I didn’t realize that I was working with the analyst incorrectly. Only when I used the right method did my business grow.

How can you solve these current confusions if you don’t know how to cooperate with analysts and how to define boundaries, and you don’t quite understand the content of analysts’ work? The author summarizes the job value and growth path of analysts to help you cooperate with analysts. If you also have these confusions, I hope the following article can answer them for you.

Recently, I found that readers are confused about how to cooperate with analysts, especially how to divide the work boundaries between products and operations and analysts. Of course, senior managers do not know the content and value of analysts' work and are troubled about how to recruit analysts. I tried to summarize the value and growth path of analysts, as well as the cooperation methods, hoping to be helpful to readers.

1. Let’s first talk about what is data-driven?

This will affect the definition of analysts and our work, so let’s talk again about what is data-driven.

If you look up various encyclopedias, you will find authoritative explanations. As for my own understanding of data-driven, you can look at its opposite. I think that the opposite of data-driven is definitely not non-data-driven. Humans have been data-driven since ancient times, starting with knotting ropes to count. Most of the time, there has never been no data-driven at all. It's just that generations have always made more use of data and used more data for analysis. The data-driven we communicate today is definitely not the data-driven 20 years ago when the Internet was not popular.

We specifically found two opposites from the company's perspective and talked about what is not data-driven?

The first is that in the past, when companies were operating, they would make operating actions based on facts. When bad results occurred, we saw them and then made modifications and corrections. This iteration cycle could take several months, or half a year to a year, but data-driven development aims to find problems faster and earlier. We find some clues from operations and problems in the business process, and then we make decisions quickly to improve business results and operating income, rather than making adjustments only after the situation actually occurs.

The opposite of this data-driven approach is that long-term cycles, big, salient facts, are no longer the factual basis for our decision making.

There is also what we call guessing, which means everyone wants to discover some clues earlier and then find conclusions within a more diverse range.

Of course, if you want to combine information from different levels to make a comprehensive decision, you often rely on people's comprehensive judgment. This kind of subjective conceptual information is mixed together, and some experience is added to make decisions. Its correlation judgment is actually non-quantitative. This actually has many problems, especially when the process is complicated, and the iteration cycle becomes very fast, when experience is unreliable, it becomes inefficient to rely solely on the business personnel's brainstorming.

That’s why today in our era we place so much emphasis on data-driven.

2. How do data analysts cooperate with business division of labor?

Most data analysis is its organizational form. The division of labor is based on communication efficiency. In other words, our team division of labor is aligned with the business side. Usually, large Internet companies will assign analysts to each department. For example, my business side has two decision-making entities, so my analysts will also be divided into two groups to connect with them respectively.

It may bring some problems, that is, sometimes everyone will do the same skills, and sometimes everyone will answer the same question. In order to ensure the best communication efficiency, this price must be paid. There are advantages and disadvantages in the division of labor in an organization. There is no perfect division method or strategy that can have both. It is more about considering whether your division of labor can support your business and maximize the efficiency of your department. Such a division of labor is conducive to the communication of the needs of each business party in each link.

3. What is the core value of an analyst?

First of all, analytical ability is a basic ability of a person, so it is not just data scientists and business analysts who have analytical ability. Business people also have analytical ability. It is not like whether you can write SQL. Not everyone can write SQL. But no one cannot analyze, especially the business decision makers in any business unit. No decision maker cannot analyze.

So analysis is not a good description of the boundaries of a job function. It is a person's basic ability, some are strong and some are weak. So we say that analysts do analysis, which is certainly not wrong, but it is a little lacking in describing a job, and it is not clear enough.

Today, let's talk about what we do with data analysis. We support decision-makers in various business units. With the presence of analysts, these decision-makers can get a wider range of information. When using different information, the method is more appropriate. When there is a need to connect or coordinate between organizations, the perspective is more diverse. With the support of analysts, the decision-makers' decision-making efficiency will be higher.

