Community group buying from 0 to 1 practical operation notes

Preface:

In early 2018, by chance, I joined a local community life company in Nanjing, where I was responsible for the overall operations of the company except for the supply chain.

This article is a review of this work, reviewing it from the aspects of users, data, activities, communities, organizations, etc. I hope this article can get more guidance and help from friends.

In addition, this project model is different from the community leader model. It is operated by full-time partners and fixed venues in the community, rather than a part-time leader who casts a wide net. Please be careful to distinguish.

1. New business, new starting point

The parent company of this company is an enterprise engaged in building and smart community software and hardware, and is a top supplier to real estate developers such as Vanke and China South. Except Nanjing, the main business in other regions is smart community. The branch in Nanjing independently carried out this community fresh food e-commerce project.

When I first joined, the business had expanded to 10 communities in Nanjing. We adopt the "T+1 pre-sale + community site" model and have our own service sites and customer service girls inside the community (the service sites only have the functions of secondary sorting of goods and small warehouses, and do not display goods or sell them on site).

The customer service girl is responsible for serving the owners in the community, cooperating with the promotion team to do door-to-door sales and attract new customers, daily maintenance of users in the community, etc.

(Figure 1: Company business model)

The online operation department has 4 dbs + 1 graphic designer + 1 new media + 1 event planner + 1 operation manager, all of whom are relatively inexperienced partners, but fortunately, they are all young people, and their hard work, intelligence and eagerness to learn are their biggest advantages. The business has just started, and they can adjust their strategies at any time, so there is a lot of possibility and space.

What I have to do is to transform the experience and methodology into practical actions, and lead everyone to achieve results. The road is long and difficult, but if you keep going you will reach your destination.

2. Find the river and build a user system from scratch

Since there was no complete operational framework to support and guide us, our products did not have any data collection, which means that the data such as APP visits, conversion rate, retention rate, repurchase rate, etc. were all blank. All operational actions had no process tracking and could only be judged based on the final results (GMV, number of paying users, etc.).

So the seemingly natural way of doing things that I was used to in big companies is obviously no longer feasible.

The game "Man vs. Wild" teaches that when you are lost in the forest, you can be rescued by finding a river and following it. My "river" is "users" and "data."

On the one hand, all strategic paths and methods of online operations are ultimately aimed at driving users to act according to our designed paths, so as to achieve the set goals; on the other hand, with the application of data technology and market competition, we should also shift our mindset from operating products to operating users.

Therefore, "user operation" is our top priority.

Because the app does not have embedded data points and does not integrate data into the background functions, all data needs to be exported to me by the backend. When conditions are insufficient, a simple and direct method is often needed, so I first simply divided the users into 3 categories: the number of new paying users + active users + silent users, and pulled them once a week to formulate corresponding operation strategies for different categories of users.

For example, the definition of a new paying user is a user who places an order for the first time within the past week. The operational goal for this type of user is to encourage repeat purchases. The corresponding operational action is to distribute low-threshold coupons for all categories online. Reminders will be given for offline door-to-door delivery, and members will be added to the group and WeChat.

In the next week, we will pull the list again. If the new paying users last week have made repeat purchases, they will be transferred to the "active user" category and corresponding different operation strategies will be applied. If there is no repeat purchase, the flow will be transferred to "silent users", which will also correspond to the corresponding operation strategy.

At this point, the framework of a preliminary user classification operation system has been established, and the user growth path has completed a simple closed loop. All that remains is to refine the operational actions step by step, conduct in-depth analysis of user behavior data, and attribute and classify them.

(Figure 2: User stratification and guidance strategy. The strategy was adjusted dozens of times based on data feedback.)

Through the changes in various data, we use different means to target different categories of users, and have iterated our outreach methods and operational strategies many times. For example, we have changed the red envelope distribution for new paying users from the initial 50 minus 10 to 20 minus 5, and from category coupons to general coupons, lowering the threshold for new paying users to repurchase.

The time for issuing red envelopes and the time for SMS notifications have also been adjusted several times. Every day, the orders of new users are manually labeled and targeted wording is edited. When the customer service girls deliver goods to new users, they are asked to recommend low-priced and high-frequency products, inform them of the red envelope discounts, and guide new paying users to remain active.

