10 Years of Product Thinking on WeChat & Big Announcement from Extra Buzz!
Dear Extra Buzzers,
Happy Belated New Year!
I have a very big announcement to make! You might have noticed that it's been a whole month since I last posted on here, so what have I been working on? Well ... we are remaking Extra Buzz! It will be a community / forum in addition to a newsletter format.
I am still in the early stages of deciding what features to include, but as loyal Extra Buzz subscribers for our first experimental year, you are invited to become a beta tester for the product. Reply to this email and if you are one of those people who like to be early into everything and don't mind error messages while I work out kinks, I will put you on the list for early access!
Therefore, I will be discontinuing Extra Buzz on GetRevue. Existing subscribers will get their first month free on the new platform, tentatively called "Insider," the cost of which is TBD, but will be a much fuller offering than the current newsletter format. The current timeline is for a launch in late February, but that is just tentative for now!
The hope is to create a place where we can learn from each other and discuss topics in depth, versus just a one-way broadcast from myself. There are lots of experts already within this community, and I believe, as I recently quoted from Li Lu's excellent book on value investing, that 1+1>4 (your knowledge, my knowledge, our exchange, and the new knowledge we co-create as a result of that exchange!).
PS I wrote the below on a beta version of the new product on Wednesday, before the release of WeChat 8.0. However, I think the key questions stand. The format gives you a glimpse into how future write-ups might look (again, less broadcast, more interactive).
Best,
Rui
Allen Zhang: 10 Years of Product Thinking at WeChat
As widely anticipated, Allen Zhang, the creator of WeChat, gave an extensive 2-hour speech celebrating ten years of the app and primarily focusing on Channels, the video function within WeChat. I am including a point-by-point summary below along with commentary from myself in italics. There are a few questions at the end for all of you as well. If you have not listened to our Tech Buzz Episode 34 on the man many consider to be the best product mind in China, I urge you to do so now.
To level set, let's look at WeChat's current daily stats：
1.09 Bn daily users
330mm or 30% use video chat (30%)
780mm or 72% view Moments feed
120mm or 11% post to Moments
360mm or 33% read articles
400mm or 37% use mini programs, GMV +100% YoY
200mm or 18% use Channels
Why Video?
In the last 5 years, users are sending 33x more video messages in chat, and 10x more videos into their Moments feed. Not everyone can write longform content for the popular Public Accounts distribution / subscription system in WeChat, but far more users can shoot a short video.
The problem is that when WeChat tried this in 2017, they realized that it would require a whole new account system, since the current Public Accounts registration system is quite strict, and frequency limited. It was made that way intentionally as one of Allen's pet peeves was SMS spam, and he wanted to make sure WeChat could never be used that way. So the project went nowhere. But by 2019, it was so apparent that video was the next big thing and simply could not be avoided that they formed a small task force to work on it.
This is pretty late as far as entrances into video go, but it should be noted that within Tencent, there were multiple efforts at competing with Douyin in short video, so in my mind, it is understandable that WeChat took a later stab than its siblings, especially given Allen's notorious reluctance to add major functions to WeChat without a clear long-term strategy. (The two revivals of Weishi and Pengyou, both discontinued in 2017 and rebooted in 2018 and 2019 respectively to go head-to-head against Douyin, failed to gain much traction.)
The most important thing to realize here is that this is a fundamentally new direction for WeChat because Channels is designed to be an interactive public forum, totally different from the closed, private social network that has characterized WeChat thus far. Put simply, WeChat Classic mostly required bidirectional consent because it was based on "friends not followers." Yes, I can subscribe to Public Accounts or mini programs, which is a "follow" action, but one of the main values created by the WeChat ecosystem for these actions and why people build on top of it -- the social sharing -- is only to my friends.
