Ten thousand words long article: How to build an operational structure?

A good operational architecture , like a good R&D architecture, can support the business from 0 to 1, and from 1 to 1 million.

01 Our misunderstanding of operations-operations and marketing

When talking about the operation system, we have to talk about our misunderstanding of operation first.

The first is the misunderstanding of the boundary between operations and marketing

Now more and more companies are recruiting for positions such as market operations, event operations, brand operations, marketing growth, and user growth. What is the difference between these and traditional marketing jobs?

The biggest difference is that operations are typical Internet era (new economy era) data-driven marketing, supply chain, and product production work.

In fact, the term "Internet era" has long been out of date. We are increasingly not distinguishing companies by Internet companies. The Internet has gradually become a traditional industry, an infrastructure that the people love and a tool used by enterprises. The difference between modern new economic companies lies in whether they have a data-based mindset to deconstruct business segments and plan efficient business operation paths, and whether they are improving their advantages by understanding the industry and users rather than by monopolizing resources.

Back to the difference between marketing and operations, we can see that operations is becoming an increasingly broad term . On the one hand, the boundary between operations and marketing work is gradually blurring, and on the other hand, the boundary between operations and product work is also gradually blurring. For example, the popular concept in the industry recently is -

Growth hacking is a complete system based on data thinking, testing as a means, users as the core, and products as implementation plans.

We, the operators, are also gradually growing into business operators.

When talking about marketing and operations, we have to talk about the "4R" model mentioned in "Marketing Strategy in the Digital Age" co-authored by Professor Qiao Lin. In my opinion, there is no essential difference between the 4R model and the AARRR model.

4R Model:

1) Digitize the user's preferences, needs, age and other characteristics as much as possible, based on the user's digitization

2) The channels for acquiring users and the way they reach users are greatly digitized. Behind this is the traffic, conversion rate, etc. of these channels.

3) Maintain the relationship with users, operate the relationship and stickiness between users to make users trust, spread the word, and finally bring

4) Income

AARRR Model:

Be data-driven, constantly generate hypotheses, and prioritize the best way to experiment.

1) Get users

2) Activate active users

3) User retention

4) Get income

5) User dissemination

Every link should be digitized as much as possible

Above, let’s see if there is any difference. In fact, the two models are both talking about:

We need to improve the efficiency of acquiring users by matching users and channels in a more refined way. After users come to the platform, we need to recommend content in a more refined way to retain users and establish a longer-term relationship with users, so that users are willing to pay, spread the word, and stay.

02 AARRR&4R

I mentioned the AARRR and 4R models above. I believe most of you are still confused. When I started to use the AARRR model, I not only found that this idea is essentially the same as the original online traffic conversion customer price , but also found that Baidu’s business growth many years ago was using this idea of ​​hacker growth.

It’s just that we didn’t call it growth hacking at the time, we called it channel rating

On the contrary, because I built a data-driven, ABTest-refined channel matching strategy on Baidu’s mobile terminal, I also applied for an algorithm patent from Baidu.

At that time, I was the product manager in charge of this project. Wang Lin and Chen Hao were R&D personnel from Baidu’s big data R&D department who worked with me. This was a perfect setup for a Silicon Valley growth hacker team. I only realized it today. How lucky I am!

How did we do it at that time?

Before we talk about how to do it, let’s share some practical background.

1) Different mobile products of Baidu have different settlement prices in channel promotion . For example, even though Baidu Search, Baidu Input Method and Baidu Maps are all cooperating with the BBK mobile phone channel, the settlement unit price and settlement method may be different. It is possible that one product is settled according to activation A, with a unit price of 2 yuan, and another product is settled according to active DAU, with a unit price of 5 yuan.

2) Different products in Baidu have different contributions to Baidu for the same daily active users . In fact, the criteria for judging whether a user is active are also different. For example, a map user may be considered an active user if he/she is used once a week, but an input method user may be considered an active user if he/she is used once a day.

3) For Baidu, the most important thing is a high-value user . The value of this user is that he can bring more users, contribute more content and active behaviors, or click more ads, making the advertiser's delivery more valuable. All subsequent user behaviors are related to user activity.

Therefore, we assume that there is a strategic model that can balance the various indicators in the AARRR model.

The logic of this assumption is:

1) A channel can bring us new users, active users, retention users, revenue, and self-propagation.

However, at different stages of the business, the importance of these five things is different. By adjusting the weights of the importance of the five things, we can clearly determine which channel has a higher rating in our opinion.

2) If a channel gets a higher settlement price than expected, it will be more willing to direct users to Baidu , so that it can obtain more users on this channel (after all, at that time, a channel can direct users to Baidu, Tencent, Alibaba, and 360 at the same time)

3) If a channel is more valuable, we are willing to settle a higher settlement price than the average.

4) This settlement unit price is compared with the user value it brings to us, that is, the ROI per user is continuously better.

Therefore, all indicators that need to be considered are included in the calculation system

The user behavior data, wireless search data, and mobile cloud channel settlement data collected from all Baidu mobile products are used to form a model of user survival, activity, and contribution value. The comprehensive dimension evaluation is carried out in terms of user cost-benefit, user quality (retention, activity, and dissemination), channel promotion capabilities (new additions), and channel strategic value.

At the time, our assumption was that the settlement price would affect whether the channel was willing to give more volume to Baidu.

The way to achieve this is to adjust the parameters of the weights of data such as volume, activity, and user value, so as to influence those channels that are better at bringing volume or more active users to give Baidu more users.

