Operation and promotion: refined omni-channel operation plan!

I’ve spent a lot of money on advertising, but why can’t I see any results? Why don’t users buy when they come? What should I do if the push efficiency is very low? Maybe you should try refined operations.

The main contents of this article are as follows:

• The original intention and necessity of enterprises to carry out refined operations;
• Four keys redefine the way of refined operation;
• Interpret the evolution of the development stages of refined operations through scenarios from an operational perspective;
• An important concept in refined operations – closed loop;
• Coordination and implementation of user level system and operation strategy;
• Technology, products and data are the three essential fulcrums for refined operations;
• Data analysis is the most important capability for an enterprise to achieve refined operations.

1. The original intention and necessity of enterprises to carry out refined operations

Why do companies need to carry out refined operations?

If you don’t have an answer right away, think about these three questions first: Why can’t I see any results after spending the budget on advertising? Why don’t users buy when they come? What should I do if the push efficiency is very low?

These are problems that most companies face every day. Essentially, this is because over a period of time, money and users are non-renewable resources, and “limited” is their biggest characteristic.

Therefore, the budget and traffic are limited for enterprises. Therefore, enterprises hope to carry out refined operations. In fact, they hope that every penny spent will have an effect. It would be best if every user who comes stays, every user who stays is active, every active user converts to paid services, and every paying user repurchases. In fact, these are the reasons why companies want to achieve refined operations.

So, what exactly is refined operation? I think there are four more important aspects. Let’s take a look at them together.

2. Four Keys to Redefine Refined Operations

1. Live within your means

The word itself means how much money you spend depends on how much money you earn. Here, it means that how much money you spend on anything depends on how much effect it can bring to you. The evaluations of channel effects, functional iterations, operational effects, etc. that we often mention in our work are all derived from this concept.

2. Take what you need

As user needs become increasingly difficult to meet, companies should leverage their needs while minimizing disturbance to users, so that every user contact can produce good results. This is also the goal that many companies are pursuing with personalized operations.

3. Make the most of everything

Everything must be maximized. For example, in common pit operations, several pits will be set up in places where the product has a large exposure to increase conversions. For example, coupons are very important marketing tools, but they are strongly correlated with costs. Companies hope that all coupons they give out can bring the greatest effect.

4. Wait for the right opportunity

The concept of waiting for an opportunity is relatively abstract. Let me take the securities industry as an example. In most cases, the operations of the securities industry suddenly remembered some lost users who had not been active for 30 days, and they recalled them by pushing news about financial products with very high interest rates on the platform. This is what operations colleagues do on a daily basis. They start pushing news when they think of a certain type of user.

There is another situation where an operations colleague is waiting for the user at a certain node and reaches him when he arrives. For example, if we find that users are on the verge of losing users during the video verification process when opening an account, our operations colleagues will recommend some financial products to users that will give them certain rights and interests as long as they complete the account opening. On the one hand, this helps to increase the account opening conversion rate, and on the other hand, it also achieves the goal of purchasing products after account opening. Both aspects are ways to focus more on triggering intent. So, I call it waiting for the right opportunity.

Another scenario is that when we are doing an activity, some users will be activated by the activity, but others will not. You can think about the reasons why those who are not activated are activated, and whether there is any way to recall them through secondary activation. Just like when we usually do optimization, we give priority to recalling people at the bottom of the funnel, because the lower the level of the funnel, the easier it is to recall them, and the higher the level, the harder it is to recall them. Therefore, waiting for an opportunity is essentially about seizing the willing ones.

The above are the four keys to refined operations that I have summarized. Next, from the perspective of an operator, I will tell you about the scenarios and development stages of refined operations.

3. Interpreting the evolution of the development stages of refined operations through scenarios from an operational perspective

1. Extensive period

During this period we did not realize that many things are limited. When scheduling activities every day, if I couldn’t schedule today, I would schedule tomorrow. If I couldn’t schedule tomorrow, I would schedule the day after tomorrow. There would always be a day when I could schedule my activities. My daily work was to schedule activities, download materials, and upload materials. Even when it comes to delivery and push, I follow suit and deliver whatever others deliver. As for whether users like what I push, I don’t care that much, because at that time, if I can reach you, that’s already pretty good.

2. Awakening

By the second stage, the traffic dividend becomes increasingly scarce, and we will find that it is not working, the effect is not good enough, the budget is limited, and the market is not as good as before. It is no longer in the same state as before. We begin to realize that we need to do it better. How can we do it well?

Do only one thing at a time, think about the purpose of the thing clearly, all means only cooperate with this purpose, and only activate a part of the people at a time. Under such circumstances, I can guarantee my results.