In order to reconcile these two different scenarios, Allen and team decided to create an entirely different identity for Channels. Anyone with a WeChat ID can create a Channel ID if they wish -- so it is much lower barrier than a Public Account -- but the identities will remain distinct from each other. That is, I can follow my existing friends' Channel accounts, but strangers' Channel accounts I follow will not become my WeChat friends. The downside is that this is a bit complicated. The upside is that Allen has kept intact the strength of the WeChat ecosystem (its social relationships) and indeed leverages that to solve the cold start problem. Immediately, not only can I let my existing WeChat friends know to follow me, but their likes become "friend recommendations" to their friends. And friend recommendations are much higher signal than generic likes. We have seen that social network apps in the West either fall into the primarily closed or primarily public category, and of course this greatly influences the type of content you would share. WeChat is trying to do both. The solution is a bit clunky, I agree, but it is a thorny problem, and Allen's logic, in my opinion, is very sound.
Another central question is the matter of recommendation. Should it be people-driven (ie by your friends and follows) or machine-dictated? WeChat incubated 3 internal algorithm teams -- presumably to have them compete against each other for the best efficacy -- and found that friend-recommended content outperformed initially. But it was not because machines don't work -- it's because when content on the platform is still relatively thin, machine recommendations will underperform.
As we know, ByteDance is effectively 100% machine-driven. (Please listen to Tech Buzz Ep. 72 with Eugene Wei for a more detailed explanation on how this insight led it to breakout success.) But that is at least partially because it did not have a social network it could leverage. And Allen's reasoning that machine recommendations really only work well when the platforms are already rich in content is supported by the fact that ByteDance's first successful app, Toutiao, effectively grew its initial userbase by kinda sorta pirating content (it was sued by multiple outlets but eventually worked out distribution deals with all of them). WeChat, on the other hand, has more choices and can implement a hybrid solution. Allen has also been openly skeptical of machine recommendations in the past (Chinese), and so it is interesting to see him introduce more nuance in his views. I wonder if the continued dominance of ByteDance has changed his mind to be much more open-minded about algorithms.
But what will the actual distribution of content that people watch be? Allen predicted it to be 1:2:10. That is, the typical user would watch on average a ratio of 10 videos recommended by the algorithm versus 2 videos liked by friends and 1 video from accounts they follow. So:
1 Follow (user-chosen) What I explicitly indicate interest in & follow
2 Friends (people-recommended) What my friends are watching & enjoying
10 Hot (machine-recommended) What might be interesting to me, at large
Incredibly, the current ratio sits at about 1:2:9. Allen's instinct was very close!
I don't know if the ratio will stick, since it's still early days, but I found this to be one of the most important insights of the whole speech. Yes, this way of segmenting the video content introduces complexity for the user, but it also does reflect, in my opinion, three very valid and different needs. Whether or not this is the best way to physically display them -- see my screenshots below -- is not something I've made up my mind about.
When you tap into the Channels function, the current default is "Friends." It shows livestreams your friends have seen on top and tells you how many friends have watched it, although it does not tell you who they are. This stream is scrolling based. Videos only play when you pause your scrolling. Then, to either side are your follows and your machine recommended content ("Hot"). Follows & Hot are very similar to what you'd see in Douyin. But the Friends Feed is something WeChat can uniquely discover, and thus is why they've made it default. The question here is, why do so, when it will only be 2 out of every 13 pieces of content that the average user consumes? Why not start at the rightmost "Hot" machine recommendation section, since that will account for the bulk of usage?
WeChat Channels function.
Allen's answer is that the types of content will differ. Machine recommendations will serve up broadly entertaining content whereas Follows and Friends will probably have higher practical value and be more centered around knowledge. This is a view he's held for many years, by the way. When people are accountable for the content they share -- I can see exactly how many friends have liked the content on this feed and who they are -- the idea is that they will share more quality content. And even though this isn't an explicit "share" -- it is merely a like -- just being able to attach a real, known person to that like is going to drive the content a different direction. At least that's the thesis. My personal experience so far is that there is a pretty big divergence between what shows up on my Friends feed versus the Hot feed. The Hot feed is notably worse, but the Friends feed isn't great either, merely tolerable. So while I can get onboard with the logic, I'm not convinced the execution has followed.
