Millet founder KK Self exposure from Microsoft to the transformation of millet from the distance

on February 22, 2014, in the "micro-innovation CEO camp-into the Millet" Innovation forum site , Millet co-founder Huangjiangi (KK) said: "We all say that killing chickens should not use a sledgehammer, but our idea was that because we were going to kill chickens, we had to use a sledgehammer, and we had to make sure we killed faster than anyone else." KK spent 3 hours at the time sharing the experience as an engineer, and from his Microsoft to the Millet experience, and said that mobile Internet innovation has indeed changed .

Since 1996, KK had been in Microsoft for more than 15 years of engineers, he did not understand Google, Facebook's rapid rise in the start-up period, and later in 2010 joined the millet, currently his focus on the millet router. KK in the Forum site exposure Millet and Microsoft completely different product development model, he thinks Millet's word of mouth is: from a single point of breakthrough-trial-and-error-collect user feedback negative Word-of-mouth-and then iterate-try again and again-until positive word-of-mouth-and finally homeopathy, will be the single point to achieve the ultimate.

KK warned at the time that the next wave of subversion would be smart devices. From the subversive to turn into a subversive, in this 18 years of KK oneself have what innovation sentiment?

The following is a KK speech dictation:

At Microsoft has been destroyed three views

From the earliest 1996, I joined Microsoft, in this industry for 14 years, learned some engineering or product aspects of valuable experience to share with you.

In June 1996, after two rounds of interviews with 7 senior engineers, I joined Microsoft as an intern. Luckily, you might not have imagined that when I was an intern I even had the opportunity to go to Bill Gates ' family party.

What I was first aware of at Microsoft was the awe of the heart, how to say it? Here are some interesting stories to share:

The first one was when I started working for a week, completed the first task, less than 200 lines of code modules, they sent 3 senior engineers, plus my boss, four people spent 1.5 hours to see, the result is that the 200 lines of code every row is picked up. It makes me realize that Microsoft is very heavy on engineers. In addition, Microsoft to people's training than I expected, I did not expect them to spend so much time to develop an intern.

The second little story is a 2000 spring night, 10 o'clock, it took me two months to finish the most complicated code I ever had, and then I showed my boss how much faster it was than before, and he said, "Yes, how do you do it?"

I told him my algorithm, my data model and so on, he said after listening to say you This method is wrong, do not fast enough. Then turned around to open his computer, began to change in the code I wrote, until about 11:30, and then all the tests passed, performance than I changed two months of code also rose 15%, before 12 o'clock completed a series of work.

This at that time gave me two shock, first, the world than you strong people are many, genius is many, talent is many. Second, because he is stronger than you, so he is your boss. It was in 2000 that I realized why the company had become such a great company.

The perfect development model?

In fact, every company in the pursuit of the most perfect development model: Start with a few months of planning, then design, design for months, and then develop, the process may only take up to 24% or less of the entire cycle, and the final test is a vast project, at Microsoft, no finished product is not available to release.

After the release of the product is the user really began to use this product, in such a development model, the product cycle will often become a "product month." In the end, every new version of Windows is almost never done in two years, not just windows, but even small products.

I felt more pain at the time, I don't understand why a start-up like Google, such as Facebook, or even some other small company, is so small that it can grow so quickly under the competitive pressure of Microsoft, and even surpass Microsoft in some ways, I don't know what the problem is.

Microsoft's talent and internal process, at that time in my view is basically invincible. To give you a simple idea is to do windows with five thousand or six thousand developers doing each version of Windows. As you can imagine, five thousand or six thousand of the top software development engineers, divided into groups, each with five people, 311 equipped-that is, every 3 engineers will be equipped with a product manager, a test.

At the back of 311 there was a 0 that I added myself. Do you know what the 0 in this 3110 represents?

It's the voice of the user! At Microsoft, the first three numbers no matter how many times you zoom in, how many people do such a project, in the development process, the number representing the user is zero.

Because Microsoft has been pursuing the "most perfect" development model, it is a development model that makes it impossible for you to make mistakes. Every cycle is so rigorous that no one can make a mistake in implementing the plan. But there is a problem in itself.

Why isn't Microsoft running so fast in the new world? Should we rethink the development model?

The perfect opportunity to reflect: the creation of millet

2010, I met a great opportunity for reflection-we started to do millet, a new beginning, the biggest advantage is that we can all over again, what we have done before we can rethink, how is our current mode of need.

What is the core problem? Let's just list it:

1. Product definition. How do you know that the product definition you are doing is right, how do you know that the function you are doing now is the most important, this is the first question.