Although the above examples are too crude, the truth is still the same:

Based on data and experimental iteration, we conduct rapid AB Testing to achieve goals such as user growth, activity, and monetization.

I think this may be the first patent for the hacker growth model , but it was too late to realize it.

Baidu’s channel growth ideas at the time can also be broken down using the 4R framework.

Recognize: Digitally identify users and form user portraits

The user bought a Xiaomi phone and first downloaded Baidu Input Method from the Xiaomi App Store. Later, he saw that the phone was pre-installed with Baidu Search, and he didn’t like it so he uninstalled it. After the user configures the phone, he starts playing games. If he downloads Baidu Maps while playing games, he can increase 200 points. So the user downloads Baidu Maps and occasionally uses it. Is this user profile vivid?

Of course, we can also go a step further and look at the user's behavior in different Baidu apps. For example, users always move between two points during commuting time, or users always enter some emoticons in Baidu Input Method. With data like this, can we have a more accurate judgment on user portraits?

In this way, no matter which product Baidu uses on mobile, we will know more clearly what kind of user this is. When we push, we can make more accurate recommendations. We will also know more clearly that if we want to wake up such a user, he will play this game, use the mobile phone input method, and use the Xiaomi App Store. We can wake up this user through this game and the activities of the Xiaomi App Store.

At that time, the Baidu mobile terminal operation system we developed within Baidu had such a function, which connected Baidu's mobile terminal products and some channels, and could integrate the overall data portrait of Baidu's mobile terminal to wake up users more accurately.

Reach: Identification of digital channel reach, such as which users this channel brings, their retention, activity, contribution value, and contribution revenue.

Since the settlement data is on our platform, we also know the cost of these users brought by these channels.

Relationship: Knowing which channel the user has a stronger connection with, and knowing which Baidu mobile product the user has a stronger connection with, we can choose to establish a more lasting connection with the user through this product, or recommend the user to use other Baidu products through this product.

Return: Of course, Baidu does not earn revenue directly from users. The user behavior contributed by users is the source of Baidu's revenue.

03 Our misunderstanding of AARRR

When I was communicating with a startup company, I found that they had a deep misunderstanding of AARRR. Later I found that many startups had this misunderstanding.

Let me briefly explain the background: This company actually only has marketing expenses and no operating income

So I asked them: The marketing expenses of tens of thousands of dollars every month have brought in so many users, why don’t they make any purchases?

Never expected

The marketing expenses of tens of thousands of yuan are all used to buy gifts for users. Users can accumulate points by checking in at the store, and the accumulated points can be exchanged for gifts

So I asked, why do we do this?

It turns out that there is such a thinking logic

First we need to acquire customers. After acquiring customers, we can know their needs. After knowing their needs, we can create good products. After creating good products, customers will stay. After they stay, we have good products anyway, so we can make money happily.

Totally wrong!

In fact, when a user walks into your store for the first time, if you cannot keep his heart and attention, he will leave. The same applies to online platforms.

at the same time

When you start a company, you need to have an initial MVP in mind in terms of what value you want to deliver to users, even if there is no market validation. Do you have any idea of ​​what value you want to deliver to users?

(Of course, I can’t help but say it a little more completely here. At least in the case above, the new company was right to acquire customers, but it used the wrong method and content. When we find that the team’s execution actions are distorted in the business, don’t doubt it. It’s the manager’s problem and the problem of the manager’s goal setting.)

We must retain users the moment they come to our platform and appear in front of us. Now it is no longer a competition for incremental markets, but a competition for existing markets.

Users are being distracted by a variety of services. What we are actually competing for is the user's attention.

If you can get the user's attention, the user will stay

Of course, behind getting the user's attention

1) What kind of product and service value do you provide?

2) How to package and deliver the value of products and services

3) Matching of this value and user cognition

Under the RARRA model, the most important thing is obvious - the value of the product and service and how to express the value

Because of the obvious value of product services and the appropriate way to express value, we can retain users.

Being able to "retain users" means that our core products have found our real user groups

The next step is to find ways to activate users, make them active, and let them spread themselves (is this the same as the Relationship in the 4R model?)

I believe that it is not difficult to earn income from a product that allows users to spread the word. It is not difficult to get more users by letting users spread the word.

Some people also mentioned that under the RARRA model, the first R can be Refer instead of Retention . We do see that more and more users’ first reaction when seeing a picture is to forward it, but we can carefully trace the user’s psychological state.

Before forwarding, do users want to get the value of this product ? If so, the user’s first R is still retention.

Before forwarding, do users not care about the value of the product , but only care about the actual economic benefits brought by forwarding? Maybe this business promotion logic works in some situations and for some user groups.

For example, in order to earn money from Qutoutiao, some people forward news from Qutoutiao, but this is not the norm. After all, there are so many news and content to read in Qutoutiao. Users "start" with forwarding for profit and "finally" recognize that forwarding a certain content carries profit value.

Under the RARRA model that is suitable for the current situation, the real growth will be as shown in the following figure:

1) Initial growth is driven by product value , because at present, we need to obtain the match between user cognition and product value, to retain users and to be willing to establish a continuous relationship with us.

2) Digital marketing is a very important second step . After a certain amount of users have been accumulated through the product’s value and the correct delivery of value, we need more exposure to attract new users.

The most important thing about such exposure is that the cost is reasonable. Therefore, if the data points are clear and sufficient, the input and output of advertising costs are continuously followed up, and the composition of advertising costs is optimized, digital marketing will have a good input-output ratio.