You might have also noticed that display format differs slightly from feed to feed. Followed & Hot are swipe based, showing one video at a time, just like Douyin / TikTok whereas Friends is scroll based. This was a result of iteration. Initially, the videos were half screen, so as to give users a chance to choose what to watch. Then, as the "hit rate" -- the core KPI at the time, it seems -- increased in 2H 2020 and the team grew more confident that users were finding content they liked, they began testing full screen. Full screen led to more views on the Followed tab, and fewer on the Hot tab, which means there is more work to be done on the algorithm. This anecdote of the product evolution as user behavior changed is also why I'm reluctant to give an absolute answer to the question of "will WeChat Channels succeed?" The velocity of learning here may be more important than a snapshot of the product. I also like how this detail reveals that maybe different algorithms are needed for each section.
It's also one of the many examples of Allen's obsession with finetuning product instead of focusing on other aspects of the business. He dismisses "operation" heavy Chinese companies in his speech multiple times. IMO, it is a clear diss on companies such as ByteDance and Pinduoduo who are known to rely as much on their customer acquisition & retention strategies as they do on their products.
A long aside here on "operations": The common saying is that if you're looking for technology, you go to Baidu, product, Tencent, and operations, Alibaba. Operations is a term that effectively accomplishes the twin goals of user growth and retention and encompasses anything that is necessary in order to accomplish these goals, up to and including organizing events, content marketing, brand partnerships, negotiating and executing on subsidies ... etc. While it is data-driven, it is not always software-driven and can in fact be highly manual, ie chatting with a single customer at a time for building rapport, and so day-to-day responsibilities can look very different from Western counterparts in similar sounding roles such as "growth hackers.“ In fact, most of us with dual US-China work experience think there is no real Western equivalent to the 运营 (yunying) or operating role in China. The fact that salaries for such positions are comparatively low (starting at sub-$1000 per month) and aren't super high-skilled per se signals to me that a reliance on operations will not go away any time soon. Some products do not show their natural advantages until they are at a certain scale or have had enough time to change user behaviors. While such products may die an early death in the US, Chinese companies will do whatever it takes, even if it's extremely time-consuming and "operations heavy," to reach takeoff velocity & maintain altitude. Allen's point is that this is an artificial boost / intervention and that he really believes the product should succeed largely on its own strengths. It's a noble ideal, because it attempts to reduce the power of capital in the equation and focus on innovation, but it is probably overly purist and unrealistic for the current climate. Whether Allen likes it or not, "operations" is very much a competitive advantage, and might even be the core of many a Chinese internet company.
If it is not clear by now, Allen wants Channels to be a place where everyone can create. At the very least, your friends will see your stuff. But what does this mean in terms of integration with the rest of WeChat?
Moments. The team had to think about whether sharing a video to moments should be more like serving a copy of of the video "file," like how we think of sharing photos currently, or as a pointer to the content in the cloud, where other users' interactions to the video as much a part of the video as the images themselves. Allen strongly believes that the answer is the latter, and that the concept of files will die off. I thought his argument here was pretty convincing. Static copies of files does seem like a very old fashioned way to think about content.
Long vs Short Video. Allen doesn't want to distinguish between them, and thinks that long videos should simply show up on the platform with its first minute as teaser, and the option to click through to view the rest. In fact, Allen's speech itself (~90 minutes) is available for view within WeChat in exactly this way. Douyin and its competitors are indeed struggling to bridge the gap between long and short video formats. Allen wants to do away with the distinction altogether. It's contrarian, but also makes sense. A lot of my favorite short videos are clips from longer ones -- why not give me the option to click through to the full piece if I'm interested?