2. Cost. The second question is how to be most effective, the lowest cost, with the fastest and most stable way to do engineering iterations, because the cost of engineers is the highest, you know, your unit inside, no matter what industry you are responsible for research and development of those employees they should be the highest cost.

3. Test method. In the past, Microsoft was looking for thousands of or even tens of thousands of people to test the team to make the work perfect, before the product sent out. The worst thing about this model is that you can't make a mistake. Now the establishment of millet, we can make a lot of low-level errors, to allow us to try and error as soon as possible, this is I think the mode with Microsoft the biggest difference, before that is to prevent mistakes, now we are encouraging, and even this model will be more and more powerful because of trial and error.

The product development secret of Millet: the Law of public praise

Next i from the product development point of view to talk about the total Internet seven words: Focus, Ultimate, word-of-mouth, fast.

One of the most important key words is which word of mouth. Because there is no word of mouth how fast it is useless. Speaking of fast, is embodied in the rapid practice of our ideas. Many developers often fall into the mistaken idea that their first edition has to do a lot of functions, to be completed before the launch, in fact, this is not fast enough, and complaints is also very high.

Our advice is to try one point at a time in a fast experiment. After this point is launched, regardless of your user volume is 1 users, 10 users, 1000 users, 1 million users, you will get feedback. If the negative on the rapid adjustment, the test, if the word of mouth, the next should be homeopathy and for.

Once you've found the user's needs, do it in a deeper way. To what extent, to the extreme.

"Single point of breakthrough-Trial and error-user feedback negative Word-of-mouth-and then iterative-try again and again-until positive word-of-mouth-and then take the advantage of the single point to achieve the ultimate."

So in the entire development model, we emphasize the lightweight input to the new market. We were very excited to get this conclusion, because Millet could try a whole new way to do the product.

Six major landing methods:

First, talent landing: Because we want to kill chickens, so we need to use a sledgehammer.

Millet new development mode specific how to spread? I think it's all about the details. If you encounter new opportunities, we judge this direction is reliable, in fact, upfront investment does not need too much. But here's the premise that this lightweight input, every member, we need to find the strongest people in their field.

When we started millet, we found the top 10 people we can find the strongest people, we said that killing chickens should not use a sledgehammer, but our idea was that because we want to kill chickens, so we have to use a sledgehammer, and kill faster than others. And these people are not killing chickens before, they are killing cattle, so they have greater potential.

As I said, my code has been optimized for my boss at Microsoft, he thinks there is a better way, I believe that if I go to him again, he can do better, I need such a person, encounter difficulties do not find excuses, the dictionary does not "have tried before, no", this is the most critical.

If you want to be lightweight and try a direction that you think has great potential, be sure to find the strongest person, or you don't know if these things are going to work out. Find the strongest people to do, do not I also recognized.

Other such people, their foundation has been very good, so can be in the rapid development of the situation, still know what can be omitted, what should not be omitted, rather than blindly just how quickly how to come. If this is the case, I think we will take many detours.

Second, cultural landing: embrace open source.

The second is to embrace the open source culture, which is actually a long-term mentality. If you do not embrace open source culture, millet will never do much. If we can not use inclusive heart, open heart to do millet, everything will be done from the beginning.

Many open source technologies that have been used by a number of companies in tens of millions of, hundreds of millions of users, so we don't need to start over, some things are worth reusing, and we want to give back to the entire open source community, starting last year, We have engineers to join the international Open source organization to become their members.

Third, the team landing: The core of the post, let users to do.

Ladies and gentlemen, what is the most difficult position in the world? Product Manager is really very difficult to recruit, why? Because a good project manager can be developed, but the product manager itself needs to have a very high savvy, but also to realize that not only he himself how to think about this matter, more attention to how users will contact the matter. So far, I think the biggest bottleneck in many company staff is still product manager and designer.

I was talking about 10 engineers. The better the product manager, yes, in our 100 product managers may only two of the millet company, in addition, there may be 98 is actually the first wave of millet mobile users, the most feverish of these users. 1 million users, there must be 100 users he is the product has his own ideas, and it is very love this product, his ideas can drive many other users.

The 100 "Product managers" into our development model, this is the key, the more critical point is the business of the general manager or the person in charge, he himself will become the number one product manager. Every business general manager must love his products, if he wants to do his own will not use products, he does not good.

Four, iterative landing: Large data analysis can be, users spit more directly!

Now we all talk big data, we will also through large data analysis what function is important, what function is not important. Although large data analysis is important, I actually believe in the user's direct spit.

If our large data analysis to a lot of users are particularly concerned about function A, but Weibo I was scolded the most is function B, I will not hesitate to solve the first time function B demand, I believe that the user direct feedback, spit slot.