(Figure 3: The backend issues red envelopes with different thresholds for different users)

(Figure 4: Lifecycle Management)

The awakening rate of silent users was very low in the initial stage, mainly because there was a lack of ways to reach users. I used a very "stupid" method: the customer service girl and the sales team went door to door with a list of high-value silent users in the community.

Because these users are our old users and are familiar with our products and service sites within the community, so in theory such door-to-door visits are more accurate. After we export the silent user list, we will manually label the silent users based on their consumption records.

For example: meat users, fruit users, housekeeping users, etc., use different products or scripts to guide them, such as 1 yuan for vegetables, 9.9 yuan for laundry, 10 minus 5 for meat, etc., so that users can place orders again face to face.

Half a month of door-to-door visits eventually awakened nearly 3,000 users. The opening rate and order rate are 50% each, which means that out of 100 owners visited, 50 of them opened the door effectively and 25 placed another order on the spot. Maybe this method seems very low and inefficient.

In fact, this is just an extension of the offline reach capabilities unique to O2O business, and the final calculated ROI, weekly retention and other indicators are also very high (in July 2020, community fresh food companies such as "Rou Federation" began to adopt the door-to-door approach, and we did it in 2018, and we did it more meticulously and efficiently).

The following figure shows the return visit statistics of users of the two communities in one week:

(Figure 5: Silent User Recall)

In the first month after users were classified and graded, the 21-day repurchase rate of new paying users was over 40%, the arppu value (average revenue per paying user) increased from 37 yuan to 55 yuan, the awakening rate of silent users was over 50%, the number of daily paying users increased by 50%, and sales increased by 40% month-on-month.

Although the colleagues are working hard, they are very happy. However, there is still room for improvement in such overloaded work. On the one hand, the awakening of silent users cannot rely on offline visits for a long time due to limited manpower; on the other hand, the subsidies for users are still relatively extensive, and user data is still manually tabulated every week, without more detailed distinction of purchasing behavior.

Although users have been roughly divided into three categories: new paying users, active users, and silent users, the dimensional fluctuations within the range are still very wide. The next step is to further subdivide users, reduce the granularity of operations, and improve efficiency.

3. Keep improving and upgrade user operations

In fact, before the birth of mobile Internet, traditional industries such as chain retail, banking, airlines, etc. had already used RFM or similar models to segment member customers.

In simple terms, the RFM model classifies each user's payment based on three dimensions: "frequency", "recency" and "amount". Depending on the business model and specific goals, the RFM model can divide users into seven or eight categories, such as "high-value users", "key development users", "high-value recall users", etc.

(Figure 6: RFM model)

Back to the project I worked on, there are many benefits to classifying users through RFM, such as awakening silent users. Before, we went door to door to activate them. Although the activation rate was 25%, it also meant that 75% of them failed.

If I can classify and evaluate through RFM, I will first look for high-value recall users with high order frequency and high consumption amount in the past. Such silent users are easier to awaken than low-value users who only placed one order a year ago, and are also expected to contribute more GMV.

For another example, my subsidies for active users can be more targeted. For users who make high purchases, with high amounts, and who purchase frequently recently, I can actually "discriminate" against them. These users have low price sensitivity and high stickiness, so I can reduce subsidies for them.

We can also issue targeted subsidy coupons based on consumer categories. For example, if user A often buys meat, we can issue coupons for other categories to encourage him to experience other products and services, thereby increasing user stickiness.

Faced with messy source data, we tried to first eliminate the high-value housekeeping orders, then eliminate accounts with one mobile phone number and multiple addresses, and finally merge the repeated orders within one day.

Finally, I came up with a more reliable RFM table. Although there are no housekeeping orders and users who place orders on behalf of others are ignored, the classification of most users is more accurate. As for housekeeping orders and users who place orders on behalf of others, there are other ways to solve them, but that is a story for later.

The following figure is a partial screenshot of the RFM user table of a certain community. To protect user privacy, the mobile phone number and related data have been processed:

(Figure 7: RFM user classification table of a community)

Through the RFM model, we divide users into different categories, focusing on "high-value customers", "key development customers" and "key recall users". During online operations, we differentiate between recommended products, distribute coupons and red envelopes in a targeted manner, and differentiate the frequency and means of reach, ultimately achieving a more refined level.