3) Brand drive will reach a turning point based on good product value and the continuous support and accumulation of efficient digital marketing . On the basis of advertising, the value of brands and products is also continuously accumulating exposure, and will eventually achieve a qualitative change.

Of course, I can also see that some excellent offline brands have also taken the step of digital marketing and reached the moment of brand drive, but for online business, digital marketing is always a necessary action that cannot be avoided.

Interestingly, this is the exact opposite of the traditional logic of brand growth.

Traditional brand growth involves extensive advertising and promotion, laying a solid foundation for the brand first.

Then there is the strong control of channels, giving channels very high commissions, allowing supermarkets and restaurants to stock the products, so that people who want to drink can only choose these wines in restaurants. To this day, if you want to enter traditional channels such as supermarkets, you need to have a gross profit of about 4 times to give the channels sufficient profit sharing space.

Finally, with the branding and strong channel coverage, users will experience the product and accept its value.

Of course, I believe that business in the new economic era is product-driven . Although the recent news about Hammer has made many people pessimistic about innovative products, user experience, and product-driven, I still have this belief - products can truly drive a company and a business.

Product-driven is not an easy task, because product value is not just about the momentary user experience, the user experience is continuous and growing.

Product value will be tested by time and space

Product value will also be tested by the iteration rhythm

Product value will be tested by collaborative networks - such as Xiaomi speakers controlling Xiaomi robots

The value of a product will also be tested by price, value, purchase path, and service details.

04 Our misunderstanding of data

Some time ago, when I was communicating with an operator, she gave me this opinion:

Data is used to beautify the market, and we prefer to listen to the specific feedback from users in the market.

But is this actually the case?

In some situations, yes

After all, we have limited resources and cannot get feedback from all users, but I always need feedback from users. There must be a user for me to study in depth what her preferences are and what kind of sales methods can impress her.

But in some cases, no

We often think that data is biased and indicative. Either we have chosen the wrong angle to look at the data, or the data collection is still incomplete and the data sampling method is biased.

Let's take a concrete example:

For example, if we are selling a facial cream but it is not selling well, we think there may be something wrong with our sales guidance. Wouldn’t it be enough to describe in an objective and rational way how this facial cream improves facial brightness through high technology?

Sure enough, I tried this sales routine and it worked very well.

So I came to the conclusion that our new sales strategy is correct based on this user's feedback.

But is this really correct?

Have you ever tried to tell users that by replenishing water with facial cream, the cells will be fuller and look whiter?

Have you ever tried to tell users that facial cream can remove blackheads, making their face cleaner and whiter?

Have you ever tried to tell users that they should apply facial cream before washing their face and use it together with cleansing products and essence products to make their skin look brighter?

Oftentimes, the effectiveness we judge is just one of thousands of effective methods.

Among the thousands of effective methods, which one is the most effective? You have to try it to know. This is the full amount of data.

At least there should be a hypothesis, and a priority of the hypothesis . Among these high-priority hypotheses, you will know only after trying them.

Misunderstanding of data

At that time, I thought that maybe startups were short of resources and manpower for data statistics, so they had to rely on subjective judgment. In order to maintain the persuasiveness of subjective judgment, they had to be very sure of this subjective judgment.

Later I found out that it was not the case. Some startups are superstitious about data.

In another communication between me and a startup company, the boss of a startup company asked me, "We already have the product, but how do we get the first step of traffic?"

I told him, let’s quickly test the market through KOLs who can bring products

He said, it makes sense. We know the boundaries of KOL’s sales and their user groups. So we can count the series of conversions, activity, stay, and consumption of users brought by this KOL, and we will know what users want from the data.

It's a beautiful thought, (^o^)/~

Many times, we believe in data and use data, but we cannot be superstitious about data , and we cannot let data make decisions for us. Does the sales of a product through a KOL, whether good or bad, really have anything to do with the user's final evaluation of the product?

uncertain!

Just like you will never buy a Yamaha keyboard at Walmart for cheaper than the one on JD.com.

Because it feels weird.

Humans are not just rational animals.

I had a similar experience when I was working for a startup honey brand called “She is Honey”.

Shemi’s honey is of very high quality and comes from the pollution-free Pingwu Ecological Reserve, so the price is three times that of conventional food.

At that time, I found a KOL (Key Opinion Leader) who I hoped could help us promote our products. This KOL was a KOL in the food circle. Dried tofu could be sold for 50 yuan a bag. I thought it shouldn’t be difficult to sell honey for more than 200 yuan.

However, the KOL rejected it the first time he communicated with me because the tone did not match.

How to use data to distinguish and identify the inconsistencies in tone

I disagreed. Your users are obviously very accurate. They love ordinary food with high unit price, have the ability to pay, and like to eat. Moreover, food and honey seem to be in the same category. If the KOL said that you really want to do it, then just do it.

As expected, none of them were sold.

Indeed, the users are very accurate. Perhaps these users are the most accurate users of "She is Honey"

However, these users may buy "She is Honey" in many places, such as boutique supermarkets and JD.com, but they will not buy it from this KOL because the tone does not match. Buying such honey from you always feels weird.

I strongly emphasize data-driven, and I have actually managed the construction of data systems in various companies. I am a product operator who grew up from the role of data product manager. Although I have a special belief in data, I still want to emphasize and remind

Don’t be superstitious about data, and don’t let data replace us in making decisions. Data is our tool, and so is our sensibility. Maybe at certain moments, the judgment made by sensibility is more accurate.

