“When you gaze into TikTok, TikTok is staring right back into your soul”

At the cost of sounding too dramatic, this is possibly my favorite way to start a conversation about TikTok. Funnily enough, it is also the key to TikTok’s explosive growth and ever-increasing popularity. It’s simple: every interaction you have with TikTok is always a two-way street. The more you swipe, comment, like and even hover, the more it learns about you. Thus, translating all this data into instant gratification every time you swipe up to watch a new video. If this concept sounds similar to every social media app you have heard of, you’re right. But TikTok does this significantly better and faster, primarily by building on top of interest graphs and not social graphs. You don’t need to follow anyone or answer those age-old questions of what your hobbies and interests are to get fresh, relevant content delivered to your homepage. It already knows. And chances are, it probably knows them better than you do yourself.

This growth is also accelerated due to the new-found spotlight on it and the short-form video space in general, thanks to the latest trend of 2020 (right after Dalgona Coffee): several countries banning the app. The traction in this space is especially more pronounced here in India due to the growing anti-China sentiment as highlighted by the government mandated purge of 59 Chinese apps in July which included all Bytedance entities. At 200MM MAUs, India was the single biggest market for TikTok globally, in terms of number of users. A majority of its users were 18-35 years old and come from tier-2 and tier-3 cities. Approximately 52% of its users earn less than Rs 25,000 /month. Propelled by the exit of TikTok, access to cheap mobile data powered by Jio, and the availability of budget smartphones has created an ever-expanding whitespace of these ‘Bharat’ users. This has resulted in a fascinating battle of video content being fought between upstarts and new offerings by incumbent Indian media / digital content companies.

Instead of trying to predict a winner, we’re going to unbundle what it takes to build a brilliant company in the short-form video app space.

Let the battle begin!

Content length on different platforms

Understanding the Consumers

A lot of the companies starting out in India in the short-form video space are hopeful of replicating the success that has been observed in the Chinese ecosystem. To make this truly happen, we also need to look at the consumer demographics and behavior.

As it turns out, Indian and Chinese consumers are a lot more alike than you’d think.

Both countries have a majority of their population living outside Tier 2 cities (China - 65%, India - 70%) and a solid chunk of their population is Middle Class (China - 31%, India - 22%).

China is not just mobile-first, but also mobile-only

India also has more internet users in the rural (320MM) than urban areas (258MM). Rural internet penetration currently stands only at 30%, as compared to 63% in the Metros. There is a huge gap here, being rapidly bridged by the “Jio effect”. This is the generation of users whose introduction to the internet has been mobile-first and majorly through videos.

If you look at this qualitatively, Indian internet users are increasingly becoming mobile-first, with a strong appetite for video content, and a growing comfort around influencer-led content. These qualities are very similar to Chinese consumers who only really differ in the way they transact (83% of the users pay via their phones). Due to these similar traits, it is not surprising that Chinese companies were among the first to identify ‘Bharat’ opportunities. This is an important point to be kept in mind- there is definitely a white space waiting to be captured, but it had already been dominated by the Chinese companies before the ban as they were catering to a similar consumer back home. The new indigenous upstarts are not the first-movers.

 Understanding the Market

To understand the upstarts in India, we want to take a step back and elaborate on why Bytedance’s TikTok being embraced globally (especially in India) is an anomaly rather than a trend. Notice here how the tech stack in China is unlike any other country in the world.

Most popular social apps across 10 countries

The digital ecosystem has evolved at a terrific pace but has mostly been contained within the boundaries of the country. On the other hand, none of the Chinese incumbents have been able to make real inroads globally. This is because of the 2 firewalls-

1) The Great Firewall of China - Chinese internet is a world unto itself, with its content closely monitored and managed by the Communist party. This makes it incredibly difficult (nay, impossible) for foreign technology companies to penetrate China.