Perhaps you would say that these users we do not pay, why do they help us to work? Sometimes the user is very simple. If they recommend a good function to you in the forum, after the good idea, you have done this function, told everybody this idea is he thought, guaranteed that he immediately the next day to help you to think another 10 better ideas, the fulfillment is a very big impetus.

Five, Test landing: Find enthusiasts, gray test in fact, the best way is to launch a large amount of users to help us do all the testing of the direction, so that not only the most efficient, the lowest level, but also the highest coverage.

What does millet advocate? We can judge what aspects to revise and what to improve according to the user's biggest pain point. More important question, which one do we fix first? There are 100 questions we should fix first 10.

Of course you will say KK you do this is not a big risk, and now the user is not small, if you so like to use the national user to do the test words out of the problem how to do?

We can use a grayscale upgrade, that is, to select 1% of users through a number of conditions, let them upgrade, and then observe their use. Test to make sure that they do not encounter problems after you turn 1% into 5%,10%, 20%, 50%, 100% all open. Of course there is a risk, but with 1% of the feedback you can basically find most of the problems.

Of course you have to let these users know that they are VIP users, they are testing with you, they are only a taste of taste, they can choose not to this 1% of the test users, but believe me, some users will use various means, is to advance the early adopters, there must be such people, these will become our testers.

A lightweight project has been used from 10 to 100 people to 1000 people are now 10,000 enthusiasts, we speak a lot of enthusiasts, why enthusiasts so important? The reason is simple, because we picked a wave of the most stringent, most picky users, you can not pass them this, You don't have to go on, but you pass them, they'll be your biggest word of mouth, because you've had the hardest times, like the 18 bronze people I saw at the Shaolin Temple, and you went out after the 18 brass and you were basically invincible, so we were very focused on finding these enthusiasts. Finding them is actually for our development.

Vi. responsibility Landing: Start with the user as a friend.

We advocate the user is our friend, our friends sound like a virtual, we just put God and friends for a change, is not very virtual? From the perspective of an engineer and a product manager, what are their greatest pressures? One of the most stressful things an engineer can get is that his friends tell him you have a problem with the product, and the designer's biggest fear is that he goes to a friend's party, and his friend tells him, "What did you design?" Why do you use this color? It's going to start challenging his taste, you know?

An employee, the biggest pressure is not necessarily from his boss, colleagues or users. If he is the user only, he will not worry, he would feel that this is just the job content.

But you will find that if a friend calls you, tell you that your phone has a problem, can not go out, or not charge, you will want to drive to his home immediately to send him a new mobile phone, I hope he don't think you this thing is not reliable. So we found that the original force is so big, the power of friends is more than all the engineering model.

In the past, we tried everything to make sure that we don't make mistakes in a very complicated process. To the end is a friend of a phone call, or a friend in micro-Borrie said, you will immediately to fix the problem.

Case: How to light the development of millet router?

Next to share some of the cases, the company last year gave me an excellent opportunity to verify that the new development model of Millet is not in the software industry is feasible.

This is a router, and we make it an adult toy. It is also the eldest of a family intelligence center. This project was actually started last (2013) April, when I talked to Ray about it, just a few minutes of communication, and then Ray always said to do it.

Then, from April to July, it was basically my team-building process, where I shared two small stories.

First, the router must be a hardware product, I should find the hardware master, but I think not, just now, I said, kill chicken to use a sledgehammer, although we do is a home router, we should look for enterprise-level router talent, because I do not just do the router, I have to do the future of the data center, it will become more and more powerful.

I'm looking for a router for more than 10 years in the enterprise, who have I been looking for? Don Mu. Tang Mu is basically Tencent's user interaction Center boss, he 10 years ago, Tencent started the user Experience Center, the industry to do the design of basically no one does not know him, I ran to find such a cow, I said to Millet bar.

If I ask him to come over and do the same thing he did more than 10 years ago, I guess I can't impress him, but when I tell him, do you want to come over here and do a router with me, and this will impress him. This is a challenge to him, he heard, on the excitement of the mess, asked me the first question is whether the router should have a display, otherwise my UI how to play out?

Then I told him, brother is not like this, your screen everywhere, you see what we do in the Millet cell phone, do the millet TV, do the Millet box, they are all have screen, we are now what era? We control a device that doesn't have to run past him, you use your cell phone, your phone's screen is your UI interaction, Even TV, what you want to do in it, or what kind of interaction, there is a lot of space to play, so basically he decided. And then drove over to report, very excited to do the router.