How Channels Show Up On Your Profile. It's simply by most recent. Same as with Moments. Allen and team felt that if people were given the opportunity to show "favorite" moments or videos, they will never change the content they choose to showcase. And it will be come very boring very quickly. Since they've allowed folks to change their Moments visibility to "last 3 days only," a lot of folks have chosen this option, which really decreases the amount of information that people can know about each other. It doesn't sound like they're planning to allow that for Channels. Again, this is more of a public space.
A user profile page of someone who has activated a Channels account.
Leading us to the next big reveal, which is that Channels is not just about short & long video -- Allen sees livestreaming as the next big content trend. Livestreaming is even easier and lower barrier than video. "It's almost just like chatting." This is pretty anticlimactic, since all the short video platforms have huge livestreaming businesses, and in fact, livestreaming accounted for over 2/3 of Kuaishou's $4Bn in revenues 1H2020. And we've covered livestreaming multiple times on Tech Buzz (Ep. 7, Ep. 68, Ep. 70).
There's already Livestreaming from People Nearby when you click on the Discover menu, but the next step will be to add a more direct pathway to Livestreaming. I presume he means this will be in the Chats or Me more "primary" menus inside of WeChat. Additionally, there will be a Chinese New Year's promotion around WeChat livestreaming, since the traditional house visits are much better represented by video than red packets. As you recall, WeChat Pay rose to prominence because of a genius Chinese New Year's campaign around digital red packets (Tech Buzz Ep. 34). This year will probably see many folks stuck at home due to the resurgence in COVID. It seems to be an excellent time to take advantage of an existing cultural tradition. I personally can't wait to see what Allen and team have planned.
Discover page currently defaults to Livestream.
Finally, as you would expect, Allen is totally against the idea of using money to lure creators onto the platform, as ByteDance and other Chinese internet companies are fond of doing. In yet another jab at "operations-heavy" strategies, he thinks a successful platform should be able to let its creators earn money without having to subsidize them. "We will never pay for content." I'm not sure WeChat needs to. It's only the creators who have signed platform exclusivities that are affected, IMO. As for everyone else, it might be easier to get some views through your friends' networks than trying to compete with every other creator out there via Douyin's algorithm, for example. And brands? Well, they are already running their mini programs here ... the ability to integrate all their sales activities where they were already doing a lot of marketing seems like a solid selling point. I do think that the mini program is the most important innovation of WeChat at this point, and it is one of the few true platforms that brands can build on top of in China, completely different from the centralized model of Alibaba. It's what's behind the biggest "invisible" trend in China in the past year, of course, the search for "private traffic," which we've talked about extensively in Ep. 75 on e-commerce SaaS.
That ends the bulk of the speech on Channels. The remainder are some points that Allen raised that give a glimpse of future features, but which I think are far less important:
Expressions. WeChat launched a "tickling" function last summer, which is now used by a few tens of millions of users per day. It's part of Allen's wish that WeChat interactions be more like physical interactions. He would also like more dramatic effects, such as animations that "blow up your screen." It is understandable that users are getting bored of just sending messages.
Statuses. WeChat should better reflect the dynamic state of its users. Allen wants to give people a place to write something brief about how they're feeling, and then discover people based on that.
Music. People are spending so much time on video now they aren't really listening to music anymore. In addition, music players have looked the same for years and years. Something could be done here, especially around the concept of showing "what someone else might be seeing when they're listening to the same song."
Less floating windows. A necessary evil, but very ugly. WeChat now has a "Read / watch later" function so you don't need to keep on switching back and forth between your chat and what you're reading / watching.
Input Keyboard. WeChat does not spy on its users messages. But many users in China use 3rd party keyboards & some of those programs do. WeChat is developing its own input system. Consequently, this was exposed a potential security vulnerability in Signal recently.
Development Speed. WeChat hasn't slowed down; it's just stayed lean. Allen still prefers small teams. The full Channels team (with its 200mm DAU) is still only 1 or 200 people. This is impressively small for China, where Series A companies can exceed this number of employees, but I'm not sure it says that much. IMO small teams don't always mean greater agility. It's still all about how you organize and lead the teams. Management is a science!