2) The Cultural Firewall - The further the cultural distance between two countries, the more challenging it is for companies in one to compete in the other. Nowhere is this more prominent than in Chinese companies trying to find their foothold in other countries. The cultural differences, especially the language barriers, and the varying design aesthetics (often driven by the differences in the rendering Mandarin versus English), have made international market expansion a real challenge for China.

But wait, TikTok has 2BN+ downloads globally. How did an app designed by two guys in Shanghai manage to run circles around apps ranging from YouTube to Facebook to Instagram to Snapchat in cultures so different from the one in which it was built? How did they build a company so good that it could pierce through the cultural barriers even when Alilbaba and Tencent couldn’t?

And more importantly- if Indian and Chinese consumers exhibit similar behaviors, what conclusions does Bytedance building TikTok hold for the current upstarts in the Indian ecosystem?

The ByteDance playbook

Their story begins in August 2012. When Toutiao launched, China’s mobile news market was dominated by censored, state-controlled media and web portals that served lengthy, boring text that was written and optimized purely for desktop. Many of the largest content providers had not yet developed mobile apps or mobile-friendly sites, meaning that true mobile-optimized information and entertainment was rare. Toutiao launched as smartphone use was taking off in China: mobile penetration increased from nearly nothing in 2010 to 65% by 2014 (Note how it grew from 65% to 99% in 2020). There was a huge disconnect between the growing mobile internet penetration and available mobile-optimised news channels. This is where ByteDance stepped in and implemented its playbook.

1. Identify the Pain Point - They realised the absence of mobile-first optimised news platforms and stepped in to fill the gap via Toutiao. None of the incumbents were doing this.
2. Build a great product with strong user retention - Toutiao started as a personalized, informative, and a mobile-first app. From the outset, Toutiao was extremely easy to use – all it took was a download. There was no need to create an account and password, to link it to social media, or to provide information on interests or preferences. It was also the first time various news articles were aggregated in one place. It used aggressive push notifications and frequently nudged users to share content.
3. Data Mining and Analysis - This is the magic sauce of Bytedance and if it sounds eerily similar to what they did with TikTok, it was absolutely intentional. From the early days, Toutiao tracked information about each user - their taps, swipes, time spent per article and location to power the recommendation engine which was built on top of its reinforcement learning algorithm. It then used the user's profile data (age, location, gender), content (if the article is trending, whether it is long or short, and the relevance of the article), and context (localities like geography, weather, local news, etc) to serve a personalized newsfeed. The use of machine and deep learning algorithms at this stage of the content lifecycle is what sets Toutiao apart from its peers, and is key to driving continued user growth and retention.
4. Growth Strategy - You can have all the technology in the world, but without an addictive product there is no data, and without data, no algorithm can make the system better. It changed the titles, cover images, and even shortened most articles. This got users to an 80% read rate on each article in less than one day, which contributed to the 45% lifetime user retention it boasted early on.
5. Monetization - All roads lead to Rome Its core business was profitable in 2015 with $220MM in revenue. It also increased the rates it would charge advertisers because it could selectively place ads in front of the highest potential customers. In 2019, Meituan Dianping spent 85% of it’s marketing budget on Tiktok + Toutiao.

Toutiao had reached maturity in 2018 with about 200MM DAUs and ByteDance realised that it’s future growth would come from applying this playbook to new use cases and products.

Enter: Short-form video apps.

 A timeline of ByteDance’s products

Taking off from ByteDance’s playbook, we have observed that you need 3 pillars to build a company in the short-form video space. We call these the 3 rights to win. If you have these rights, you will come out as a winner, despite the intense competition. We’d also argue later in this piece that a lot of players in the Indian ecosystem currently are yet to achieve the first right to win. 