Besides, even if the data indicators are good, does that mean your business is definitely doing well?

05 Your misunderstanding of operations

Many bosses in traditional industries are looking forward to finding an operator who can solve current business problems with little effort, find hot spots, write popular articles, and make 100,000+ in minutes.

But in fact, this is also a temptation that is often difficult to give up.

1) You can write a popular article and get 100,000+ views, but what is the cost behind this and how many users can be retained?

2) You can find a way to achieve a lot with a little effort, but this is often because the original business was done too poorly. In fact, in the entire business process, a little improvement in the user experience process can achieve results.

A typical case is the love jewelry that was very popular with the "koi"

At that time, industry insiders calculated that the gift value of 100,000 yuan attracted more than 1 million people to participate (the final data was 1,220,845). Judging from the data at that moment, this was a very cost-effective activity, with an average cost of less than 0.1 yuan per new user .

This is also an activity that has sparked a lot of discussion in the operation circle (tempting activities, high-value gifts, low-threshold participation methods, and the concept of copying koi fish without market education)

But when we continue to pay attention to the follow-up, what is the user retention situation?

To be more objective, I selected articles published on the AiLian Jewelry WeChat public account on December 3rd, which is more than a few days away from the current date, and also selected similar topics for comparison - celebrity topics

Students who are new media editors must have this feeling (in fact, this feeling is also caused by data increment). Three days after the article is published, if there is no big V forwarding, it will be difficult to get more readings, so I chose the articles for comparison according to the above ideas.

Article reading and interaction volume before the event:

September 11, Zhou Xun topic, likes (user interaction) 31, reading volume (traffic) 3592

October 16, Zhao Liying and Feng Shaofeng topic, likes (user interaction) 110, reading volume (traffic) 24009

On October 22, the Koi event was launched

On October 29, the lucky draw for the Koi event

December 3, topic about Joey Yung, likes (user interaction) 35, views (traffic) 7420

The evidence is below:

So, let’s do a simple calculation.

An investment of 100,000 yuan, an increase in user traffic of more than 3,000, and an increase in user interaction of a single digit. The average cost of retaining a user is 3,000+ yuan, and the average cost of an active user is 10,000 to 20,000 yuan. Is this investment worth it?

It may be worth it. After all, in the jewelry industry, the average customer spending is high, so it can acquire users at a higher cost.

However, is this user traffic still worth it in terms of conversion?

This is not necessarily the case. After all, how many of these users who keep reading can eventually convert into sales? If the conversion rate is 1%, then the above cost will be multiplied by 100 times.

So, is this event a failure?

As an event with such a big influence, it must be a success first of all. What is really important is - what was our purpose in doing this event?

Is it to expand influence and create topics? Then we are successful

Is it to accumulate more users to convert consumption and gain profits? Then it's a failure.

Newly-started companies may also ask this question: Is it really useful for us to spend money to expand our influence?

Here, I also want to answer affirmatively, it must be useful

The expansion of influence can reduce the cost of acquiring users through other channels, increase the possibility of cross-border cooperation, and empower our content more effectively.

But if you want a quick return, and this return must cover the cost, I am afraid it will not be possible.

Here are two small points of knowledge:

1) We need to pay attention to LTV>CAC (Life Time Value>Customer Acquisition Cost), that is, the revenue brought by the user's life cycle should be greater than the cost of acquiring a user.

The cost of a user should be estimated as (the cost of expanding influence, the cost of acquiring users, and the manpower cost of doing two things) / number of users

Of course, if the posture of expanding influence is correct, it can also reduce the cost of acquiring users and increase consumption during the user life cycle.

2) Data can never be judged in a time slice . For example, at the moment of this activity, the data is beautiful.

But if we look at it over a longer period of time, are the judgments and evaluations we make based on the data still consistent?

A good operator has no magic tricks. He needs to have continuous management of LTV and CAC, have continuous assumptions about each key node in the operation process, and have continuous attention and follow-up on user psychological changes, communication points and consumption stimulation points.

Operation itself is a protracted war. Building a strong fortress and fighting a dull battle are the key to success in an operation-driven business.

Of course, the contempt chain from marketing and products will also reach operations, but it doesn’t matter.

Indeed, the starting point of operations is often those supporting work contents, but gradually, insightful operations will connect the various business departments of the company like lubricating oil.

Ultimately, we will be guided by business goals and design the most reasonable path to achieve the goals.

Based on data and experiments, we will continue to iterate and update to achieve faster growth and a longer-term future.

Many people are good at using operational actions to carry out small growth activities one after another, but they tend to overlook the use of operational systems to replicate, scale, and coordinate to truly expand and solidify the business.

A good chess player has no brilliant moves throughout the game

06 What should a good operational structure look like?

A good operational structure can replicate success, coordinate the whole process, and be quantified and controlled.

Only when it can be quantified and controlled can it be managed and operated

Of course, before we talk about the good operating system, let’s first talk about the common problems in the operating system, okay?

Company B is an unmanned shelf company. Their operations are placed within the shelf operation team, which means that the focus is on shelf operations.

The feedback mechanism of this matter itself is very slow

What is her feedback mechanism like?

Data collection is complete -> Data analysis gives the judgment result -> Which different products need to be placed in different positions on this shelf -> Procurement needs to be purchased immediately -> Delivery needs to deliver this product in time

But in fact, I could have a prediction from the beginning

A company with more boys and a company with more girls will have different products on their shelves.