WeChat in 2 words: Connect and Simple. This perhaps got the greatest pushback from the public, especially "simple." The sheer number of functions available aside though, it remains the only app that I could teach my now 88-year-old grandma in one go.
Overall, I'm a lot more bullish on Channels than I was a few months ago. I'm now interested versus disengaged. It may still fail -- WeChat Time Capsules, anyone? -- but at least I understand why this is the direction Allen is going in.
My Key Takeaways:
WeChat is serious about Video / Livestreaming. It is however, going to go about it in its own way, leveraging its unique, irreplicable asset -- the largest social network of close / real-name relationships in China. This is where everything defaults to -- what your friends are liking, watching, saying. This is its competitive advantage and won't be change any time soon, if ever.
It wants ALL the content. All types, all formats. Consistent with its past decisions, it wants to be as close to a platform as possible, and not a medium. Whereas apps like "TikTok" connotes a specific type of content (punchy, meme-y), the strategy for Channels is to be as open-ended as possible. It wants to be the container for all your public-facing expressions, and video / livestreaming just happens to be the lowest barrier, highest opportunity. Seeing all the other video platforms bump up against each other as they try to expand beyond their category -- Douyin into longer formats & more serious content, Bilibili breaking into the mainstream -- I do think this is the right decision for WeChat, which is used by basically every single netizen in China. With that broad of a user base, there is really no option except to start off with accommodating everyone, because everyone is already on your platform. You cannot start off with appealing to a narrow niche community when you onboard a new function onto a billion-user app. It does have to work for everyone. It's true that Channels is a new initiative, but it cannot operate like a startup like Douyin did.
Allen is determined to be patient and grow the ecosystem organically by iterating product instead of relying on operational "hacks." He can afford to; WeChat is indispensable in China. I'm less certain about this strategy, but it has worked out well for him in the past -- case in point: mini programs. Which brings me to my next point ...
Channels is an evolving product. I hadn't realized how many changes and experiments had been made in just the last year, its first. My personal content feed is still fairly mediocre, but I'll be keeping an eye on it to see if it improves. If not ... I guess I'll blame my friends with poor taste? 🙂
And lastly, here's a parting thought that keeps on nagging at me, though it's not an immediate threat per se. Channels (and everything else within WeChat, really) is built around the real-name social web of relationships that it has accumulated over the last decade. It is definitely something special. But are we so sure that in the long run, follower-based platforms aren't just as powerful in establishing a sense of connection? Aren't Gen Z already having 100% virtual relationships, a trend that's currently accelerated by the pandemic? (Yes, I know that the article is talking about the US, but you'd be surprised how applicable it is to China.) I know I've made a lot of friends online -- some of you in this community I talk to nearly daily and don't think I even know your full name -- and it all came out of follower-based social media. I'm not saying that the follower-based platforms are facilitating connections well -- they're generally pretty terrible at it, although they keep on trying, ie ByteDance Duoshan -- but supposing one of them eventually figures it out ... might there be enough of a change in how younger users especially relate to each other that WeChat's most valuable asset is no longer that valuable? And is that day sooner than we think? Maybe more of an opportunity for someone else than a threat to WeChat at this point. Let me know what you think!
Questions for you:
1. What are your main objections to Allen's speech? Specifically, what do you disagree with or think are his blind spots re: product development and strategy?
2. What do you think is Channels' biggest risk?
3. Do you think Channels and Douyin are not comparable, or actually directly competitive?
4. What are you excited about re: Channels, if anything?
Addendum: Since the above was written, WeChat 8.0 has been released and the specific points Allen talks about re: expressions, statuses, etc. are now in the latest version. You can check out my tweet below for a video of some of the changes.
Rui Ma 马睿
@ruima
WeChat 8.0 changes thread: all your emoticons are now animated, a few have full screen type effects.
I guess that’s cool? Not sure how my mom will feel about me throwing 💣 at her tho? https://t.co/kOkk5SAOJS
8:23 AM - 21 Jan 2021