Feature clips预设、密钥、可用性等做成 30 秒 GIF,方便二次分发Presets, API keys, and uptime as 30-second GIFs built for resharing
  • 模型日系列Launch series上线说明 → 调用代码 → 成本对比 → 完整教程 → 用户案例,48 小时内出齐Launch note → sample code → cost comparison → full tutorial → customer stories, all within 48 hours
内容节奏Content cadence

早期累计 4 万+ 长内容。叙事随用户升级:尝新 → 工程接入 → 成本与密钥治理 → 工作区 / Agent 工作流。矩阵要发的每一条,都先有对应成片。40,000+ long-form pieces early on. The narrative stepped up with the user: try-it → engineering → cost and key governance → workspaces / agent workflows. Every post had a finished piece behind it.

结果Results

<5% → 约 35%留存率,工程化内容上线后Retention after engineering content shipped
约 50 万 → 500 万+About 500K → 5M+注册用户Registered users
4 万+40K+早期长内容累计Long-form pieces early on
案例 BCase B1→10 · 规模化放量1→10 · Scale up

日本内容冷启动Japan content cold start

AI Agent 产品 · 日本AI agent product · Japan

北美内容搬不过去。英文「自主 Agent」直译无感,声量份额约 0.3%。我们先按当地职场场景重写短视频和图文,小范围测共鸣,过线的角度才批量供稿。North American content couldn’t ship as-is. A literal “autonomous agent” line fell flat — share of voice was about 0.3%. We rewrote short videos and posts for local workplace scenes, tested for resonance in a small batch, and only scaled the angles that passed.

做成了什么内容What we made

  • 场景改写Scene rewrite不翻译,改说法——把「自主完成任务」写成「短时间做完一天的工作」Rewrite, don’t translate — “it does the task itself” became “finish a day’s work in a short sitting”
  • 短视频实测Demo videos程序员 / 职场视角动手演示,让能力看得见Hands-on demos from developer and workplace angles so the benefit is visible
  • 图文种草Seeding posts多角度体验分享,承接短视频后的二次讨论Multi-angle posts that keep the conversation going after the video
  • 转化文案Conversion copy落地页主标题、行动号召与社会证明对照测试Landing-page headline, CTA and social proof A/B’d in parallel
胜出单条Winning piece

程序员实测「用它短时间写完一天的代码」。测出共鸣后,按这一角度扩成系列,再放大。A developer demo finishing a day’s code in a short sitting. Once it resonated, that angle became a series — then it was scaled.

播放约 42 万~420K plays 互动率 6.8%6.8% engagement 收藏率 4.1%4.1% saves 测试均互 3.1% ≈ 基准 ×1.96Test avg 3.1% ≈ 1.96× benchmark

结果Results

0.3% → 6.8%声量份额,内容放量前后Share of voice, before vs after content scale-up
约 680 万About 6.8M总曝光,8 周累计Total reach over 8 weeks
约 9.8 万About 98K候补名单注册Waitlist sign-ups

价格Pricing

按素材、量级和授权报价。起订量就是一次测试,过线后再按阶梯放量。Quoted by material, volume, and usage rights. The minimum is one test; we scale by tier only after it passes.

交付形式Format 制作类型Type 内容说明What's included 单价Price 起订量Minimum
短视频Short video 切片Clips 单一素材截取,简单剪辑、配乐、字幕Cut from a single source with simple editing, music and captions $5 / 条/ piece 500 条起From 500
混剪Remix 多来源自有素材重新拼接编排,不改素材本身Own footage from several sources re-sequenced; the material itself is unchanged $15 / 条/ piece 500 条起From 500
AI 创作AI creation 基于已有素材重做表达,或 AI 从零生成画面与脚本Existing material re-expressed, or visuals and scripts generated by AI from zero $50 / 条/ piece 100 条起From 100
真人 + AIHuman + AI 真人口播或展示,搭配 AI 背景、特效或产品画面On-camera talent with AI backgrounds, effects or product visuals $70 / 条/ piece 100 条起From 100
真人拍摄Live-action 全流程真人实拍,无素材复用、无 AIFully live-action; no material reuse, no AI $110 / 条/ piece 100 条起From 100
图文Posts 图文Posts AI 生成图文内容AI-generated posts $30 / 篇/ piece 100 篇起From 100
信息图Infographics AI 生成信息图内容AI-generated infographics $60 / 篇/ piece 100 篇起From 100
长文Long-form 研究资料Research-based 基于研究资料撰写Written from research material $1,000 / 篇/ piece 1 篇起From 1
采访Interview-based 基于采访撰写Written from interviews $3,000 / 篇/ piece 1 篇起From 1
实测体验Hands-on tested 基于实测体验撰写Written from hands-on testing $5,000 / 篇/ piece 1 篇起From 1
  • 短视频与图文量越大,单价越有空间;长文满 10 篇 8 折;发布单独报价。Larger volumes of short video and posts earn a better rate; 20% off long-form from 10 pieces. Publishing is quoted separately.