The red envelope subsidy item in operating costs has decreased by nearly 20%, while indicators such as GMV, ARPPU, and the average daily number of paying users have continued to grow. By the way, it also solved the problem of awakening silent users. Our customer service girls and promoters no longer have to go door to door. Instead, they can take the list of "key awakening users" given by the operation, and conduct targeted private WeChat chats, phone calls, and group identification.

Online operations will also distribute coupons or red envelopes with different strategies and reach users through text messages. Without visiting the site, the awakening rate of silent users actually increased by about 10 points.

After completing the RFM model, I re-examined our user funnel and optimized each process. Standardize and institutionalize the actions of each process.

(Figure 8: AARRR model)

After the initial establishment of the user operation system, all user operations have achieved a closed loop.

A further step is to automate user operations, that is, to turn data that requires manual processing into system functions and form an open capability. This is also the concept of the middle platform that has become popular in the past two years. Due to limited space, I will not elaborate on the concept of "middle platform" here.

(Figure 9: original table)

The above is a backend table data, where the fields include mobile phone number + name + time + address + product + product category.

Our previous RFM models, coordinate models, targeted coupon issuance, different operational strategies, etc., all started with sorting out business needs, identifying core indicators and key influencing factors, building a model, and then manually pulling the various values ​​we wanted to fill in the model. Finally, the itinerary strategy assisted in the implementation of operations.

In fact, all of the above are my own home-grown methods of simplifying things when there is insufficient technical support. In companies with relatively good technology, this work is completed through the collaboration of N roles such as product, data, algorithm, development, and testing. It is a systematic, dynamic, and automated process.

For example, at Didi Chuxing, they use algorithms and machine learning to issue coupons to users. The screening combinations of user features can be processed through bucketize or vocabularize, and finally a classification model for issuing coupons is obtained.

In terms of improving transportation capacity, the EM algorithm is used to classify drivers and identify different types of drivers to provide support for the formulation and implementation of operational strategies.

At the same time, when doing user retention, SHAP + XGBoost is used to explore and quantify the factors that affect user activity, and finally form a coordinate graph with the vertical axis being each measurement dimension and the horizontal axis being the feature value.

All information is output into a large table, where operations personnel can find corresponding operation scenarios to improve operation efficiency and influence user behavior. Of course, this is a more technical question, you only need to understand it roughly.

As an excellent operator, all operational actions must be digitized, visualized, online, quantifiable, explainable and predictable.

Operations may not understand technology, but they must have this awareness and be clear about what you want, how product development is achieved, what the process is, what the influencing factors are, and what dimensions and factors operations must determine. This is efficient, high-quality, and advanced operational capability.

4. Eternal Theme: Retention and Growth

Retention and growth are always the core goals of Internet business, especially in business units with fixed physical scenes such as communities, campuses, and cities. Retention and growth are core life indicators.

In a fixed physical scenario, our total number of users, user needs, user behavior, user life cycle and other parameters are relatively fixed. If this type of business is to scale, the first step is to expand the physical scenario, and the second step is to increase the penetration rate and improve the frequency of demand.

For example, Didi Chuxing. When Didi's transportation capacity in Beijing reached a certain level, it maintained a dynamic balance. Then, if Didi wanted to expand its business scale, it had to open up new cities. However, the expansion of the physical business scenarios means an increase in operating costs, which is one aspect of increasing the business scale.

Improving the efficiency of transportation capacity in covered cities and increasing user frequency is the other side of expanding scale, which is the side of reducing costs and increasing efficiency. The theme of this side is retention and growth.

My personal understanding is that retention is included in growth, just like the classic math problem in elementary school: a pool, a faucet releasing water, and a hole leaking water, how long will it take to be filled?

Filling the pool = overall growth, draining it = new user growth, leaking it = user churn. If you want to increase the overall growth, you must divide it into new user growth + old user retention + lost user recall. These three parts complement each other and are closely linked, and none of them can be missing.

We have discussed user recall strategies above. Now let’s talk about retention.

There are many ways to improve retention, which can be summarized into two important parts: one is to start from the user life cycle and improve the retention indicators at each link to improve the overall retention rate. The second is to start from user behavior, optimize every link of the user behavior path, find the core parameters that affect user retention, and thus reduce the overall churn rate.

