Monthly Archives: February 2021

About Processes

People who know me know that I have a biased view of processes – and that’s probably because I’m working for Facebook at this moment, and generally speaking larger companies have more processes than smaller ones. While I have a lot of ideas while discussing with my coworkers about processes, I’d like to put down some more structured thoughts in this article.

First of all, what I believe that makes a company / organization / team successful boils down to two very fundamental things:

People, Domain.

  1. Key people is a multiplier to an organization. They unlock potentials and make impossibles possible. They are people of integrity who would consider faking or even taking shortcuts unacceptable – and they would quit if forced to do so. They don’t blindly follow instructions, and have their own thought of thinking. They have the domain knowledge, and willing to master new domains whenever needed. They ask rest of the org to be the same of them, or at least a portion of them.
  2. Domain defines how well the people power can be unleashed. Even super talented person cannot fight against trends. A hot domain may make someone less talented to look like they’re super talented. However, it’s worth noting that if you have the right people, talented people can find the right domain (think the shift to mobile and the companies who succeeded and failed)

The rest plays much less critical roles. Even with a bad process, the good ones will still figure out how to navigate it and find the answer, while a good process cannot magically turn something shitty into great.

Steve Jobs had a great comment on process vs content – which I take the content as “the content great people produce”. I can’t agree more with him.

When I think about the greatest CEOs in the tech industry, they all seem to care a lot about details, and they have a lot of thoughts on them. Take Elon Musk as an example, even as busy as an CEO (or CEOs), he still knows how a rocket works in details.

So am I against processes? No. I think they are important, though from a different way.

Processes are there to make the best use of the talented people.

While we want to work with as many talented people as possible, the sad fact is that we only have a few talents. Processes exist because they give them power, or reduce the friction of having the talented people taking care of the rest. Aka, letting the great people to become more efficient in creating content.

Let’s take an example. A team used to have unstructured roadmapping processes. This worked great because we have a great TL who understands product and engineering. This worked until the team started to grow, potentially by 2x or 3x – the TL can no longer cover every roadmapping if nothing changes.

To address this problem, we can create such processes:

  1. Make sure all the product ideas are vetted and ordered upfront.
  2. Make sure all product ideas have well-documented plan so that it’s super clear to the TL what needs to be built.
  3. Have this TL go through the docs and give early feedbacks on each project.
  4. Have a review session to finalize on the roadmap.

In this process, 1 reduces prevents low quality ideas to go through the roadmapping session. 2 reduces the back & forth between product & engineer, and also pushes product to think through the project in more details. 3 turns the back & forth into an async communications and 4 ensures the “eventual consistency” between product & engineer. This process introduces more work for the product team, but increase the value of time this TL puts into it.

While this example this talent is an engineer, it could be anyone – including yourself. It could also be an important highly productive team in the org.

Processes don’t make sense if you don’t have talents on the team, or they’re not a constraint resource. Processes increase the amount of work for people overall – and only when we can shift some work from the talented to the rest of the team processes start to make sense.

Alright, to end this post, here are my two principles on processes:

  1. Processes are useless if you don’t have the talented person.
  2. Built processes for that person so they create more great content.

There will be only one winner in Food, Drink & Local Services

Players in high frequency need can beat low frequency need players extremely easily.

Yelp will definitely agree to this statement – they’re getting killed by Google.

Food is the highest frequency need, then drink, then other local services (grocery, autos, karaoke, kart racing, kayaking, wedding, etc etc).

You will probably know you can also order drink from the app where you ordered food, rather than an app where you ordered grocery delivery, even though the nature of the business are not much that different.

Doordash will almost certain enter the business of grocery delivery, killing instacart and shipt.

Doordash has to face Yelp, if it wants to keep its high evaluation after COVID. Yelp is currently only 2.67B while Doordash is 65B. Buying Yelp will be a smart move that will almost pay back immediately after COVID.

Ride sharing will be the next one. Isn’t it nice if you can send some riders to restaurants, then immediately pick up some food from the very same restaurant (or in the same plaza), deliver it to the residential area, and then getting another ride request from the residential area to the business area?

The list goes on and on. The path is clear, the rest is just execution. Doordash had amazing execution in the past (and strategy as well for sure!) – otherwise they would not have won the delivery war.

可能“恰饭”才是互联网广告的未来?

恰饭,就是大V或者UP主收费发软广。有时候这种软广不是单独发出,而是夹在正常的内容中间,通常伴随着满屏的“防不胜防”的弹幕。

互联网广告的发展,从最开始的牛皮藓广告+弹窗,到后来的信息流广告,广告的样式也变得越来越原生:原因很好理解:原生广告更贴合平台内容,更容易吸引点击。与此同时,广告的定向性也随着效果广告的发展而变得越来越重要。然而定向广告需要收集大量的用户数据,而这个商业模式正在被不同程度的挑战着:

  1. Apple 举起隐私大旗,Android 厂商将被迫跟随 – 小米是第一个对此下手的 Android 手机厂商,但绝对不会是最后一个。
  2. 欧洲等国家因为这种商业模式收不到税,开始以各种理由限制对个人数据的收集并加大罚款力度。
  3. California Props 24 的通过表明普通人的隐私意识已经开始觉醒。

定向广告将会遇到越来越多的困难,与此同时“恰饭”类型的广告,已经开始慢慢崛起:

  1. 比原生广告更原生,融入内容。很多创作者的恰饭广告质量也很高。
  2. 用户与创作者的联系让用户甚至很乐于看这样的广告。
  3. 不依赖于精确的数据收集,但是同时又具备一定的定向性。

主要的问题还是在于广告的门槛变得更高,对于新广告主,试错成本变得很高,而且质量控制困难,scale 也比较困难,如果这些问题能被解决,可能恰饭广告的春天就会来临。写一写我的基本思路:

  1. 广告主提供模版,创作者只需要按照模版读广告词即可,或者利用 AR 把 UP 主形象放在模板里。
  2. 创作者预先录制好相同长度内容不同的广告,可以在播出时进行简单的定向投放。
  3. 中间平台进行广告的质量控制 – 类似于抖音的小范围分发来确定质量 帮助创作者制作出效果好的广告