• Uncategorized
  • 0

How To Do With User Growth?– My Practical Experience With Uber And Mobike

In this article, I roughly collected my past experience in Uber growth teams, such as Ed Baker, Andrew Chen, and mobike’s business experience, and try to avoid revealing any trade secrets, and talk about the current thinking in the industry. The passage is very long, and I divided into the following four subsections:

1.What is user growth? 2. What did the company do during the outbreak? (mobike example) 3. What is the system of mature companies(Uber example)? 4. What do you do from zero to one?

Over the past year, many friends have told me about their startup ideas and asked me what else I could do in the post-Internet age. I’ve always said that 2017 is the point at which online traffic prices reach a critical point, and it’s time for startups to fit into one of these two:

A. or you can extract high value from each user(High LTV, such as cash loans, big games)
B. Or there’s a way to get customers that’s so cheap that you needn’t pay for it.(Low CAC)

If you can’t find a reasonable way to realize the unit economy, or do not have an unique channel to get customers, do not resign easily, returned overseas, the future of the Internet industry belongs to the giant (TATBMD ranking), not entrepreneurs.

For example, a friend used Xu Xiaoping’s angel to invest as a skill payment platform (Contrast Thumbtack) ,it was very delicate and well run, but it took nearly 100 yuan to obtain an effective user, and there was no natural amount to be done,so he pulled up stakes last year. Similarly, strategy games are now on the front page of a download-activated CPA ($120), which is expensive enough to cast doubt on life.

This is why everyone brushes the bar, sees the advertisement is almost all krypton gold mobile phone game (plays the blue moon, the Chu liuxiang), the English training and the second-hand car. Without exception are in line with the condition a I said, to a user can make a good deal, so can afford to pay for direct advertising.

Why did mobike and ofo get up in 2017, because it qualified condition b. In fact, the two companies rarely advertise for information flow or effect ads for app stores. Most of the effective passengers are from the subway entrance, bus station, road side of the body of the QR code. As a result, both companies hit a large number of subway stations, bus stop signs, to enhance the main access channels, so that users choose theirs rather than competitors when they are out of the subway.

In the post-mobile Internet era, the most important means of production is traffic and capital, technology and products themselves have been difficult to form barriers (such as AI, VR/AR, block chain). Look Douyin and Muse, the two products look exactly the same, unit data (unit price, length of stay) difference of several times, a period of time after the number of users a thousand times. Although Douyin is a plagiarized copy of Muse(Musical.ly), but the apprentice wins. Sure enough, in November, the headline (the parent company) bought Musical.ly. the latter was typically a big deal, but fortunately, overseas users were big enough and the price wasn’t bad.

Readers may ask, this pixel plagiarism product data is so different? This is the difference between a team of approachable masters and a team that doesn’t enter the stream or is not approachable. In this era, having a solid growth team and growth methodology is crucial. In this article, I roughly collect in the past in the Uber growth team from Ed Baker, Andrew Chen and other master understanding, and mobike business experience, to talk about the current industry ideas.

What is “growth”?

Growth refers to the marketing of mass Internet products with the programmer’s mind. The first wave of Internet companies employed traditional advertising agencies and marketing directors, with more than half of the ads in the Super Bowl coming from Silicon Valley in 2000 and almost all of them closed down during the bubble. When Jack Ma promoted Taobao in 2003, he started with television ads, print media and outdoor advertising.

There is a famous saying in the traditional advertising industry: “I know that 50% of this advertisement is wasteful, but I never know which 50%.” Gradually, people found that Internet products can be done in new ways, allowing engineers to take the lead, using product iterations and AB tests to reduce waste, increase conversion rates, and eliminate “self-improvement ads”. At home, this is called flow playing. Slowly everyone wakes up and opens up their former marketing director. They are replaced by product managers, product operations and programmers.