Of course, the decision maker may be a group of people, but that’s okay. This is why we need to provide this kind of comprehensive decision support.

4. Data-driven model matching business model

Usually in the traditional consumer field, data including business and financial data are also used, which is the classic 4P theory: channel, price, product, and user. These four elements are organized in this way to form a closed loop. Through this closed loop, the consumer market penetration is improved and revenue is increased. Companies like Apple are relatively new IT companies, but they all build their analysis methods according to this traditional consumer market methodology. The iteration speed is obviously slow. Of course, this has a lot to do with their business characteristics.

5. Growth Hacker Model

This model is the growth hacker model of Silicon Valley. The general organizational form is a good business framework, which divides the traffic and then iterates quickly. The biggest advantage is speed, and each business unit is divided and ruled. It does not require much business knowledge and market cognition, because the environment after the division is closed. You can quickly answer the direction of the next round of iteration in a very small range. In fact, the comprehensive ability requirements for people are slightly reduced, so the talent threshold is low.

However, this model also has disadvantages. The first disadvantage is that it is not so universally applicable. It actually has high requirements for traffic segmentation. For example, Douyin, Facebook and Instagram are more suitable for this method. The key is to see whether the demand on one side will affect the changes on the other side. For example, Didi will have fewer vehicles on the supply side after you place an order. For e-commerce and video, we can roughly see that the supply is infinite and is not affected by the demand side. For complex bilateral businesses, because traffic cannot be divided, many of the so-called AB tests we see are actually different from the business meaning of AB tests of pure traffic-based companies in Silicon Valley.

The second disadvantage is that once you fall into this path dependence, pursue this kind of rapid iteration, and rely on this business framework, you will not be so sensitive to major market changes. For example, when Tik Tok entered the United States, it had already set a sample for Facebook to see. With relatively large traffic and someone having already done a demonstration, the speed of copying its products was relatively slow.

This is actually the case with big market changes, and those who do fast iteration and experimental decision-making are not that sensitive. Such big changes are actually devastating, but growth hackers usually don’t answer such questions. For example, if you want to adjust the business from the operating structure, and want to extend the upstream and downstream businesses, because the closed loop is small enough, from the perspective of business focus and organizational mobilization, local iteration organizations are prone to ignore these problems, or fail to use these dimensions as business handles to do something.

If some new market opportunities appear, we cannot rely solely on experiments. To put it simply, the laboratory is answering the users in your network, that is, the users who are already in your network, what choices they make, and what potential users, users on the opponent's side, that is, the relevant users who may enter this field in the future, think? Experiments don't know, this is the advantage and disadvantage of growth hacking.

6. Behavioral Measurement Model

The so-called new retail, represented by Alibaba and many new and small retail companies, focuses on using behavioral measurement, that is, traffic selection to replace user portraits, which is behavioral measurement. Of course, this method is controversial among product managers and operations. Many books I have read still encourage product managers to create user portraits. Product managers collect user portraits when they interview users or read user comments.

So what is psychometrics? Psychometrics is about your wealth level, your preferences, and some basic attributes, such as whether you are male or female, and your age. Through your profile, we can predict your behavior, the recognition of various products, and the choices you might make. In fact, it is definitely slow in terms of iteration.

However, it is possible to measure key businesses based on a large amount of traffic behavior. This new method of Alibaba. I may not know what you prefer, but I know what you are likely to choose when faced with each traffic choice. In other words, I am not targeting you as a person, but I am only targeting this traffic source, and what kind of choice the person from this traffic source will make when he encounters this traffic stage.

This method is very suitable for studying some niche markets. For example, in the lipstick market, the brands that originally ranked seventh to tenth were actually inefficient in the past. It is possible that three or five brands cover the entire market because the information channel is so small. If you want to do research on consumers of niche brands, its efficiency is very low. But today, with this behavioral measurement and the interaction of offline information and traffic, you will find that small brands in many fields have begun to operate.

