Like it or not, TikTok is now a global phenomenon, not just in terms of cultural relevance, but also as a real business success. (As we all have seen with Twitter, you can be very culturally relevant and still struggle as a business.)

Based on the latest round of earnings reports from all the large social media companies, everyone is having trouble growing and competition from TikTok is a big reason why. As popular as TikTok may be, how it got to where it is today is poorly understood – until this Chinese language article was published last week. Titled “TikTok Behind the Scenes: Zhang Yiming’s Global Journey Among Giant Waves” (original Chinese title: 《TikTok内幕:张一鸣的巨浪征途》), this deeply-sourced, well-reported, and beautifully-written article told the story of ByteDance’s global ambition that dates back to at least 2015 – long before TikTok was conceived.

For more of our previous writings on TikTok, see “Zhang Yiming’s Last Speech” (premium content) and the ByteDance tag.

There is a ton of materials worth unpacking in this 22,000+ character long article. Today, we will focus on one part that is particularly interesting to me – how TikTok strategizes and invests in its global expansion. (My day job is driving global expansion strategy at GitHub, which is, of course, a very different type of company and product than TikTok.)

Market Ranking Framework: S/A/B/C/Others

To prioritize all the markets TikTok wants to expand to outside China, it established a “S/A/B/C/Others” market ranking framework. Here is how this framework works:

  • S (as in “strategic”, highest priority): United States, Japan, United Kingdom, India (now banned)
  • A (2nd priority): Germany, Brazil
  • B (3rd priority): a group of 15-20 countries. Examples include France, Italy, Spain, the Netherlands, Russia, South Korea, Indonesia, Mexico, Canada, Australia.
  • C (4th priority): a group of 20-30 countries. Examples include Thailand, Vietnam, Saudi Arabia, UAE, Egypt, Argentina.
  • Others (lowest priority): rest of the world

The author of this article, Zhang Xiaoyun, made a helpful graphic (in Chinese) summarizing this market ranking model and the reasoning behind each tier.

Some of the inputs TikTok used to rank these markets are quite common: population, GDP, mobile internet penetration level, size of digital advertising market. Other inputs are based on proprietary data (possibly illegally obtained) and competitive dynamics.

As the article revealed, ByteDance lured many Facebook managers away in 2019, when TikTok was also spending millions of dollars advertising on Facebook. Along with those ex-Facebook managers came Facebook’s internal, proprietary data that mapped not only its daily active users (DAUs) in every country, but also its Average Revenue Per User (ARPU) by country.

Having this data helped TikTok populate its “Tier B” countries, some of which might have been ignored if Facebook’s data did not reveal their commercial potential. Russia and Latin American countries received priority, simply because ByteDance’s Chinese competitors, Kuaishou and Likee, were also expanding into those markets. This “S/A/B/C/Others'' framework was fully operationalized in early 2020, when TikTok the product was reasonably mature (after merging with and Zhang Yiming was hyper-focused and hyper-aggressive in realizing his global expansion ambition.

Back then, his own OKR was “No Gaps Allowed, Full Court Press” (“不留空档、全面压制”).

Every Two Months: Adjustment with Execution

TikTok uses this framework to allocate user growth spending, but every two months, these countries’ tiering is evaluated and adjusted with learning and data from execution.

ByteDance is famous for its “every two months” cadence, which permeates every corner of the company, not just engineering or product. Zhang Yiming’s own executive-level OKRs follow the same cadence. This is fast, very fast, especially for a global company that now has more than 130,000 people working for it.

Some interesting examples of countries being adjusted:

Indonesia: it started out in “Tier B”, but was elevated to “Tier A”, because TikTok discovered that its Indonesian users have a stronger penchant for creating content than other countries – 10% versus Chinese users’ 5% and Japanese users’ 2-3%.

South Korea: it started out in “Tier A”, but was dropped for a while to “Tier C” after lackluster DAU growth, then later elevated to “Tier B”. South Korea has been a difficult market for TikTok. Some employees posit that Korean users are very nationalistic and generally resist foreign products. Others believe Korea’s entertainment market is so mature and full of choices that it is more difficult for a new product to gain traction.

Iraq: it was always in the “Others” tier, but the way TikTok gained traction there was particularly noteworthy. The “Others” tier buckets all countries, where TikTok does not believe requires any spending to grow, because they will grow organically as downstream markets of other countries. Sounds like a nice theory, but does it work in reality? It did, and surprisingly well, in Iraq. In the Middle East, TikTok spent some money acquiring users in Saudi Arabia and the UAE to trigger this downstream effect, and its traction in Iraq skyrocketed. During the early days of the pandemic, TikTok’s DAUs shot up to 6 million in Iraq, out of a total mobile internet population of 25 million – a 24% market share without spending a single cent!