The products that should be displayed on the shelves of a business company and an Internet technology company must be different.

So assuming that we can rebuild this operation system , I think a more reasonable way is that the operation team should be placed in the business team from the beginning.

When I signed with this company, I should have communicated and predicted with this company, and even conducted employee surveys in advance - you choose what product you want, and as long as you choose, I will give you a special price for the first order of this product. Anyway, there were quite a lot of special prices at that time.

At the same time, operations must be placed on the supply chain system

I want to influence the premium right of supply chain procurement based on shelf dynamics data

Including my distribution system, snacks are consumed very quickly in some companies, and very slowly in some companies. Plan the delivery route to maximize the efficiency of operations

The above are the problems that Company B encountered in operation scheduling and configuration at the business level.

Another one is Company C. The operating system problems encountered by Company C are common in many companies. The problem found at that time was that the KPI did not match

The reality behind this is that everyone has different goals. You do yours and I do mine. We each have our own KPIs, and when combined, they do not add up to the KPIs of the entire company. Many companies may have this problem.

Lack of overall goals. The team did not break down tasks around the company's overall goals. As a result, resources were scattered and efforts were not focused on one thing.

So how do you build a good operating system?

The basis is to have a sense of the overall situation at the business level and a holistic view at the data level.

In other words, it means having a good thinking foundation and data foundation

We need to build an operating system based on this

The operating system is a bit abstract, it is nothing more than the deployment of troops in the organizational structure and the coordination of goals.

, but what work needs to be done to put it into practice?

1) Use process management thinking to ensure that goals are achieved

2) Use B-end + C-end to balance supply and demand

3) Use metrics to align business goals

Here, process management is actually based on the in-depth exploration of a specific business. What are the key nodes on the path to achieve this specific business? This needs to be thought through clearly.

After thinking clearly, you should jump out and look at how to balance and collaborate between the B-end and the C-end of the overall business. Then you need to use data to abstract goals and use data to unify business goals so that the entire team can collaborate happily.

07 The foundation of a good operational structure

A good operating system is a complete business system . Usually, the business is composed of three parts: the supply side, the product service side, and the user side.

The value provided by our platform is mainly reflected in products and services. Products and services are for users on one end and for suppliers on the other.

We continuously refine our products and services and choose different ways and channels to deliver the value of our products and services to users in order to retain, activate, spread and consume our products and services.

On the other hand, our products and services are also provided by suppliers.

If we are an e-commerce company, our suppliers are our purchasing sources.

If we are a game company, our suppliers are our material designers, R&D or IP

If we are a consulting company, our suppliers are the consultants or teachers behind the scenes.

Behind all products and services there are suppliers who provide “this product and service”

We continue to encapsulate the value provided by these suppliers to form our product services

The operation on the user side is what we often refer to as the C-side operation, and the operation on the supplier side is what we often refer to as the B-side operation.

Therefore, to have a holistic view of the business, we must first be aware that our business is composed of the B-end, the C-end, and the product and service end.

So when we often say that we want to achieve operational growth, it is not just about user growth, but also product and service growth, as well as supplier growth.

Take Xiaomi as an example . When Xiaomi's growth was sluggish, it was not just about changing its channels that could improve user growth.

Because at that point in time, can the channel be successfully reconstructed? Can the value per user continue to grow? There is another very important factor behind this.

This is: Product (Or Service)

Whether the products (or services) delivered by Xiaomi to users can expand the original value is the key factor in whether Xiaomi can land in the new offline traffic scenario.

After all, users who choose to buy from Xiaomi stores and Xiaomi Homes may not necessarily be interested in such a "fever machine"

At the same time, the suppliers behind the products (or services) are also closely related to the value delivered by Xiaomi.

Xiaomi's growth this time, let's first look at the logic related to our current business operations

Xiaomi's growth can no longer be achieved simply through innovation in channels or changes in user acquisition, retention, conversion and monetization methods. Only by continuously expanding the "brand boundaries" and the "product value boundaries delivered by Xiaomi to users" can it usher in such a new wave of growth.

so

1) At the brand level , Xiaomi has broadened the boundaries of its brand, from "born for enthusiasts" to "letting everyone enjoy the fun brought by technology"

2) At the channel level , we expanded the number of Xiaomi stores in major cities and Xiaomi stores in third- and fourth-tier cities.

3) Product level : broadened product line, all-inclusive

This is the overall picture at the business level. Let’s take a look at the overall picture at the data level.

What data do we need to collect from the user side? This is based on user portraits and the channels from which users come, such as channel traffic, conversion rate, etc., as well as a series of indicators related to user portraits, such as user consumption sensitivity, service sensitivity, etc., which are also user-side data indicators in our data system.

If we take Xiaomi as an example, we can design data indicators on the supplier side.

The purpose of collecting data from the supplier side is to have a series of hierarchical management logic for Xiaomi's suppliers. Under this logic, it is necessary to collect and count some key indicators, such as the supplier's inventory, delivery cycle, quality, price, after-sales service, etc. Including the compatibility of my hardware, whether my chip can only be matched with this motherboard.

Supplier data is collected in full, and there are many dimensions for my reference. I can evaluate suppliers. I also need to design data related to product and service. Similarly, these data collection is also for - I have the ability to manage a hierarchical system for product and service.