3. Hysteresis Measurement

After companies realize the role of focus, the need to collect data becomes prominent, and part of the company's data enters the measurement stage. At the beginning, there is a lag in data, such as reviewing user reach rate and conversion rate after an activity is completed. If the effect is good, the operation will use these data to infer what to do next time. If the effect is not good, there will be no follow-up after looking at the data.

This stage usually starts with channels. Since channels are directly related to money and have obvious costs, companies will first pay attention to effect evaluation.

However, operation is a hidden cost or opportunity cost, which is not so obvious. Therefore, data awareness begins to emerge at a relatively later stage, and the focus begins to shift from effect evaluation to user operation, product operation, event operation, etc. Therefore, the third stage is lagging measurement.

4. Small-scale testing phase

The delayed data cannot fully satisfy us, because companies hope to measure the effect with the minimum cost and invest the effect in the maximum market, so we need to adopt small-scale tests, such as product and function tests, and not invest money until good results are tested.

The demand for data in this link is actually more clear and strict. We need to obtain timely data. We cannot wait a week before doing an activity and then review and launch it. We hope to get the data within a day or two and iterate quickly. Therefore, the basic demand for data at this stage is T+1.

5. The era of personalization

Enterprises begin to have the ability to obtain data in real time, so they can see user behavior and feedback at any time. Then, with the support of data and technology, based on tags and portraits, we can do a lot of things, such as personalized recommendations and push notifications. So, the era of personalization is the fifth stage.

Looking back on this evolution process, it essentially corresponds to the evolution of data, and has gone through several stages: no data, some lagged data, certain time-sensitive data, and available real-time data . Therefore, refined operations and big data technology are developing in a coordinated manner.

4. An important concept in refined operations - closed loop

Regarding refined operations, I think "closed loop" is a crucial concept. If everything does not have a closed loop, the effect will be greatly reduced. Let's take a look at how to build a closed loop for refined operations.

Step 1: User stratification

Enterprises need to think clearly about how many layers of users are divided into and the corresponding operating strategies for each layer. User stratification essentially represents the user growth system. If companies want users to become loyal users step by step along their growth trajectory, they need to build a more reasonable channel.

Step 2: Event Planning

Companies need to design targeted activity objectives for each layer of users, and then determine the materials, content, time period, and contact methods based on the objectives of the activity. Each strategy is different based on different layers.

Step 3: Internal acceptance

Internal acceptance can be better understood through this example. When evaluating the channel, the operation found that in addition to traffic from Douyin and Toutiao, the traffic also came from Bilibili. The operation wanted to keep these people and decided to increase the investment in Bilibili in the future. However, data analysis revealed that users of Bilibili frequently browse product details, but the purchase rate is very low. This is presumably because its users are relatively young, perhaps children who like the two-dimensional world and have a desire to buy but do not have more financial resources.

Therefore, although Bilibili is a good channel, if it wants to carry out large-scale investment, the product must have a core support, and it cannot directly recommend super-big luxury brands that exceed the user's purchasing power. It can add some light luxury and fast fashion brands to the recommended products.

Step 4: Technology and Products

Realize tag management, user profiling and precise push notifications at the technical, data and product levels.

Step 5: Data Analysis

Collect and understand data in real time, review activity results, adjust strategies in a timely manner, and prepare for secondary reach.

(I will explain the fourth and fifth steps in detail later, so I won’t go into details here.)

Step 6: Strategy Accumulation

Record the degree of match between operational strategies and user groups, such as evaluating whether operational methods match user stratification, and based on the results, enrich the strategy library in experiments. Therefore, stratification is very important. If it is unscientific, the company needs to readjust the stratification and gradually turn it into a scientific user stratification.

V. Coordination and practice of user level system and operation strategy

Many companies emphasize user portraits, which is a typical user stratification. In fact, user portraits have an evolutionary process. In the beginning, user portraits are constructed through some externally acquired data, such as: consumption capacity, life stage, interests and hobbies, social status, education level, etc. This is the earliest user portrait.

Later, it gradually evolved into labels. Now many operators say that they will apply labels to some activities. I asked them what labels are?

Their answers are usually labels such as gender, age, occupation, and region. In fact, in my opinion, these are not truly complete labels, or they are not the most important part of the labels. These are called static labels. Static labels are not absolutely static. They are just labels that users will not change in a short period of time. The corresponding ones are semi-dynamic labels and dynamic labels. Semi-dynamic labels will indeed change.

However, we do not record the process of change. For example, the account balance will change, but we only record its final result: the account balance is 10 yuan, so I don’t know whether the user’s account balance has increased or decreased at this time.

Dynamic tags not only change, but also record the process of change, such as recording what was purchased, how many times it was purchased in the past three months, and how many coupons were used. These are dynamic tags.

Once an enterprise has relatively complete labels, it can combine them to generate more complete and timely user portraits.