These market expansion learnings and tiering adjustments illuminated a profound insight – content downstream effect.

Unlike building a social network with a real social graph, content can be pooled and leveraged cross-culturally in a much more scalable way, if you know which (downstream) countries like which (upstream) countries’ content. TikTok’s data and experience yielded these patterns:

  • Chinese content has downstream effects in East Asia and Southeast Asia
  • Saudi Arabian content has downstream effects in other Middle Eastern countries
  • American content has downstream effects…everywhere!

That’s why TikTok was hellbent on growing in the United States, even if it will cost a fortune and unprecedented political scrutiny.

A Survival Guide for Meta?

As TikTok’s global expansion strategy and methodology becomes better understood via resources like this article, the biggest beneficiary may be TikTok’s biggest nemesis, Meta fka Facebook.

Meta’s one product that can go up against TikTok is Instagram Reels. While there are some video length differences – TikTok video limits are longer than Reels’s – the two products are very similar. In fact, much of the best performing content on Reels are originally from TikTok, with TikTok watermarks.

Recently, Instagram officially announced that it will demote Reels that have a TikTok watermark, likely a desperate defensive measure. But for IG Reels to go on the offense, it should tap into the same framework and methodology that TikTok is using, which was built on Facebook’s internal data anyway. The content downstream effect can work for IG Reels too, as long as it has the content and knows to which direction its algorithms should push.

Before it became Meta, Facebook was mostly asleep at the wheel in 2018, as TikTok poured millions of dollars into buying Facebook ads to fuel its global growth. Even though it cloned TikTok with Lasso in mid-2018 (shut down in mid-2020), TikTok was such a huge customer that Facebook happily ignored its own role in seeding its largest competition.

For all the Meta shareholders out there, let’s hope it won’t make that mistake again and take more pages out of TikTok’s playbook instead. After all, Facebook has never been shy about copying anything.

TikTok 的全球扩张之路

(本篇中文版文章是读者 Ben Yu 做的翻译,我做了一些修改而发表。非常感谢Ben的贡献,《互联》的读者们真伟大!)

不管你喜不喜欢,愿不愿意,今天的 TikTok 已是一款全球规模的产品了,这不仅体现在 TikTok 在多个文化迥异的国家中被广泛使用,也体现在它获得的丰厚商业回报上。(一个反例是 Twitter,虽然 Twitter 也被多个国家广泛使用,但它一直没有找到好的商业模式。)

各大大型社交媒体公司最新的财报显示,每一家公司都或多或少都遇到了业务增长的挑战,而 TikTok 是这挑战的一大原因。TikTok 是如何一步步走到今天的?人们对此一直知之甚少,直到上周一篇调研详尽,非常好读的文章《TikTok 内幕:张一鸣的巨浪征途》 为读者们讲述了字节跳动全球化扩张的历程。这个故事可以最早追溯到 2015 年——此时距离 TikTok 的成立还有很长一段时间。

我们此前也写过很多有关 TikTok 的文章,你可以阅读《张一鸣的最后一次演讲》(付费内容),或者查看 ByteDance 文章标签。

在这篇超过 22000 字的文章中包含了大量资料,今天我尤其想讨论其中最令我感兴趣的部分——TikTok 如何制定它全球化扩张的投资战略。(我的日常工作是推动 GitHub 的全球化扩张的战略,当然GitHub和 TikTok 是截然不同的两个公司和产品类型。)


为了对中国以外的所有市场做优先级排序,TikTok 构建了一个“S/A/B/C/其他”的市场排名框架:

  • S:最高优先级,如美国、日本、英国、印度(受到地缘政治的影响,目前 TikTok 在印度被封禁中);
  • A:第二优先级,如德国、巴西;
  • B:第三优先级 ,有大约 15-20 个国家,如法国、意大利、西班牙、荷兰、俄罗斯、韩国、印度尼西亚、墨西哥、加拿大、澳大利亚等;
  • C:第四优先级,有大约 20-30 个国家,如泰国、越南、沙特阿拉伯、阿联酋、埃及、阿根廷等;
  • 其他国家:最低优先级。


TikTok 用了一些常见的估算判断市场潜力的维度:人口规模、GDP、移动互联网渗透率、数字广告市场规模等。TikTok也用了一些专利甚至是竞争对手的机密数据(很有可能是非法获取的)和其他竞品的竞争情况来做判断。

文章中提到,2019 年 TikTok 在 Facebook 上投入了数百万美元的广告费,同时也挖了很多 Facebook 各层管理人员加入。字节除了得到这些人才外,还得到了 Facebook 内部的独家数据,例如每个国家的日活(DAU)和每用户平均收入(ARPU)。