The previous article has actually roughly described some of the operational methods and strategies we have implemented in these two aspects. In terms of the user life cycle, we continuously adjust the strategy to improve retention at every step of the user's life.

For example: We have optimized the 2A3R model. In the first step of user registration, we guide users to complete their first paid order through exclusive prices and exclusive products for new users, so that they can directly and completely experience the product process and services. This step is to improve the "activation rate".

After new paying users come online, low-threshold coupons are issued to these users to guide them to make repurchases. This step improves the retention of new paying users.

After that, we will incorporate new users into our user classification system, classify them through RFM and other models, and then use different operational strategies to promote repeat purchases. I will not go into details here.

Regarding improving retention, I would like to introduce an important method: Cohort Analysis. The first time I came across this method was around 2010 when I saw an academic article on MSN SPACE. It was used for sociological research, but I was not impressed at the time.

The next time I came into contact with this feature was in early 2015 when Google GA launched it. At that time, I was building my own foreign trade website in my spare time, which was mainly used for traffic source analysis, specifically in terms of keywords, countries, devices and other dimensions.

This method is to observe the changes in historical retention rates, find the key parameters that affect retention, and continuously revise operational strategies through data insights.

The picture below is a retention chart of new paying users on a weekly basis that I made casually. The format is rather casual, but the meaning is the same. The top one is the number of retained users, and the bottom one is the retention rate.

There are several parameters in it. The first column is the time, from the first week to the ninth week; the second and third columns are two variable parameters: new customer acquisition budget and coupon discount rate; the blue part is the retention number and retention rate.

(Figure 10: New user weekly retention group analysis)

After this table is made, its most intuitive function is retention warning. When you find that the retention data of new users this week has dropped significantly, you must immediately trace whether there is any problem with the entire operation process.

For example, in my actual work, I encountered a situation where the weekly retention rate of new users dropped sharply by 17%. After looking at the group analysis tables of various communities, I found that the loss was mainly concentrated in a certain community.

After sorting out the online behavior paths and troubleshooting product bugs for the same product, and ruling out problems with the product itself, it may be a problem with the product or the quality of attracting new customers. After further ruling out the quality of attracting new customers, the only problem left is the product.

Through verification with offline store staff and user follow-up visits, we found that there was a problem with the products used to attract new customers last week. In the summer, we used fresh vegetables to attract new customers, but we did not use perforated packaging. The vegetables went bad in the plastic bags, which created a poor experience for users. As a result, the second-week retention rate of new users in this community dropped by 17%.

We promptly made up for the retention rate in this community by providing free re-distribution of vegetables and maintaining customer relationships. We also improved the packaging of vegetables and reminder language to avoid such problems from occurring again in other communities.

There was also another time when we found that the next-week retention rate of a certain community was close to 100%, so we immediately found the customer service girl at the community store and asked her in detail how she attracted new customers, the strategies and methods used in each link, etc.

In the end, we found out that it was because the customer service girl communicated with the users strictly according to our script, introduced in detail the repurchase red envelopes, coupons, specific products, etc. that we gave to new paying users, and actively urged them to place orders.

Through this case, we improved the coupon reminder strategy, and proactively unified the push and SMS messaging for new paying users’ coupons before they expired, forming a standard. We also shared and trained the girl’s case so that customer service girls in other communities can learn and improve together.

In addition to observing retention rates and discovering superficial problems, how can Cohort Analysis play a more in-depth role through data insights?

We can take a closer look at the two variable parameters "new customer acquisition budget" and "discount rate". We found that when the new customer acquisition budget increases, the number of new paying users increases. When the coupon discount is large, the retention rate increases, and these two are positively correlated parameters.

Now we can draw a rough conclusion: when our operational goal at this stage is to improve retention, we can increase the budget + increase the discount. This seems to be a well-known and correct nonsense, but the core value of this table lies in "quantifiability".

For example: This week’s operational goal is to add 100 new paying users and increase retention rate by 10 points.

Then I can calculate through past data that to achieve this goal, I need to increase the new customer acquisition budget by 1,000 yuan and increase product discounts by 5%. Now that I have the data forecast, the next step is to break down the goals again. In which channels should I invest my additional budget of 1,000 yuan? To which products does the additional 5% merchandise discount apply?