Till then, here are the 3 rights to win in this space-

1. A solid product, technology stack, and the use of data science - To some extent, technology is the product here. The personalization of your feed depends upon how well your algorithm understands content and context. This involves creating outstanding creation tools, building on top of an interest graph, having the tools to collect the data and analyse it, and finally using this data to create a hyper-personalised news feed of videos.
2. Growth Strategy - Effectiveness of a machine learning algorithm isn’t a function of the algorithm alone but of the algorithm being trained on some dataset. The very types of video that this algorithm needed to train on weren’t easy to create without the app’s camera tools and filters. For an algorithm to become as effective as it has, the content produced on the platform became its own source of training data.
3. Monetisation - It is often pointed out that India is the MAU/DAU farm for global apps. It’s a great place to amass users for your platform and grow vanity metrics, but monetisation has been a tough nut to crack in India. Traditionally, monetisation has been achieved through relevant in-feed ads. India’s digital ad spend totaled $1.93BN last year, a tiny fraction of the $108BN advertisers spent online in the U.S, the world’s largest ad market. Add to that another hurdle: frequently found non brand-safe content on these apps hinders the ability to attract a lot of the traditional brands to advertise on their platforms. Monetisation is not a given, it remains a right to be achieved.

Interestingly, ByteDance’s corporate team does not follow your traditional department structure based on business lines. Their internal team is split among only three core functional departments: User growth, Technology, and Commercialization. These are well aligned with our 3 rights highlighted above.

Source: Attention Factory: The Story of TikTok and China's ByteDance

The 3 rights to win

Let’s try to understand this better from a TikTok perspective and then apply this to the upstarts in the Indian ecosystem.

1. A solid product, technology stack, and the use of data science

This brings us to the most important piece of technology Bytedance introduced to TikTok: the updated For You Page feed algorithm. Remember how Bytedance realised that their technology from Toutiao needed a new use case? This is it. Note how Bytedance has been perfecting and iterating their algorithms for 7 years now. In the short form video world, your technology is your product.

Thus, all elements of your platform should be carefully crafted to strengthen the algorithm.

All the elements that strengthen TikTok’s algorithm

The algorithms powering your feed form the core element of your platform. There are two kinds of reinforcement algorithms:
a) Explore - An explore algorithm tries to broaden your exposure to more than just what you’ve shown you like
b) Exploit - An exploit algorithm will show you more of what you like and have watched before

Prior to TikTok, YouTube is said to have had the strongest exploit algorithm in video but in comparison to TikTok, YouTube’s algorithm now feels primitive. Before Bytedance bought Musical.ly and rebranded it TikTok, its Musical.ly clone called Douyin was already a sensation in the Chinese market thanks in large part to its effective algorithm.

At the time when Musical.ly got rebranded as TikTok, its use case was quite niche: it was dominated by teen girls doing lip sync videos. TikTok, aiming to be a generalist video platform, sought to attract an increasing variety of subcultures. That’s where the combination of Bytedance’s enormous marketing spend and the power of their algorithm came to the rescue. To help a network break out from its early adopter group, you need to bring lots of new subcultures into the app (that’s where the massive marketing spend helps) but also find ways to help these groups to a) find each other quickly and b) branch off into their own spaces. How does one do it? By harnessing the power of interest graphs.

All the popular apps you might be using on a daily basis - Facebook, Linkedin, Twitter etc. are built on social graphs. The onboarding process mainly revolves around you filling up your profile and interests and the platform nudging you to add more friends. The content popping up on your newsfeed is mainly prioritised by the accounts you most interact with, not necessarily what content and interests you like from the accounts you follow. The long-term idea here is to approximate an interest graph network from the foundational social graph. The problem with this is multifold- it assumes that the user is going to stick around long enough on the platform for the approximation to kick in and negative network effects that kick in at scale for social graphs. The problem remains that you’re rarely interested in everything from any single person you follow. This leads to context collapse.

TikTok skipped the approximation process and directly built its platform on an interest graph - merely by watching some videos, and without having to follow or friend anyone, you can quickly train TikTok on what you like. This is where the beauty of its algorithm comes in. Its algorithm is a hyper-efficient matchmaker that assembles the graph without imposing much of a burden on the user at all. Since 98% of all videos on Tiktok are only 15 seconds long, this significantly helps shorten the feedback loop and also sets up the stage for virality in these bite sized videos. By personalizing everyone’s FYP feeds, TikTok helped to keep these distinct subcultures, with their different tastes, separated. Compare this to Twitter, a much more volatile platform, all because subcultures with polar opposite views aren’t kept separate from each other in their own bubble by the algorithm leading to frequent clashes.