先出方案,再做测试。Plan first, then test.

免费出一份内容增长测试方案。确认之后,才进入测试批次。A free content-growth test plan. The test batch starts only after you confirm.

免费获取测试方案Get a free test plan

ProvenMart 优势Why ProvenMart

01

先测再放量Test, then scale

累计 80+ 服务案例。内容先在真实平台验证,过线的才进入下一档放量。80+ clients served. Content is proven on real feeds first — only what passes moves to the next tier.

02

一次交付三段One delivery, three stages

样片、批量、内容测试报告分段验收、分段付款。制作、发布和报告由一个团队交付,你只对接一个团队。Pilot samples, batch, and content test report — each with its own milestone and payment. One team owns production, publishing, and the report, so you only deal with one team.

03

账号不必自备Accounts optional

有账号用你的,没有用我们的。号池现有 TikTok 300万+、Reddit 2万+,Instagram、X、Facebook 等 5万+。发布失败、被删或账号异常,我们补齐。We post from yours, or from ours if you have none. The pool is 3M+ TikTok, 20K+ Reddit, and 50K+ on Instagram, X, Facebook and others. Failed, removed or restricted posts are made up.

04

出厂前三级自检Three checks before it ships

制作自查、品牌与母语二审、版权与法务兜底。版权严格把关。Production check, brand and native-language review, then copyright and legal. Copyright is strictly enforced.

ProvenMart 团队The team

核心团队横跨商业增长、B2B 营销、海外品牌、内容营销与 AI 技术。成员来自 TikTok、谷歌、阿里巴巴和 AI Infra 头部厂商,兼具闽浙沪经商背景,平均从业超过 10 年。The team spans growth, B2B marketing, global brands, content, and AI — people from TikTok, Google, Alibaba, and leading AI infra companies, with commercial roots in China’s coastal hubs (Fujian, Zhejiang, Shanghai) and 10+ years’ average experience.

商业增长Growth

从 0 到规模化操盘From zero to scale

AI Infra:注册用户 50 万 → 500 万,营收增加 20 倍。AI Agent:付费用户 2 万 → 20 万。AI infra: registered users 500K → 5M, revenue up 20×. AI agent: paying users 20K → 200K.

深度服务多家头部 AI 公司冷启动,从验证走到放量。Cold starts for leading AI companies, from proof to scale.

B2B 营销B2B marketing

全链路低成本获客Low-cost acquisition

在 Google、Meta、TikTok、X、Reddit 有压低获客成本的记录。A record of driving down acquisition cost on Google, Meta, TikTok, X and Reddit.

从流量获取做到询盘成交。From traffic through to qualified inquiries.

海外品牌Global brands

冷启动到品类第一Cold start to category lead

4 个月打造品类第一;3 个月海外社媒冷启动,粉丝破 30 万,曝光超 8 亿。Category No.1 in 4 months. An overseas social cold start in 3 months: 300K followers, 800M+ impressions.

主导 TikTok Shop 泰国从 0 做到品类领先。Took TikTok Shop Thailand from zero to category lead.

内容营销Content

细分行业内容第一Category-leading content

发布 50 万+ 图文,年获企业付费线索 10,000+ 条。500K+ posts published, 10,000+ paid business leads a year.

从 0 搭建覆盖全国 8,000+ 商家的内容获客平台。Built a content-led acquisition platform covering 8,000+ merchants across China from scratch.