Why do we do "millet Router 1 dollars public test"? In fact, this is to achieve before I said how to let users participate in testing for us. such as the network equipment such as routers, China each province, each county, each city inside the network basically strange, how to ensure that the router of the millet in China can be captured in every corner? The best way is to get the test.

The price is 1 yuan, this is an attitude, because I really appreciate these users, they are also very happy, spend a dollar to buy a high-end router, and then there are 1 T's hard drive, as well as the Thunder Platinum Membership card, value, this is an attitude, they know I respect them.

What is the second attitude? This is a toy, what does it mean? We want users to play with it instead of just using it, and we encourage them to think about how they're going to play the router, how they're going to use the hard drive, how to use other hardware, and the software we're offering them, All of a sudden you will find that we have a lot of "testers" and "product managers", they are beyond my expectations, there are even professional designers.

You really have to admire the design of the user, very beautiful. The open mode of allowing users to participate in product testing is reliable. At present we have done two rounds of the public test, we will do the third round, and the third round of products have made some adjustments in front of the two rounds, so the first two rounds of the value of the public test I have got.

A lot of talk today, is my own past 18 years of a course. We had the opportunity to do things that others didn't have the chance to do, and I was really lucky with my team. Can make their own use of a very cool product than any other return to power.

How does the traditional enterprise overkill?

Gold wrong knife: KK said a lot of software development and hardware innovation of dry goods, just now I feel particularly valuable is the use of Microsoft's past development model to the development model with millet to do a contrast, is simply blood and tears of experience and lessons. More is Microsoft's blood and tears, I would like to ask, whether it is to do traditional software or hardware, overkill this is the pattern of millet, many people can not use, some may be able to use. I wonder if there is a specific proposal for software companies how they use the Internet to innovate?

KK: I do not recommend a mature company to promote such a new model, will be affected by all aspects of the pressure, which basically all of the staff have a very comfortable way to do things, so I suggest that if you are willing to try this model, you can choose a little out of a small project, And find someone who really has a sense of identity, do not need to give too many resources, so that they have the resources to find themselves, from a small to cut, to see whether he has the results, if the results of the words to enlarge it, in the company's various departments to promote the deduction, I think the effect will be very big, A month or two may have been immediate.

Gold wrong knife: now many people in the hardware, a lot of hardware companies have anxiety, especially afraid of the layman suddenly killed in, although the layman does not understand the industry, but killed in. For hardware companies, what advice do you have if you want to use the Internet as a way to innovate?

KK: I think the hardware company you go back immediately can organize discussion, if you get this concept, you have to subvert your domain what you will do, what are you most afraid of? This is the first step. The second is to discuss it, after you really have to turn your nightmares out of the possibility of subversion, and talk about Internet technology, I can guarantee that no matter what the field, at least three or more companies are trying to use the Internet model to subvert the area you are doing.

Where's the next tuyere?

Micro-Innovation students: In the beginning of the 2010 millet, Ray is always the way to persuade you to join the millet business?

KK: I was back in 2005 from the United States, 2010 years just with Ray always do this thing, the answer is actually very simple, I think later Ray always used the tuyere to describe the trend, I feel really is very correct, in fact, why was it from the United States. The first reason is that I think Asia is the outlet, this is the first. Why choose Mobile phone, especially mobile area, because this is the air outlet inside the Chinese tuyere, because I see these two are opportunities, so did not hesitate.

I was with Ray always meet, chatted for several hours about products, mobile phone these, are chatting products, no other, but I go, I was very impressed, I and Lei embarrassed I have things to go first, but no matter what you do, count me.

Q: Asia is the tuyere, you are back, the phone is the tuyere, you also caught. Where do you think the next tuyere is?

KK: Each industry has the tuyere of various industries, but if you ask me I see that tuyere, I have no doubt that the smart device is a immediately to explode the tuyere, but also include smart home, any device to intelligent it, he will do something. including air-conditioning, soy milk machine This, this is actually why I do.

Q: We do smart home and community businesses, now there is a big confusion in hardware iterations, whether our development center or our chip center, as well as our stock center, it will have a relatively long period of time, in the millet this piece, the hardware here to iterate and update the product is how to do? Thank you.

KK: Hardware has a natural feature is from your design to open mold, to access the board, it must be in the middle of the cycle is more than the software, change that, I do not have a good plan, we can only do the middle can be iterative process to do faster. But there is a suggestion, because you are doing these hardware, I believe that the software is certainly part of the inside, the software can actually be synchronized with it, many traditional hardware development is the hardware to complete the software, but in many cases does not need to do so.

Source: Innovation Faction Dictation: KK (Huangjiangi, founder of Millet)