2. Growth Strategy

a) Marketing Spend - To put things in context, at over $100BN in valuation, Bytedance is currently the highest valued startup in the world having raised over $7BN. So, it simply opened up its wallet and started spending on user acquisition. TikTok spent over $300MM on Google ads in 2018. It is rumored to have spent $10MM/month on Google Ads in India. The ubiquity of TikTok ads was mind-blowing. At the peak of it’s spend in September 2018, 22% of all ads seen on Facebook’s properties by Apple devices in the US were for TikTok. Eugene Wei put it perfectly - “If Bytedance could have purchased ads on the back of my eyelids at sub $20 CPMs I don’t doubt they would have done so.” One thing becomes clear - This is a sector where a sky-high spend on marketing is inevitable for the platform. In a business with such high cash burn, being well capitalised becomes your moat.

b) Subtle growth hacks in the product - After two loops of a video, the share icon automatically lights up and starts blinking. To build good content, you need robust creation tools and camera. But what stopped users from simply using this app as an editing tool? Simple - You couldn’t export a video off TikTok unless you had uploaded it. Even when you do export it, it comes with it’s famous watermark directing more traffic back to the original platform.

c) Strong focus on the community - It made creators first-class citizens, giving them full-service support that included weekly content suggestion emails, 1-on-1 demos with TikTok’s staff, and live in-person events to create a community and encourage collaboration. This is the most overlooked aspect of building a platform and all the upstarts need to have a strong focus on finding their community and then engaging and empowering it. It started out by focusing on a very specific community of lip-sync videos, slowly growing to be the generalist video platform as we know it today.

3. Monetisation 

 Even for a giant like TikTok, Monetisation has been tough to achieve in India. Although its journey was abruptly ended, TikTok India recorded a revenue of merely $13MM in FY20. It was expected to do $500MM in the US in 2020. This has also been noticed in other ‘Bharat’ apps with Sharechat, ShareIt, and Likee recording abysmally low revenue with regards to their scale. The only players who have been able to figure out monetisation through ads successfully are the Big Tech companies. For comparison, YouTube India with 265MM MAUs recorded a revenue of $267MM in 2018. Monetising traditionally through ads hasn’t proven successful in India, and Asia-Pacific in general. To overcome this, video content platforms are now monetising through hashtag challenges, brand takeovers, and most importantly video commerce. But as things stand, this is where TikTok struggles in securing the third right to win.

The Indian Ecosystem

Since the government’s decision to ban the Chinese apps, it’s been fascinating to witness the sheer pace at which things are moving in the short-form video space in India. In a market so fiercely dominated by TikTok, there is a new ruler waiting to be crowned. Broadly, these new entrants can be classified into two groups-

1) New offerings by Indian media / digital content companies - These are relatively new players to the market. Wanting to capture the 200MM users void, these offerings are a good way for the parent company to capture an attractive opportunity and further strengthen their footholding in the consumer internet space by cross promoting their various assets. TakTak by MX Player, Moj by Sharechat, Josh by DailyHunt, Roposo by InMobi, HiPi by Zee5, and Reels by Instagram are the biggest names here.

These existing parent companies have well-oiled scalable channels for distribution and well-entrenched user bases. Moreover, they already have the experience and know-how required to scale consumer social products, possess a ton of data on Indian users, and are well-capitalized. Sharechat and DailyHunt have raised $263MM and $125MM respectively. Access to capital not only solves the marketing problem but also lets them invest in technology by hiring the best available engineering talent. We must not look past the experience that these platforms have accumulated in growing which gives them a significant lead over various other upstarts. Several of these platforms have actually tried to launch similar platforms at the time of TikTok’s entry to India. While they failed then, that experience will undoubtedly hold them in good stead.