Under this logic of hierarchical management, there will be some key indicators, such as inventory and price, including whether this product is an isolated product, whether it will be produced later, which category it belongs to, and what is its situation compared to its competitors, its cumulative sales, recent sales, conversion rates, and topicality are all data that needs to be collected at the product and service level.

It would be more interesting when we string together the data of the entire system . We may find that some users have very high price sensitivity when buying Xiaomi's power bank, but their sensitivity is not high when buying other products.

Of course, in addition to the data being viewed together, some students may have such questions - is it the person who controls the overall situation who needs to have a global awareness to see the data on the entire disk?

Of course, not!

If we do a specific activity, we also need such a global awareness.

When doing an activity, what products should be placed on it and how should these products be priced. At this time, the activity itself is a product and service provided to users. So how to design this event is actually how to design our products and services

In order to do this activity well, we should collect and pay attention to the number of views, click-through rates, participation rates, forwarding rates, etc. of the activity.

For example:

Recently, there was a knowledge festival event in Himalaya. Teachers of Mayu and Teacher Fan Bing are the suppliers. They provide their courses. The activity itself is the product and server side. The user side is undoubtedly the users participating in this event, Himalaya users.

If Himalaya strings this data together, it may also find an interesting phenomenon. Teacher Xiaomayu has a very strong ability to deliver goods, and the course price is average, but she can bring a lot of users. Teacher Fan Bing's course is a unit price higher than the average, but the user conversion rate is particularly high.

If we are the designers of this activity, we should balance how to recommend Teacher Xiaomayu and Teacher Fan Bing so that their characteristics can help me with this activity, which can not only participate in enough users, but also achieve good sales.

Here are two more knowledge points to remind you

1) All small things can have a more global perspective

Because any small thing has upstream and downstream, and has its users and supply side. Even if we do a small collaborative work in the company, there are objects we need to deliver, that are our users, and there are sources we need to support, that is our supply side

2) In fact, it is meaningless if the data is placed under a certain time slice.

Only by following the trend in a longer time dimension, looking through a finer granularity, and analyzing the correlation between data, is meaningful

08 Operational architecture based on the basis

The above briefly describes the relationship between the three modules of the operation system based on the basics. Here we will talk about the operational architecture based on the basics.

We must have strong process management to ensure that we can get the results of this business and achieve our goals .

What is process management?

In most cases, the traffic conversion rate is the average customer price = our sales volume

When my product is relatively stable, the average customer price may be relatively fixed. In this case, the key indicators of the business results I can get.

If I want to disassemble the process indicators, I should track how much my traffic has been done, how much my conversion rate has been done, and can I achieve the business results I want this year through such traffic and conversion rate.

Of course, is it OK if I disassemble the process indicators fine enough?

not enough.

Because in each specific business segment, only by breaking the goals that need to be completed into specific actions and following up the progress of this action can we ensure that the goal is achieved.

For example, I already know that the conversion rate should be 5% to be OK.

This is a very important process indicator , but can I ensure that it is achieved with this indicator?

I need to break down the 5% target into many things and projects to be implemented, and keep following up with the progress of these projects. The progress of these projects is the process indicator for me to achieve a 5% conversion rate.

Therefore, we can dismantle the process indicators that can be implemented and continue to follow up to ensure that we get business results

But what to do with collaboration? This ushered in the second step in building the operation system, the coordination between the B-end and the C-end

I just mentioned that many companies have problems with collaboration. Let me give you a problem with the supplier side of online products. For example, if you want to do an activity, you need to have a home page pop-up screen on the APP to do some home page traffic for the event to promote sales, because the KPI indicator for operation is that so many sales of the event this year.

But at this time, R&D stopped doing it. R&D said that my KPI was to control the time when the APP was opened to within 0.01 seconds. If you add a pop-up screen, I can't achieve my KPI.

(R&D is our supplier in many online businesses, which provides the functions of our products)

So when our process management indicators are broken down enough, we must start to pay attention to collaboration, pay attention to how the BC side balances supply and demand relationships, and get the business results that the company wants most together.

When we can balance the BC side well and have clear ideas for process management, we must use indicators to unify our business goals.

09 Strong process management to ensure the achievement of goals

The process is the guarantee of the results. When there is a problem with the results, it is too late to change all this.

Good managers and good operators must be good process management experts

But there are too many processes to achieve one thing, and there are often many links. Which link is the link we need to dig deeper through perspective? What corresponds to our deep understanding of the business

In this business link, which link is the process management that we need most to achieve this matter. It is usually the most uncertain link for us, but the accurate judgment of uncertainty and risks requires our familiarity with the business to "assist" it.

What is the key link? What is process management?

Suppose we want to discuss cooperation with B-side customers and sign a contract to obtain income. Contract income is the business result we want to get the most.

At this time, some targets will be dismantled, and the contract revenue will be dismantled to different partners. This is the dismantling of the target, not the judgment of the key link, nor is it the process management

Let's get into the complete process of this matter and take a look-

What is the most important prerequisite for us to obtain a partner who can sign a contract?

There are enough potential partners to discuss cooperation

What are the prerequisites for having enough potential partners?

It is possible to have an inventory of partners

How to take stock of potential partners?

Based on the characteristics of cooperation, we can sort out the partners who are most likely to establish cooperation. These partners are our potential partners.