How does the company do during the outbreak? (example of mobike)

I went through a period from 3 million to 25 million a day in mobike. Before the outbreak, the founding team had found a smooth way to get customers: the subway entrance swept the bike body QR code registration. Until I left mobike, this channel is still the largest and most effective new user source. With 299 yuan of deposit and more than a car, mobike achieved a magic positive cash flow model:

Cost per vehicle < deposit * number of deposits per vehicle
The guest can also get positive cash flow! Such good things, everyone in the dream of the Internet to laugh wake up. Because of this, old investment institutions such as Sequoia Kaoling Tencent entered the building.

Now that the unit economy has been verified, the second step is to expand the business while improving efficiency. So we continued to increase the number of cars, while we started to increase the conversion rate of activation.

In the new user registration process, we use Growing.io to see the drain funnel for each page. After reading clearly, decided to cut off the guide page, the previous four steps of mobile phone authentication, deposit, ID card, invitation changed to three steps. The deposit before ID card verification is a very nice design: for users, hundreds of deposits are handed in, privacy is still what? So the conversion rate is very high.

What I’ve just said is that app is already installed. In the actual usage scenario, the bigger step for conversion is to download the installation. All outdoors, traffic is not rich, see jump app storehouse, a lot of people close the page. Mobike was the first to make a WeChat Mini Programs, and in Mini Programs re-engraved the new user process, the conversion rate further soared. How exaggerated is this channel? For half a year, Tencent told us that more than 50% of Mini Programs’s total amount came from mobike.

Optimized for guest and activation, it is time for ascension to remain and active. There are many strong competitors and our primary goal is market share. Mobike, who was born in Shanghai, competed against Yang Yujie, a former RGM RGM in Uber’s North West end, against Chen Wei, who is a return from Uber’s headquarters. Mobike and Ofo are both experienced veterans. In 2017 we’ve all seen how veterans fight subsidies: training, recruiting meat, burning money. In this period of outbreak, we summed up all of these things:

Rush back-free day-fission free day-egg cart-treasure box car-red charter car v2-member-month card v2-red charter car v3-monthly card v3

Ofo is also basically similar, with the intermediate Ofo trying GXGY more than we did, but with modest results.

These moves, “everyone is a product manager” website has a college student summed up a particularly good, can read mobike’s operational strategy review and thinking.

Up to now, no one on either side was able to win the other side, but killed a small player,Bluegogo small blue bike, although the car is easy to ride, but the product growth can not be done, in this era also afraid of deep alleys.

To sum up, mobike basically conforms to the AARRR model, finding a good guest model(Acquisition), using various ways to improve activation(Activation), and then using the product matrix to do retention(Retention), active engagement) and good friend recommendation (Referral). Because the technical threshold is too low and the competition is too fierce, I think it is very reasonable to skip the AARRR revenue step. In this tuyere, only to step on the throttle to grab market share to further take financing, near the giant, the battle is not finished absolutely can not rush to do revenue.

Talk about a small episode , in this battle, speed and execution are of great importance. Both had a poor code architecture at the end of 2016 and had to drag a technical debt bayonet in 2017. Both stayed there for a week and a second, but the outages were too heavy, the overtime was not enough, and I had a hard time chatting with Ofo’s CTO for morning tea. Later, they couldn’t resist it. Both of them had a tacit understanding of changing it to two weeks and one shift. Mobike also had the opportunity to do a refactoring (using Kotlin to rewrite Android and launch the Manhattan project that Liu Yao and I were thinking about.) (the core UX/UI was revised); the client of mobike also won the 2017 design award of pea pods.)

What is the system of mature companies?( Uber example)

I used to code in the growth division of Uber headquarters, initially as a user active agent, and later as the leader of the loyalty Program. Our division’s director, Andrew Chen, is a blogger who has spent more than a decade writing about how to do growth, says Eric Ries, author of Lean Entrepreneurship the Lean Startup.His blog is “One of the best entrepreneurship blogs of all time.” Kalanick bought his failed company directly to get him into Uber. He met Marc Andreessen in his early years because of his good blog. After 17 years of infighting, everyone was depressed. Last month, Andrew left Uber to join A16z, Silicon Valley’s midday rookie venture capital firm. He made one particularly good point:

–all growth schemes are being used more and more indiscriminately, and their effects are getting worse and worse.(law ofshitty clickthrough rates)

Therefore, the effective method everyone hides private, any writing in the blog style, has become obsolete, not very good use. Overall, then, the growth sector cannot rely on one or two whimsical techniques, but must have a methodology for systemically discovering, validating, and promoting new skills.