Despite these possible upsides, we have struggled to find product differentiation within all these players. Their UI is identical and reminiscent of TikTok and as a user, the experience within all these apps remains largely similar. While this works really well in the short-term for user onboarding, they need to evolve with time. Having a TikTok-esque UI also leads people to expect a similar quality experience and product. Users reward you for building a good product, not being the first to build it. Which begs the question - is building a TikTok 2.0 the answer to fill the void? If yes, once TikTok does come back, will their growth stagnate and their audience churn out?

2) New Startups - Although, yes, some of these players have existed months and years before the ban, these are the players who witnessed a sharp rise in their users post TikTok’s exit. In stark contrast to the second group, these startups are not backed by any big Indian media companies. Players like Mitron, Chingari, Trell, Bolo Indya, Changa etc. are the top players in this group.

It’s an uphill battle for them because they’re up against incredibly well funded players already making inroads with ubiquitous marketing (I think we have all seen Badshah dancing and singing on TakaTak) and yet to be differentiated technology. One of the ways some startups have tried to differentiate in this space is by going after a niche community - beauty, lifestyle, travel etc. The problem here is targeting and finding your audience soon, otherwise they risk facing the musical.ly fate. The other way is going down the longer-form content videos (>90 seconds) where you’re competing more with YouTube rather than TikTok. We think this battle won’t have a single winner. The user’s attention will be fragmented over several of these apps due to the niches created.

We keep comparing them to TikTok’s technology stack because we believe it to be the gold standard in recommendation engines. We know it’s rather unfair to compare their technology to that of TikTok, who took almost a decade to perfect it. While we are not expecting them to immediately match up to TikTok’s reinforcement algorithms, all we want to see is some evolved product and an engaged community in the short term, which in itself is a tall order. Because of how nascent the space is in India and thanks to the plethora of competition, it is hard to filter out the companies who are simply riding the ‘Make In India’ hype wave. Thus, they are yet to achieve the first right to win.

 It’s very easy to get carried away by the number of downloads, but the metrics which truly matter are user retention and engagement. They are miles behind what TikTok boasted of and the user churn is rather disappointing even in an absolute way.

Source: August 2020 numbers from Morning Context, Economic Times

Wrapping Up

This space is still very nascent in India and hence, there is a lot of room to grow and evolve. There are new updates and launches every day. Our major problem with this space is exactly this - it is hard to find product differentiation and clarity owing to its early stage currently. To put this visually, let’s try to imagine them on Gartner’s Hype Cycle.

Gartner’s Hype Cycle

The sudden change of status quo with the fall of TikTok has opened up a plethora of innovation and competition.

Currently, we think these startups are at the top of the curve at ‘Peak of Inflated Expectations’.

What lies ahead of them are very interesting times. We truly hope they come out of the Trough of Disillusionment with more differentiation in product and experience, proven technology, an empowered community, who are engaged and retained. Meanwhile, we’ll be tracking these startups and the space, hoping these changes happen and the platforms evolve over the next few months and achieve the rights to win.

Epilogue - A déjà vu?

While we were researching to understand the space better, something really struck our eye. Till early 2018, local startups like Clip App (short video), Samosa (short video), Roposo (short video) and ShareChat (social networking) were blossoming. In early 2019, as competition heated up, they needed to consolidate to hold their power. Samosa shut down soon after. Roposo was acquired by InMobi and Clip was integrated into Sharechat and later shut down. The player who they lost market share to and were hence forced to consolidate?

Bytedance (TikTok + Helo). Go figure.

During our deep-dive into this short-form video space, we came across a few incredibly insightful pieces which set the foundation for how we looked at this space and its nuances, understood interest graphs, and tracked the growth of ByteDance. This has helped us apply their learnings to the Indian ecosystem. We can’t recommend their writings enough. Thank you Novak Turner, Eugene Wei, and Anu Hariharan for your brilliant work!