In fact, if companies want to really make good use of labels, they need to make a final layer of abstraction. Which indicators are combined in what way to determine whether this ability represents user characteristics such as consumption preferences, consumption behaviors, activity levels, and access preferences. Therefore, building user portraits after label abstraction is a complete process.

Let me explain this with an example:

The above picture is the framework of the label system of the e-commerce or new retail industry, which is divided into several categories, including basic information, member characteristics, consumption characteristics, active characteristics, temporary labels, etc.

• Basic information can be divided into demographic information and sociological information; member characteristics can be divided into member type, member life span, and member information;
• Consumption characteristics are divided into total consumption scale, cross-consumption, consumption capacity and frequency, and activity preferences;
• Active features are divided into online activity and offline activity; temporary tags can be obtained from different directions such as reach, stickiness, and conversion.

Of course, although the above labels are comprehensive, it is never enough for a company to have them. Let me explain the reason with an example. Suppose the operator wants to plan a customer feedback event in Beijing, what kind of people should be selected as the target of the event promotion?

First, we need to select customers who are registered in Beijing and are old customers with high visit frequency and purchasing behavior. In summary, they have visited the website more than 3 times in the past 3 months, have made at least one purchase, are registered in Beijing and have not bought Shiseido products in the past week. They can be considered loyal old customers.

Why do you require that you haven’t purchased Shiseido products in the past week?

Because the purpose of the operation at this time is to reach and give back to users, make users feel that the platform values ​​them and will give them some customized discounts. If users have bought the product in the past month, it means that the demand has been released, and pushing it to such users will not satisfy them. I didn’t write this example very rigorously, but it basically outlines how I target user groups when I run an event, so having labels is far from enough.

Next, let’s look at user stratification. Before doing user stratification, we must first have a user portrait, integrate the user portrait, and conduct a secondary analysis to do the stratification. Only for such stratification can we have a specific strategy. For example: you can stratify according to the user life cycle: new users, mature users, loyal users, and churned users.

We have different operating strategies for different groups. We have some stronger guidance for new users and push some high-frequency, low-threshold products to quickly complete conversions. We need to recall lost users and make some interventions in the middle.

Enterprises can also classify users into high-frequency high-price, high-frequency low-price, low-frequency low-price, and low-frequency high-price according to the frequency of user visits and the value of their contribution to the enterprise.

High-frequency and high-price users are the group of people that companies value the most. Companies will try their best to meet their needs, and there will not be too many of them. High-frequency and low-price users are the largest group of people on the platform, with almost 80% of them. It is difficult to make any targeted operations on them in terms of operational means. Generally, the corresponding optimization operations will fall on the product, which will be optimized through the product's functions and mechanisms. No intervention will be made in the operation, allowing it to grow wildly.

However, operations need to be monitored. If some users begin to migrate to low-frequency and low-price products, there needs to be an early warning to tell us that intervention measures need to be taken. Low-frequency and high-price users are potential groups. They can bring you high value, but they are not your loyal users yet. These people are important targets at the corporate PR level and need to be cross-communicated with them. Regarding low-frequency and low-price users, targeted operational measures may not be taken for the time being if labor costs are limited.

At the same time, enterprises can also be divided into promotion fanatics, promotion sensitive, promotion neutral and promotion indifferent according to their sensitivity to promotions.

Those who are indifferent to promotions will almost ignore promotions and only look at their own needs. If they have needs and can afford them, they will buy them, and if they cannot afford them, they will not buy them. Those who are neutral to promotions will be interested in certain promotions, so promotions can be pushed to them for recall when they are on the verge of losing customers. Those who are sensitive to promotions will be motivated by most promotional information, so promotional activities can be pushed regularly to increase conversions. Promotion fanatics will look for discount information on their own and enjoy the process of finding coupons, so there is no need to push coupons to them.

Of course, there are other classification methods, such as grouping users based on their level of enthusiasm, activity, conversion methods, etc., and then using some targeted operational methods.

6. Technology, products and data are the three essential fulcrums for refined operations

The above picture generally summarizes what refined operations are doing. User groups are selected through tags, activities are customized, and the effectiveness of the activities is evaluated. Finally, the experience related to the activities is summarized, so that the next activity can be formulated based on the basis.

Generally speaking, enterprises need support from three aspects to complete the closed loop of refined operations: technology, products and data. Technology can be provided by the enterprise itself or through a third-party platform such as Sensors Data. I will not elaborate on the technology in detail.

1. Data

There are four things companies need to pay attention to regarding data:

• First, companies must collect user behavior data;
• Second, enterprises need to connect behavioral data with business data, because sometimes after users have performed front-end behaviors, they may not complete business conversions later. If there is only front-end behavior, what can be done is very limited;
• Third, after connecting behavioral data and business data, correct user identification is required;
• Fourth, companies should reasonably construct user tags, user attributes, etc.