AI 技术AI

百亿级系统与 AgentLarge-scale systems & agents

核心技术来自阿里 P8 序列,主导过支撑字节跳动百亿级 GMV 的系统架构。Senior Alibaba engineering background (P8-level). Led architecture for ByteDance systems handling tens of billions in GMV.

落地 AI Agent 与 RAG。Ships production AI agents and RAG systems.

常见问题FAQ

ProvenMart 适合谁?Who is ProvenMart for?
已经有产品与站点、正要出海的 AI 与科技团队,我们按验收指标先小批量验证,再放量。AI and tech teams with a live product and site who are going global. We test small against your success metrics, then scale.
测试方案是什么?What is the test plan?
免费的一份内容增长方案:怎么测、测什么、起步怎么报价。确认之后才进入测试批次:先出样片,再发布一批内容到真实平台,最后用数据告诉你哪条有效。规模按你的产品阶段和 brief 来定。A free content-growth plan: what to test, how we’ll test it, and a starting quote. After you confirm, the test batch begins — pilot samples, a batch on real feeds, then data on what works. Scope is set by your stage and brief.
多久出数,怎么算过线?How soon is the data, and what counts as passing?
样片几天内交付。测试批次按已交付案例,两周内能看到数据。有效与否事先写进 brief,例如千次曝光进站成本、互动率、订阅转化,对口径,不对感觉。没过线就换脚本、换角度或换打法再测,直到过线,不靠加预算硬推。Pilot samples land in a few days. On completed projects, test-batch data is usually in within two weeks. Pass/fail criteria are written into the brief — cost per visit, engagement, sign-ups — judged by numbers, not gut feel. If it misses, we change the script, angle, or approach and retest until it passes. We don’t force it through with more spend.
需要我提供素材吗?Do I need to provide material?
看你选的档位。有长视频或自有素材,选切片或混剪,更轻更便宜;没有素材,从产品从零做,投入更大。It depends on the tier. With your own footage, choose clips or remix; without it, we build from your product up.
内容版权怎么保证?How do you handle copyright?
版权是严格审查项,三级自检的最后一层就是版权与法务审查。我们不搬运、不镜像、不变速改写他人视频;音乐、肖像、影视 IP 与品牌素材都需授权,成片以原创为准。AI 生成内容会保留工具、Prompt 与输入素材记录,可随时核对。Copyright is strictly reviewed — the last of our three QC layers is a copyright and legal check. We don't re-upload, mirror or lightly edit others' videos; music, likeness, film/TV IP and brand assets must be licensed. For AI-generated content we keep records of tools, prompts and input material for review.
可以修改吗?Can I request edits?
每条内容原则上包含一次常规修改;因我们未达到 brief 要求而返工,不计入修改次数。Each piece includes one standard revision. Rework caused by missing the brief doesn't count.
发布要用我的社媒账号吗?Do you publish from my social accounts?
看你有没有。你有可用的账号,就用你自己的账号发布;没有,就用我们的矩阵账号发布,不用你开通、授权或交出任何账号。如果发布失败、内容被删或账号异常导致有效条数不足,我们会补齐。两种方式数据都回收进内容测试报告。It depends on whether you have them. If you do, we post from yours; if not, from our account network — nothing to set up or hand over. If posts fail, get removed, or an account is restricted, we make up the shortfall. Results land in your content test report either way.
怎么付款?How does payment work?
分段付款,每段对应一个验收点:样片、批量交付、内容测试报告。Paid in stages, each tied to a milestone: pilot samples, batch delivery, and the content test report.

把产品发给我们Tell us about your product

产品网址、阶段、目标市场、月内容量、联系方式。Product URL, stage, target market, monthly volume, and how to reach you.

免费获取测试方案Get a free test plan

提交之后会发生什么What happens next

  • 01
    我们看你的产品、阶段与目标市场We review your product, stage and market
  • 02
    24 小时内免费回你测试方案与起步报价Within 24 hours you get a free test plan and a starting quote
  • 03
    确认 brief,先出样片,再进入测试批次Confirm the brief, get pilot samples, then start the test batch
免费获取测试方案Get a free test plan