So, if we want to ensure that we get so much contract amount and divide the contract amount to different partners, it will not help us ensure that the business results are achieved

Going forward with the key process - have we taken stock of a large enough market, have we clearly defined the key elements that promote cooperation after the market is reviewed, and have we taken stock of sufficient potential partners based on these factors? This is the key process indicator we need to control

Go back to the Double 12 activity in the figure, how do we manage the process

Judging from the Double 12 activities, the business indicator we need to achieve is - the sales of Double 12 activities

But judging from the 50% discount on the store, what key processes do we have that will affect the sales of the Double 12 event?

The main influence factors of this activity are

Registered Merchant *Effective Merchant Conversion Rate =Effective Merchant

Valid merchants *Unit customer price = event sales

Therefore, the registration merchants, effective merchant conversion rate and customer unit price are the three most critical indicators that affect the final event to achieve

But from the perspective of Double 12 activity operation , do we need to follow up on the process of all three key indicators?

Of course

So what are the processes?

1) During the Double 12 registration period, continue to follow up on whether the number of registrations has been reached

2) After the registration deadline, continue to pay attention to the conversion of email merchants

3) On Double 12, continue to pay attention to the sales of each merchant

But from the perspective of the overall helmsman of the Double 12 event , do we need to pay attention to the process of all three key indicators?

No, we need to pay attention to the most important influencing factor, that is, the link with the highest uncertainty

We review the following important influencing factors of Double 12 activities:

Registered Merchant * Effective Merchant Conversion Rate =Effective Merchant

Valid merchants *Unit customer price = event sales

Is it difficult for my big Alibaba Double 12 to register for the merchant? No

Is the average customer price very difficult to affect my Alibaba Double 12? Yes, but the average customer price is relatively fixed. Of course, there will be a certain increase on Double 12.

Effective merchant conversion rate is a very difficult thing for me, Alibaba Double 12? It is not difficult, but it is particularly important. If my effective merchant conversion rate can be penetrated in the actions that affect the conversion rate, it can actually affect our average customer price if it is overdue.

What are the key elements that can penetrate the effective merchant conversion rate and the average customer price of these merchants?

It is the Relationship in the 4R model

It is the Activation in the RARRA model

And what are the actions that can lead to this key element?

Is there 100% posting posters, is there 100% coverage of store training, and is there a corresponding incentive mechanism for store clerks?

It is necessary to have a responsive incentive mechanism for shop clerks to have a response, because this can enhance the link between the store and Alibaba and enhance the activity of the store.

But that's not enough

If a user enters the store and does not know that the store has a Double 12 event, how can he ensure that when the Wuyang Wuyang crowds into the store, every user can be promptly and accurately conveyed the information about the Double 12 event?

If a user enters the store and WeChat gets angry that day, the same subsidy activity is given, and the user knows that WeChat Alipay pays the same price and chooses WeChat Pay, how should the store’s service staff complete this conversion?

If the user uses a complex set of offers (you also know that the activities are getting more and more complicated now), how can the phone check out the packages purchased by the user?

Then 100% material coverage, covering the best view

100% of the training of clerks, knowing how to deal with it in different scenarios and how to write off various discount packages, will become another key process for getting sales results for Double 12 events.

Of course, it is just a key process goal, because 100% material coverage and 100% clerk training still need to be asked a question by us, how to achieve it?

The key process has been completed by half of this step, because the most critical process has been counted out by us. The next step is to drill down and drill down to find the specific actions that affect the results of this most critical process.

Taking 100% material coverage as an example, our key process is:

1) Material laying training 1,000 people

2) 500 people have passed the above training examination

3) More than 500 people bring 20 pieces of materials to lay 10 stores every day

The same 100% training coverage is also a similar key process:

1) 1,000 people training for activities and speech

2) 500 people have passed the above training examination

3) Each person trains 2 stores every day, and finally one store can cover more than 3 training times

Only when we drill process indicators to this point can we get a truly manageable process

Only by following up whether so many people have passed the training and whether these people who have passed the training exam have carried materials to lay so many stores, can we ensure that we can get 100% of the material laying results?

Only then can we ensure effective merchant conversion results, and only then can we ensure results of sales of Double 12 events

The process is the guarantee of the results. When the process is not good, the result will not be good. When we find that there is a problem with the results, it is too late.

010 Balanced supply and demand relationship at BC end

As mentioned earlier, our growth is usually not just user growth , and often when our users grow to a certain boundary, we need to expand and optimize the content to achieve a new round of growth.

Content growth and user growth are often complementary

I would like to remind you here

1) We often spend a lot of time thinking about how to acquire more users, but how much time we spend thinking about the value provided by our products can better deploy troops through content?

2) We often spend a lot of time thinking about how to expand and expand our products and services, but how much time we spend thinking about how we should optimize our products and services?

A good business system architect should keep these three things in mind

1) Who is the user

2) What is the value of the product and how to pass it on

3) How to continuously optimize the value of the product

Going back to the relationship between our content and user side, how to identify the main problems at present? How can we solve it?

I have an immature method:

Let's first look at whether the current bottleneck is whether the user does not come, or whether the user does not convert after the user comes.

If the user doesn't come, we can find opportunities from the following small points

1) What key users can bring users?

2) What old users can look back?

3) Can the user group further sink

Thinking from these three dimensions is essentially to see what potential there is still room for exploration for existing users.

Whether it is spreading or exploring the consumption potential of old users, it is all about looking at the potential of users.

We can give users some coupons exclusive to this user and ask them to send their exclusive coupons to her friends

We can also give old users some special solemn thank-you activities to awaken the recognition of us that this user has experienced in the past

At the same time, can we do their business if the user has not penetrated in the past?