2. Product Features

Product functions vary from company to company. Taking Sensors analytics products as an example, user segmentation and attribute analysis are both good supports for refined operations. We have also developed an enhanced version of label management that can manage messy and complex labels in an orderly manner. At the same time, we are further exploring the integration of marketing plans with Sensors products.

In short, with the support of powerful products, operations can devote more energy to exploring new breakthroughs in operations rather than repetitive chores.

Let’s take a look at some pictures below:

7. Data analysis is the most important capability for enterprises to achieve refined operations

Under refined operations, data analysis can be divided into four categories from the perspective of follow-up tracking: evaluation of reach effects, evaluation of internal acceptance effects, evaluation of marketing tools (such as how many people used the coupons and how many people repurchased), and sensitivity testing of marketing tools (such as typical coupon pricing).

1. Evaluation of reach

The above picture simulates the data indicators that I value when evaluating the reach of an activity. In fact, although the activity carriers may be different, such as SMS, APP messages, banners, H5, etc., in essence, what companies track are how many people are reached by the activity, how many people click on the link, how many people browse the page, and then evaluate the click-through rate, conversion rate, etc.

Sometimes operations will consider the ratio of new and old users. For example, in order to attract more new users, H5 is launched externally. If it is found that 80% of the users are old users after H5 reaches them, even if the UV value is very high and there are many people, it does not mean it is good. Therefore, the ratio of new and old users needs to be listed, as shown in the lower right corner of the above picture.

In addition, when I evaluate the conversion effect of an activity, I generally look at indicators such as the completion of the conversion funnel, seven-day retention, and next-month retention. For example, if the purpose of a mobile banking activity is to increase the number of users who complete registration and account opening, what I care about is whether they have completed the funnel after entering the activity page (as shown on the left side of the figure below). At the same time, I will also evaluate user retention to see if the stickiness has increased because of this activity.

2. Evaluation of internal undertaking effect

The evaluation of internal contracting effects has been introduced before, so I will not repeat it here. In a word, to make the user's first purchase process smooth, the product must be designed according to the user's preferences, that is, a good user acceptance work must be done within the product, and the content, category, and price must be consistent with the activity target portrait, so as to minimize the number of users who go through one round.

3. The evaluation of marketing tools can be divided into the following categories according to the different evaluation purposes:

(1) Evaluate the popularity and user groups of each voucher

At this time, our evaluation indicators are generally the collection rate, usage rate, usage frequency distribution, proportion of new users, etc., and the evaluation dimensions are store, type of coupon, coupon denomination, etc.

(2) Evaluating usage preferences of different vouchers

What kind of people are willing to use coupons, new users or old users, high-value users or novice users? Similarly, analysis can be conducted from dimensions such as stores, platforms, coupon types, and coupon denominations.

(3) Efficiency enhancement of assessment vouchers

We will think about whether there will be any changes in the GMV and retention brought to the company if users use coupons or not? Evaluation indicators include average order value, repurchase rate, etc.

4. Sensitivity testing of marketing tools

When testing marketing tools, we are more concerned about whether the pricing of the coupons is reasonable, and hope to attract users with the lowest price. For example, if a 5-yuan coupon is enough, and a 10-yuan coupon can bring a 5% increase in average order value compared to a 5-yuan coupon, I may be more willing to choose a 10-yuan coupon.

Regarding the effectiveness of evaluation coupons, there are different evaluation indicators and evaluation methods based on popularity, new customer acquisition ability and profitability, long-term effect and cost, which can be perceived through the following pictures:

Final words

The above is basically what I want to share with you. Now let me take you back to the first three questions:

First, companies have spent money on investments, but why don’t they see any results?

This issue actually involves many factors. If the company sets up data tracking, it can view data such as the number of views, downloads, registrations, actual conversions, etc. of the delivery channels. After such a data monitoring process, the company can more easily find at which stage the users are lost, because the loss process and reasons for loss corresponding to different channels are different, and the corresponding solutions are also different.

Second, why don’t users buy when they come?

Earlier, we explained the closed loop of refined operations. There are many reasons why users may not purchase. But through data analysis, such as funnel analysis, it can clearly tell you at which step the user is lost, so as to locate the reason for not purchasing. For example, when I see that a user has lost, I generally divide it into a cycle. If he does not complete the action I want him to complete, where will he go and what things attract his attention, so as to determine what can be done to win him back.

Third, what to do if the push conversion rate is too low?

This issue has been discussed a lot before. Through refined operations and testing, we can find out what kind of groups have the greatest response to what kind of activities, so as to push them accurately and increase conversion rates.

The above is all my sharing, I hope it will be helpful to you!

Source: Sensors Data