When I joined the Uber, I had just started to reorganize, break up the enterprise structure that had been divided by type of work, and become a separate business department with a whole set of products, front end, back end, operation, data. I was going to follow my LinkedIn boss into the mobile group, but instead I went straight into the growth department and worked under Yang Ji. (Yang Ji later went to Guazi as a senior director and head of the growth department). In fact, the reorganization was led by Ed Baker from Facebook. There have been several subsequent reorganizations, but in general, our growth has been reduced to four teams:

Acquisition, activation, engagement, retention.

After Andrew came to see, the lack of invited friends, so added a group

3. engagement
4. Retention
5. Referral

Let’s go into detail about the KPI for each department, what attempts have been made, and what techniques are available.

Guest acquisition:



What about going from zero to one?

The biggest difference between established companies and startups is that startups don’t know if they have found Product/Market Fitps, nor do they have a clear channel of access. The core is to build the growth of this growing wind and fire wheel.

Anyone who has read lean start-ups knows that startups have one, and only one, way to die: the money is spent. At this stage, if you find PMF, the momentum to do it, hot money must be crazy to find you. So, the happy company has all kinds of happiness, the unfortunate company is all the same unfortunate:

–the money is spent, yet to find real demand and constant growth.

Do not grow, is the curse of the start-up company, as long as not growing, the user base will shrink, slowly more and more bleak, referring to Renren, Jumei. In contrast, Tencent, Facebook, do 1 billion DAU, but also rose, as if there is no end.

This aspect of finding real needs is too big to write another article. This is mainly about growth. How can we find a steady stream of growth before the money is spent? The old 80-20 rule and Andrew Chen’s law are getting worse and worse.

The 80-20 rule is that 80% of the customers of each product at each stage come from the same channel. The more bad the rule is, if you find a very efficient channel, you must not rest easy, good days can only be three months to three years, quickly expired.

These two laws, must cooperate with the start-up company’s capital chain to practice. Startups are so short of money that every penny has to be split in half. There are three steps:

1.Spread the net and try all possible ways to get customers at the lowest cost (no more than tens of thousands of dollars) and get the gold pipe that is reliable at this stage.
2.Open the golden pipe to the maximum, and turn on the silver pipe around you.
3.Before the golden hose goes out, repeat the first two steps

Creativity is very important, comply with the characteristics of the product Synergyis very important, before the most effective way is fission red envelopes (pinduoduo, DiDi, mobike), GXGYinvited GXGY (mobike); new methods emerge in endlessly. Remember, a good solution must be used less.

In addition, Cheng Hao said earlier that all successful startups had a wave of dividends, so the trilogy must have been done according to the times. Zynga could start with Facebook dividends in those years, and Uber could start with mobile Internet dividends in 2010. Last year pinduoduo can rely on WeChat dividend start, but the dividend of the mature platform is basically rubbed off, gold pipe must be in the new platform.


These are three levels of growth that I have summarized: start-up, outbreak, and maturity. In general, it is the initial period as frugal as possible to find the three axes suitable for their products, and then gradually transition to the maturity of a set of systems.
References: Understanding Uber: Driver Growth Loop http://ayeshamascarenhas.tumblr.com/post/141532139520/understanding-uber-driver-growth-loop
Stefan Benndorf: 5 steps to grow your app user basehttps://marketingland.com/5-steps-grow-app-user-base-lifted-uber-playbook-221812
Andrew Chen: How to build a billion dollar digital
Andrew Chen:
Startups and big cos should approach growth differently (Video)

Share to:

You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *