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爆款，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 to find breakout hits, then turn winning content into global traffic — for AI and tech teams going abroad.

免费获取测试方案Get a free test plan
PROVEN
示例数据Sample data

示例项目·TikTok·19天·测试+首轮放量·累计39,496条Sample project·TikTok·19 days·test+scale·39,496 posts

短视频Format条数Posts均播Avg进站率Visit注册率Sign-up下一步Next
真人出镜On-camera3,1171,883接近near0.72%接近near9.3%接近near优化后复测Retest
产品介绍Product7,635225达标pass0.41%偏低weak5.4%偏低weak暂不放量Hold
UGC 内容UGC28,7441,020达标pass1.43%达标pass14.8%接近near放量Scale
合计Total39,4969341.27%14.2%
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 TikTok · Reddit · X · YouTube · LinkedIn · Instagram · Facebook
团队Team 华人团队 · 全球交付Based in China · delivers worldwide

基本单元效果Base unit results

自然曝光Organic views30万–150万–300万300K–1.5M–3M1,500 条 × 均播 200–1,000–2,0001,500 posts × avg plays 200–1,000–2,000（基本单元：30 个账号×25 天×2 条内容）(Base unit: 30 accounts × 25 days × 2 pieces)
进站Site visits2,400–18,000–42,000自然曝光 × 进站率约 0.8%–1.2%–1.4%Views × visit rate ~0.8%–1.2%–1.4%
注册&下载&询盘Sign-ups · Downloads · Inquiries240–2,880–8,400进站 × 转化约 10%–16%–20%Visits × conversion ~10%–16%–20%

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

参考示例，点击下方可查看Reference examples — click below to view

示例为海外平台链接，需外网访问。Examples link to overseas platforms and may require VPN access.
短视频Short video 切片 · 混剪 · AI · 真人Clips · remix · AI · live
切片Clips
混剪Remix
AI 创作AI creation
真人 + AIHuman + AI
真人拍摄Live-action
图文Posts 图文 · 信息图Posts · infographics
图文Posts
信息图Infographics
短文Short-form Reddit 短内容Reddit posts
短文Short-form
长文Long-form 研究资料 · 采访 · 实测体验Research · interview · tested
研究资料Research-based
采访Interview-based
实测体验Hands-on tested

发布Publishing

有Yes

用你自己的账号发布Post from your accounts

统一发布格式、发布时间与发布习惯：我们给出可执行的排期与文案，你按自己的节奏在账号里发出去，粉丝与数据都留在你这边。One posting format, schedule and habit set: we hand you the calendar and copy; you publish at your pace. Followers and data stay on your accounts.

  • 发布格式：文案结构、编号与追踪链接统一对齐Format: copy structure, IDs and tracking links aligned
  • 发布时间：按平台与你的节奏排出日历Timing: a calendar matched to each platform and your pace
  • 发布习惯：自助发或约定协作方式，数据回收进报告Habits: self-post or a workflow you prefer; results go into the report
无No

用我们的账号发布Post from our accounts

按你的目标平台与行业配置账号，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。Accounts matched to your platforms and industry, scheduled and multiposted. Nothing to open, authorize or hand over — and no need to build accounts before testing.

  • 号池：TikTok 300万+ · Reddit 2万+ · X、Instagram、Facebook 等 5万+Account pool: 3M+ TikTok · 20K+ Reddit · 50K+ on X, Instagram, Facebook and others
  • 覆盖 TikTok · Reddit · X · YouTube · LinkedIn · Instagram · Facebook · 小红书Covers TikTok · Reddit · X · YouTube · LinkedIn · Instagram · Facebook · 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
制作Produce

按 brief 交付成片。Finished pieces delivered to the brief.

  • 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
发布Publish

每条内容对应一条发布。Each piece has a matching post.

  • 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)
  • TikTok · Reddit · X · YouTube · LinkedIn · Instagram · Facebook
  • 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 03Stage 03
数据Data

回收发布数据并分析，交付内容测试报告。We pull back the publishing data, analyze it, and deliver a content test report.

  • 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
验收点Milestone报告交付Report delivered

示例数据Sample data

示例项目 · TikTok · 19 天 · 测试 + 首轮放量累计 39,496 条Sample project · TikTok · 19 days · test + first-scale total 39,496 posts

短视频Short video渠道Channel条数Posts播放（均播）Plays (avg)进站（进站率）Visits (rate)注册（注册率）Sign-ups (rate)vs 基线vs baseline下一步Next step
真人出镜On-cameraTikTok3,1175,868,5871,883（接近）1,883 (near)42,2750.72%（接近）0.72% (near)3,9239.3%（接近）9.3% (near)三项接近All near baseline优化后复测Optimize & retest
产品介绍Product introTikTok7,6351,716,866225（达标）225 (pass)7,0470.41%（偏低）0.41% (weak)3825.4%（偏低）5.4% (weak)均播达标，转化偏低Avg pass, conversion weak暂不放量Hold
UGC 内容UGCTikTok28,74429,315,1391,020（达标）1,020 (pass)418,5171.43%（达标）1.43% (pass)61,96414.8%（接近）14.8% (near)整体达标，注册接近Mostly pass, sign-up near放量Scale
合计Total39,49636,900,592934467,8391.27%66,26914.2%