Then I can add a few more variables, trace back the retention rates of various user source channels, and the retention rates of various products, and get the final conclusion: In order to achieve the corresponding goals, I need to increase the new customer acquisition budget by 500 yuan in channel A, 200 yuan in channel B, and 300 yuan in other channels.

The discount for meat and seafood in the fresh food category increases by 5%. Substituting this into the project I am actually operating, I need to make a group analysis table for all communities, observe the retention situation of each community, the impact of variable parameters on retention, and the impact of consumer categories in different communities on retention, and finally conclude that in order to achieve 100 new users and increase retention by 10 points, my total budget is XXXX yuan.

I will increase the budget for attracting new customers in certain communities, increase the budget for field promotion by 700, increase the budget for bringing in new customers by 300, increase the discount for meat by 5% in communities ABC, and increase the discount for vegetables by 5% in communities DEF.

The final implementation step is to break down time nodes, identify milestones, determine personnel division of labor, and prepare resources.

The Cohort Analysis model can add different variable parameters and be refined to different business units.

Learn from one example and apply it to other situations: for example, if I were the headquarters operator of Didi, I would make a general table of all cities and a sub-table for each city, and add several core variable parameters, such as "capacity", "response rate", "order completion rate", "subsidy rate", "penetration rate", etc. Businesses like Didi may also need to do a group analysis of all DP companies or drivers, which I will not enumerate here.

In addition, Cohort Analysis can not only observe retention rate, but also observe different indicators such as "recall rate", "churn rate", "number of orders", "frequency", etc. It can also reduce the granularity and take basic business units as the observation objects (communities, schools, cities, partners, branches, platforms, etc.) to comprehensively analyze and correct the current operation status.

In short, group analysis is the most intuitive means for us to observe and warn of the current operating status. It is also a quantifiable, predictable and executable growth strategy. The granularity and dimension can be freely adjusted to observe the impact of all variables on the overall business.

5. Hasty launch of online activities

With the establishment of the user system, our GMV and number of active users have increased significantly, and our partners’ abilities have also grown in the process. It’s time to hold a full-platform promotion as an acceptance test.

In the last week of May, I pulled all-nighters for two consecutive nights and finally finalized all the marketing activities for June.

(Figure 11: June event framework, the final version has some additions and deletions)

The entire event will last for 21 days from June 4 to June 24, divided into four time periods. It will start with high-priced, low-frequency products and gradually warm up. During the two middle time periods, we mainly sell high-frequency, low-unit-price products such as eggs, milk, fruits, vegetables, etc.

The entire event is centered on the brand, combined with marketing strategies and outreach methods targeting different users, and supplemented by a variety of sub-line gameplay to increase activity.

Once the framework is finalized, the next step is to determine the content and connect with supplier resources. Determine which suppliers will participate, how big of a discount will be given, how many goods will be sold, what the demands are, etc.

We also have to coordinate with the supplier's delivery date. Some deliveries are made once a week, some are made twice a week. In short, this big problem of suppliers was barely solved until recently.

The next step is to submit product requirements. Online gameplay requires technology to achieve. Although our APP has functions such as flash sales and group purchases, there are occasional bugs and problems with the mechanism design. We can only find third-party software to do the raffle and bargaining.

I also gave up on the last 1 yuan purchase because I was worried about BUGs. In fact, looking back at the entire event, the various gameplay methods of the event were achieved only by human labor and third-party technology. If there is sufficient time and resources, the effectiveness of online activities will increase exponentially.

(Figure 12: Deadline schedule for the previous activity. Due to space constraints, only a small part of it is included)

Although there were bumps in every link every day, the event was successfully completed according to the framework settings. During the event, indicators such as total sales, arppu value, new paying users, silent user activation rate, etc. increased greatly, and have initially met our expectations.

The event ended successfully, and I felt relieved. During the review, I found many problems, such as insufficient manpower, time constraints, weak technical support, insufficient execution, etc., which resulted in the deadlines of all the scheduled tasks and links not being completed 100%, affecting the final effect of the event.

In subsequent monthly theme activities, these problems were gradually improved, laying the foundation for the establishment of the platform activity system.

6. Activate existing resources and bring offline activities into the community

Although the rhythm of online activities has basically taken shape and is gradually getting on the right track, offline activities are also a very important part based on the attributes of the community. In the past, we only carried out some daily ground promotions based on service points within the community, which was completely sales-oriented.