From the point of view of means, we can unravel the product's layering

For example, in the past, we did a customized tour, and next time we could try to do a 6-person group on the same route, or a 20-person group to cover those people below the top pyramid population.

For example, Xiaomi began to use offline channels to lay down young people in small towns and make the user group sink.

If the user comes and does not convert, we can also check whether there is a breakthrough by a few small dots first.

1), special price

2) Value-added services

3) Competitive products are sold

None of the above is said, this article is long enough, after all, the focus is not here

Of course, there will also be more advanced gameplay, constantly creating scenes that users may use to describe this product and service, and using the most reasonable description method to convince users, is another professional system for event content planning and copywriting.

In the entire operation system, how can we better balance the growth of B and C ends? What is more important behind it is the operation of B and C ends, and how to achieve the coordinated operation of B and C ends, and how to achieve the coordinated operation of B and C ends.

1) There is a need to judge whether the supply and demand ends are balanced (that is, whether the user does not come or the user does not convert when it comes)

2) Our BC side should have similar or consistent judgments on the business rhythm, so that it is easier to move together.

3) Collaboration is also needed at the organizational level

At the organizational level, I also recommend putting the B and C-end operations under the same operation system

What the C-end operation core needs to do is also a bit similar to the traditional marketing

First, shaping value, then transmit value, and then growing users

B-end operation and C-end operation complement each other

Because C-end operations must be based on our products and services to shape value. B-side operations must ensure that the value of products and services required by our users is the delivery standard of the B-side.

011 Indicators Unified Business Objectives

When we understand process management clearly, we must have a balance between B and C.

It's relatively easier for us to use indicators to unify business goals

Let’s take Xiaomi as an example. Assume that as Xiaomi’s current operation, we must do a good job in revenue, disassembly the allocation of our entire indicators, and think clearly about who should memorize the traffic and who should memorize the conversion.

So, behind the disassembly of traffic and conversion rate is to think clearly—

What roles are affected by these data fluctuations, and what factors are affected by the user's average customer price.

The conversion rate must be affected by the supplier , because it determines the quality, and the average customer price is also affected by the supply , because the quality of the supplier determines whether the quality of the basic product he provides me can be sold at such a price.

Who will be affected by the transformation?

Operations convey the value of my product well, which will affect conversion

The platform that provides basic tools is easy to use and will also affect conversions.

Who is affected by traffic?

Will be affected by the person in charge of user growth

Usually, the person in charge of user growth is doing the value transmission, and uses the right way to find the right channel to convey the value of products and services.

At the same time, traffic needs to be flowed between products and platforms. If the user closes this page, he will remind the user to try another product. If the user completes the payment, he will remind the user to follow the service number to understand the service after payment.

Many times, traffic growth is like this

We have laid many pipelines in front of users. Choosing to lay the pipeline in front of which users is determined by the work selected by our channel. What kind of content is recommended in the pipeline to attract users to enter this pipeline and come to our platform. This is determined by the content planning work, and the thickness of this pipeline is determined by the brand.

Go back to our indicator design

If you consider clearly what actions you are related to and who are related to these actions, it is easy to design this indicator .

For example, C-end operations must be based on indicators such as conversion rate, sharing rate, and retention rate. The role responsible for user growth must be based on traffic indicators.

We can also disassemble the traffic of old users and the traffic spread by users, and indicators like this

Of course, operations also undertake traffic and conversion, and they need to follow the indicators of product turnover. From this point of view, is the operation moving forward with a load and taking a short walk to a thousand miles? O(∩_∩)O Haha~

B-end operation and C-end operation have many overlapping indicators. Although everyone has completed it together, B-end operation also needs to directly bear the corresponding indicator pressure.

For example, B-side operations must bear the performance indicators of supplier settlement and the conversion rate of products and services.

In order to complete such indicators, B-end operations need to constantly understand the market situation of the supplier, the supplier's products and services to competitors, and the supplier's production process service standards optimization capabilities

When B-end operations communicate with suppliers and business in order to better "make higher" product and service conversion rates, I believe that more sparks can be created that can enhance the highlights of products and services.

When B-side operations communicate with suppliers and business with the goal of "making high" supplier settlement turnover, I believe it will definitely be easier to gain the trust of suppliers.

Of course, the product technical team also has to follow the conversion rate, sharing rate, and retention rate. These are indexes related to experience and technology related to the indicators.

At the same time, in the business process and in the specific connection with suppliers, they need to be responsible for the supplier's quantity and quality and supply cycle, in order to achieve better product quality and better retention rate.

012 A summary of a super long article

1. AARRR does not need to attract new products first and then convert, but to convert and retain them at the first moment of meeting a user. The current user growth model has changed from AARRR to RARRA.

2. Data should be considered in a comprehensive way. When data is just data sliced ​​in time, it is meaningless. It is necessary to penetrate the data into the business process and analyze it in the breadth of time. Data is a tool that cannot but make decisions on our behalf.

3. A large operation system is an overall business system.

4. B and C are a process of dynamic balanced common growth. Now, growth is product-driven first, then digital marketing-driven, and finally brand-driven. Some excellent offline brands may skip the step of digital marketing and go directly to brand-driven growth

5. A good operational architecture should be based on a good data foundation and business consensus, have strong management capabilities for the process, strong collaboration capabilities on the BC side, and the team has unified indicators and decomposition of indicators to clarify goals and concentrate resources.

Source: Operations Blackboard News