放量基线Scale baseline

接近或超过视为达标Near or above = pass

真人出镜On-camera均播 2,000Avg plays 2,000进站 0.8% · 注册/进站 10%Visit 0.8% · sign-up/visit 10%
产品介绍Product intro均播 200Avg plays 200进站 1.2% · 注册/进站 20%Visit 1.2% · sign-up/visit 20%
UGC 内容UGC均播 1,000Avg plays 1,000进站 1.4% · 注册/进站 16%Visit 1.4% · sign-up/visit 16%

价格Pricing

交付形式Format 制作类型Type 对原始素材依赖度Source-material dependence 制作投入度Production effort 内容说明What's included 单价Price 起订量Minimum
短视频Short video 切片Clips
依赖高High
投入低Low
 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
依赖高High
投入较低Lower
 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
依赖中Medium
投入中Medium
 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
依赖较低Lower
投入较高Higher
 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
依赖低Low
投入高High
 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
依赖中Medium
投入较低Lower
 AI 生成图文内容AI-generated posts $30 / 篇/ piece 100 篇起From 100
信息图Infographics
依赖中Medium
投入中Medium
 AI 生成信息图内容AI-generated infographics $60 / 篇/ piece 100 篇起From 100
短文Short article 短文章创作Short article creation
依赖中Medium
投入中Medium
 Reddit Post（AI+人工审核）Reddit post (AI + human review) $15 / 篇/ piece 100 篇起From 100
长文Long-form 研究资料Research-based
依赖中Medium
投入高High
 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
依赖较低Lower
投入高High
 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
依赖较低Lower
投入高High
 基于实测体验撰写Written from hands-on testing $5,000 / 篇/ piece 1 篇起From 1
  • 按条数计费：制作多少条，收多少条的费用；量大更优价，长文满 10 篇 8 折，社媒发布按条数另计。Billed per piece — you pay for what we produce. Better rates at volume; 20% off long-form from 10 pieces; social publishing billed per post.
  • 依赖度高需更多素材；投入度高单价通常更高。Higher dependence needs more source material; higher effort usually costs more.

合作流程Process

  1. 01填写内容需求Share your content needs
  2. 02评估&对齐产品、阶段、内容需求、增长目标等Assess & align product, stage, content needs, growth goals
  3. 0324小时内免费回你测试方案与起步报价Free test plan & starting quote within 24h
  4. 04确认brief，先出样片，再进入测试批次Confirm brief → pilots → test batch
免费获取测试方案Get a free test plan

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

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、报告三段验收；按条数计费——做多少条收多少条，发布按条数另计，制作、发布和报告全部由一个团队交付，你只对接一个团队。Pilots, batch and report — three acceptance checkpoints. You pay per piece produced, plus per post published from our accounts, all delivered by 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+，X、Instagram、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 X, Instagram, 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 TikTok、Reddit、Google、X、Meta 有压低获客成本的记录。A record of driving down acquisition cost on TikTok, Reddit, Google, X and Meta.

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.

ProvenMart 介绍About ProvenMart

ProvenMart 提供从切片、混剪、AI 创作到真人实拍的全形态内容生产，覆盖短视频、图文与深度长文。依托 AI 内容流水线与标准化制作团队，单条成本与交付周期相较传统工作室大幅降低，支持千条级批量稳定交付；所有内容按海外平台原生语境与算法偏好制作，开箱即可用于矩阵分发与广告投放，帮助客户以最低成本快速完成素材测试、持续放大跑量内容。ProvenMart produces full-spectrum content — from clips and remixes to AI creation and live-action — across short video, posts, and long-form. An AI content pipeline and a standardized production team cut unit cost and turnaround versus traditional studios, with stable delivery at thousand-post scale. Every piece is built for overseas platforms’ native tone and algorithms, ready for matrix distribution and paid ads — so you can test creatives fast at the lowest cost and keep scaling what works.

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，提交后 24 小时内回复：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, back to you within 24 hours: 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 2 天交付，测试批次 2 周出数。验收指标事先写进 brief：曝光、进站成本、互动与转化，接近或超过放量基线才算过线——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 in 2 days; test-batch data in 2 weeks. Success metrics go into the brief up front — impressions, cost per visit, engagement and conversion — and near or above the scale baseline counts as a pass. 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.
内容出厂前怎么审核？How do you review content before it ships?
每一条成片发布前过三道：制作自查（画面、字幕、第三方 Logo 与幻觉）、品牌与母语二审、版权与法务兜底。三道都通过，才进入发布。Every piece clears three checks before publishing: a production check (visuals, captions, third-party logos, hallucinations), brand and native-language review, then copyright and legal. It ships only after all three pass.
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?
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，不用你开通、授权或交出账号。用我们的账号发布：数据由我们回收进内容测试报告，发布失败、被删或账号异常我们补齐。Post from your own accounts if you have them; if not, ours — nothing to open, authorize or hand over. From ours: data comes back to us and into the content test report, and failed, removed or restricted posts are made up.
怎么付款？How does payment work?
按条数预付：制作多少条，就收多少条的费用，单价见「价格」表。发布按条数单独计费（采用我们的账号发布时）。Prepaid per piece: you pay for the pieces we produce, at the per-piece prices in the Pricing table. Publishing is billed per post when we post from our own accounts.

把产品发给我们Tell us about your product

联系方式、产品网址、阶段、内容需求。Contact, product URL, stage, and content needs.

免费获取测试方案Get a free test plan
免费获取测试方案Get a free test plan