This type of ground promotion is inefficient and has poor results. Community residents are often only in the community for two hours, including young people going to and from get off work and elderly people going out to buy groceries in the morning.

Therefore, the daily ground promotion effect is not good, the flow of people is low, the fresh products sold are not easy to preserve outdoors, and the loss is very high. The final result is often that the work is done, but the goods are not sold and the people are not retained.

We did several ground promotions together and sold hundreds of boxes of durian fruits. I felt that this should not be done this way. Ground promotion should be transformed from a sales-oriented one to a new customer acquisition and activation-oriented one, from a kitten fishing to casting a net to catch fish. Daily ground promotion should still be done, but there must be a medium to large-scale event every week.

So we selected a better community for the pilot. The ARPPU value and GMV of this community are relatively high, the residents have close relationships with us, the online and offline activity is also high, and the user base is very good.

Originally, the property management here did not allow large-scale commercial activities, but we took advantage of the Arbor Day and worked with the property management to organize a parent-child tree planting activity for homeowners. We provided the manpower and money, and the property management was responsible for organizing the owners to sign up and participate, planting saplings in the green spaces in the community. We also made albums of each family's photos and gave them to the owners.

In this way, the property management did not have to spend much manpower, but also improved the relationship with the owners, and the community also had more green space, so the property management agreed readily.

In this event, we arranged two scenes, one was planting saplings in a green space, and the other was a food market in the community square. Users must first sign up for the event on our app, and then sign in at the market to receive saplings and other items. Our staff will then take them to green spaces in batches for planting.

During the registration and waiting period, we conducted cooking tastings of featured products on site, distributed coupons on site, and required users to register on the app to receive them, with all users online. We also had an on-site communication session, setting up links for sharing red envelopes, photo frames and so on.

In the end, this event attracted 170 new users, the community received 300 new online orders that day, and GMV reached a new high. We formed a good relationship with the property management company. From then on, when we held various activities in the community, the property management company not only did not charge us any money, but also helped us with door-to-door promotion.

After this event, we quickly replicated it and organized many similar events in other communities, including local specialty fairs, home appliance cleaning events, Labor Day fun labor competitions, Mid-Autumn Festival rice dumpling making competitions, and so on. These activities helped us better connect with community users, formed waves of word-of-mouth communication, allowed us to take root in the community, and gained a large number of loyal users.

After community activities became a regular occurrence, we planned and organized a series of surrounding tours to bring owners out of the community and into our cooperating farms and reservoirs for on-site experience and sales.

We organize owners to pick vegetables, grapes, and watermelons, fish in reservoirs, visit modern chicken farms and pig farms, etc., which further deepens users' recognition and confidence in our products and gives users a better experience.

7. The significance of community: from transformation to connection

We have established WeChat groups in every community, with each group having at least a hundred and as many as three or four hundred people. We also have new media editors who edit the wording and product images every day and post them in the groups. However, the community is not very active, there is no good content, and there is no complete operational logic and framework.

I made two adjustments to our community operations. The first one was conversion-oriented. We formed conversion goals by creating active groups through content. In terms of content mechanism, we did not have any plans before. When purchasing had new products, we would inform operations temporarily, and we would publish them when the copy and design were completed.

In addition, the operator cooks a dish himself every week and sends the pictures, texts and product links to the group. All content is limited to this.

The first step I took was to establish a content production mechanism, requiring purchasing departments to submit unified reports on the main new products in each category on Tuesday morning. Then, based on the promotion intensity, profit, volume, weather and other dimensions, I would uniformly arrange the time for pushing the products within the group. I would also put product promotions, event publicity, cooking pictures and texts, and other matters into a table, and also include the source of materials, production standards, copywriting posters, topic creation, customer service scripts, etc.

This has initially standardized the production and release of content. The community has a set of basic operating rules.

(Figure 13: Community Operation SOP Table)

After the first adjustment of the community operation strategy, the group activity and user conversion have improved significantly, but still failed to achieve the expected results.

After a period of data statistics and review, we tried a second adjustment, changing the nature of the group from "sales" to "connection". Sales are easy, but connection is difficult. All community group buying groups, including ours, have a problem: even if there are hundreds of people in a large group, there is no high-frequency and effective interaction.