这些数据帮助了 TikTok 判断 B 级的国家有哪些,仅依靠自己 TikTok 很可能会忽视掉其中一些有潜力的国家。俄罗斯和拉丁美洲的优先级相对来说比较好理解,因为字节的竞争对手快手和 Likee 已经在这些国家有推广。这个“S/A/B/C/其他框架“在 2020 年初开始大范围使用,当时 TikTok 在和 合并后,产品趋于成熟,张一鸣也在这个全球化扩张的阶段中表现出自己高度专注,富有野心的一面。那时候,张一鸣的个人 OKR 就是“不留空档、全面压制”。


TikTok 虽使用这个框架来分配用于用户增长的预算,但每两个月,这些国家的优先级就会基于执行中得到的数据反馈,被重新评估调整。

字节以双月作为节奏,并且渗透到了公司的每一个角落,不仅仅是工程和产品上,张一鸣自己的 OKR 也遵循这一节奏。对于一个现在有超过 13 万员工的跨国大公司来说,这种节奏是非常迅速的。


  • 印度尼西亚:最开始是“B 级”,后来升到了“A 级”,因为 TikTok 发现印度尼西亚用户比其他国家更喜欢创作,大约有 10% 的用户是创作者,与之相比,中国这一比例为 5%,日本则是 2~3%。
  • 韩国:一开始是“A 级”,但在 DAU 增长乏力后,一度下降到“C 级”,后来又升到“B 级”。对于 TikTok 来说,韩国是一个困难重重的市场,一些员工认为韩国用户有非常强的民族主义,普遍抵制非本国产品,另一些员工认为韩国的娱乐市场已经非常成熟,消费者有足够多的选择,因此新产品很难有突出重围的机会。
  • 伊拉克:它一直处于“其他”中,值得注意的是 TikTok 对这些市场的策略判断。TikTok 认为“其他” 等级中的一些国家不需要投入额外用于用户增长的预算,因为可以通过文化相近的上游国家输出内容。伊拉克是沙特和阿联酋的文化下游,TikTok 没有在伊拉克花一分钱,而是投放在了它的文化上游。结果在疫情期间,TikTok 的伊拉克 DAU 激增至 600 万,而伊拉克的移动互联网总用户数也才 2500 万左右——相当于日活渗透率达到 24%,月活渗透率高达 40%。

通过不断的市场反馈及时调整的市场扩张策略,TikTok 洞察出了一个极为深刻的产品扩张方法--内容下流效应。

与通过真实生活的社交关系构建的社交网络不同,如果你知道哪个国家(下游)喜欢另一个国家(上游)的内容,那么内容就可以用一种拓展性更强的方式跨文化地被聚集和使用来增长产品,TikTok 的数据和经验总结出了以下的“上下游”关系:

  • 东南亚是中国文化的下游;
  • 其他中东国家是沙特阿拉伯的下游;
  • 而美国的内容则是全世界其他的放的上游!

明白这一点,TikTok 不惜代价也一定要拿下美国市场的原因就显而易见了,即使花费大量财力和需要承受前所未有的政治压力。


随着 TikTok 全球扩张战略方法论的流出,最大的受益人可能是 TikTok 的最大对手,Meta(即 Facebook)。

Meta 唯一可以与 TikTok 拼一战的产品就是Instagram 的 Reels 模块。虽然有一些视频长度上的差异,例如 TikTok 的视频限制时长要比 Reels 要长,但总的来说,两个产品是非常相似的,实际上 Reels 上大部分最受欢迎的内容都带有 TikTok 的水印。

最近,Instagram 官宣将对带有 TikTok 水印的视频降低推荐权重,这可能是孤注一掷的防御策略。对于 Reels 来说,它应该使用和 TikTok 相同的市场排名框架和策略,因为 TikTok 的这套策略本身就建立在 Facebook 的内部数据,而且内容的上下游效应很适用于 Reels ,只要它有内容,且知道算法应该如何优化。

在2018年,当时 Facebook 还没改名成 Meta 的时候, TikTok 投入了数百万美元购买 Facebook 的广告推动其全球增长,而Facebook 却并没在意TikTok的增长。尽管 Facebook 在 2018 年年中拷贝了一款一模一样的产品 Lasso(2020 年关闭)进行回击,但 TikTok 依然是其广告业务的重要客户,以至于 Facebook 在巨大营收的诱惑面前,任由自己最大的竞争对手的崛起。

对于所有 Meta 的股东来说,只能希望它不会再犯这种错误,好好学学 TikTok。毕竟,Facebook 从来不是一家羞于抄袭的公司。