Although these users live in the same community, they are different in gender, age, job, hobbies, living conditions, etc. Apart from public issues and product and service issues within the community, there is a lack of topics in other dimensions to connect them. So we decided to establish different small groups while retaining the large group.

A good community must have three characteristics: small number of members, common interests, and the ability to guide.

So we subdivided the group and asked our younger sister to add friends and create small groups. We set up small groups based on different dimensions such as online and offline activities, popular products, user classification labels, etc. The number of people in these groups does not exceed 50, and the group topics are dominated by group members. Our operators only provide guidance, set rules, provide services, etc.

After two community adjustments, although our overall sales and activity conversion rates within the group have increased significantly, this is the difference between operating products and operating users.

BTW: There is a great example of community operation here: Kids King. If you have children at home, you might as well go to Kids King to apply for a membership card and see how their users and communities operate.

8. Upgrade organizations and empower individuals

After the online operation system was initially built and each operation module was operating according to the SOP, I kept thinking about two questions:

1. How to improve the ability and willingness of customer service girls to better reach and serve users in the community?
2. How to reduce the operating costs of each community so that the single-station profit model can quickly reach a better position?

Our customer service girls are generally married women aged 35+ who live in this community or nearby. Their previous jobs may have been basic sales such as supermarket promoters.

Their advantages are that they understand the products, have strong affinity, good service, and have a better understanding of users in the community. The disadvantage is the lack of initiative in using the tools and the ability to deeply manage users. At the same time, each service station must have two customer service girls, and the monthly labor cost is about 7,000 yuan, which is a huge obstacle to the profit model.

After clarifying the core issue, I made a decision: to convert the customer service girl who was originally employed by the company into a full-time partner in a cooperative relationship, cut her base salary, and increase the profit sharing ratio.

The company pays for site costs, fixed equipment investment, goods, tools, subsidies, operational guidance, etc. Full-time partners only need to maintain community users and make sales, and abide by basic cooperation rules, such as not being a third-party platform, only selling products provided by the company, accepting the company's coaching, training and assessment, etc.

The core logic is that the company invests money in individual entrepreneurship, hard work leads to wealth, and the fittest survive. After three meetings, 80% of the customer service girls were willing to become partners with no basic salary.

We have also set up a training system to re-train new partners. The core of this system is to help them refresh their thinking and improve their capabilities. There is only one factor in their assessment: growth.

We also conducted a detailed review of the existing users in the community, made portraits and labels for each user and gave them to partners. We also asked them to enrich the portraits during the operation process. The dimensions include the number of people in the user's family, the age of the children, the number of rooms, air conditioners, and televisions, the newness of the house decoration, and other labels.

We guide partners to completely transform from "managing products" to "managing users". After receiving the portraits of partners' feedback, we will develop user needs in a targeted manner.

For example, for users with children, we will specifically recommend maternal and child care and K12 education products from third-party suppliers. For users with old house decoration, we will promote our renovation business point-to-point through SMS, DM inserts with goods, WeChat and other channels. For users with floor heating at home, we will push floor heating cleaning and maintenance activities in late autumn, and so on...

Meet user needs from all dimensions and help partners increase their profits.

After the implementation of the partnership system, the partners without basic salary guarantees not only did not see a decline in performance, but their income also increased significantly compared to when they were in an employment relationship. Of course, there were a few who could not adapt to the partnership system and were eventually eliminated and replaced; but from the perspective of the company's operating costs, business improvement, user satisfaction, and partners' personal income, this was a successful reform.

9. Final Thoughts

This article simply reviews the operation of the project in more than 10,000 words, but compared with the overall workload and work content, it is just a drop in the bucket. I hope these superficial experiences can give readers some small help.

Finally, this article was formed in early 2020. At that time, community fresh food was not paid attention to by giants. If you want to compare it with giants, the main difference between the business I run and theirs is that Didi Meituan Pinduoduo is operating traffic, and professional group buying companies such as Shihuituan are operating products, and my core concept from beginning to end is operating users.

In the next article, I will talk about my understanding and judgment of the community fresh food track in detail. Welcome to continue to pay attention. Recently, I plan to pull the team to continue fighting on this track. Investors or practitioners who are interested in community business are also welcome to contact us.

Author: oy1010

Source: skyline0510