This shows that operators of new brands are exploring a new path. This path is very universal and can enter many, many niche fields. Moreover, it is very efficient in the process of combining offline information with traffic selection.

7. Business Measurement Model

This model is particularly universal. Its core is to convert business into indicators and measurable data parameters, manage corporate business and operating strategies through quantitative methods, predict corporate data changes, and measure the impact of each major operational action of operators. This is the most commonly used type of model, which we call a business measurement model.

These three modes are not independent and can be used interchangeably, but they are difficult to use for some businesses. For example, if your business users have relatively little online interactive behavior, then our choice is inherently insufficient when we want to obtain more traffic expression from users.

Secondly, for businesses where the user side and the supply side affect each other, its use will be more restricted, because the strategy you make on the user side will in turn affect your supply side. There is even no way to separate market research and product innovation.

It is difficult for me to answer what the market wants first, such as high certainty and short service response time. But the cost of doing this is very high, which may require sacrificing commissions or actual perception. The whole system is a domino effect, and it is difficult for me to break it down into various fields at the top-level architecture level to achieve their respective goals. All goals may rely on other resources during the experiment.

It is precisely because of this impurity that I was ultimately unable to transform my operating model into a purely experiment-driven model, nor could I rely heavily on consumer feedback and repeatedly iterate to study various market segments.

I don’t even dare to say that we can make an early division of labor, what should be done on the supply side, what should be done on the demand side, how many points should be improved in transaction matching this year, and how much the efficiency of pricing and smart subsidies should be improved this year. After the division of labor is completed at the beginning of the year, we can check this goal at the end of the year. This is not possible because everything is intertwined. These things determine that if we want to achieve this improvement in business efficiency and promote business model changes, there is only one way to choose, which is to provide this kind of comprehensive decision-making support.

This requires our team's capabilities, which actually require a comprehensive view. You must be able to understand the market, understand your competitors, and then you must also be able to understand the key points of the solution process.

8. Analyst’s Responsibilities

Analysts need to provide comprehensive decision support, but many teams certainly don’t have this capability by nature. So there are several major changes. This is also the value that analysts usually have.

The first change is in goal alignment. The entire team must be responsible for the business process and then be responsible for the business. That is, within the organization, you can also focus on business results, rather than someone telling the business side to get data, or the business side saying to look at the dashboard, or the data must be accurate, or someone telling you that you have to connect two data now, because this is not being responsible for the final business, but being responsible for an engineering process.

In this case, you are not actually involved in decision-making, which is the biggest change in the value of analysts. First, you are more of a decision-maker, a decision-maker analyst, and then a data analyst. Data is one of the many ways and tools I use to support decision-making. After talking about the role from accepting needs to assisting decision-making, let's talk about the three major responsibilities of analysts:

1. First, help the business side establish observation

Analysts need to have a comprehensive understanding of their upstream and downstream, competitors, markets and their own evolution process. They must have all the necessary information needed, whether it comes from online or offline, or from users of friendly teams, or information collected from other operations teams. This is the responsibility of establishing observation.

For example, the public opinion of suppliers is lurking in a group of more than 100 suppliers. We listen to what the suppliers are saying and find out what we think is important information. Then we verify whether it is true on the intranet or in the internal system. If it is true, we immediately raise our hands to tell the business side that this matter needs attention. Some suppliers are comparing us with our competitors and asking where we are failing. Then we verify that the internal data may be true. It does not come from field data and does not involve complex engineering processes. However, this is our responsibility because we need to establish observations.

2. The second responsibility is to evaluate how to allocate limited resources

Each business team, who is doing well and who is not doing well, each of them decides whether the previous evolution and iteration direction is right or wrong, whether to turn or stop in this direction, this must be a decision, and the process must be scientific, and these questions must be answered in a relatively convergent environment. Where is this number? This is different from establishing observation. Establishing observation is more about looking outward, looking at the business from multiple levels and angles, and evaluating what is important in allocating resources? It depends on how to divide your limited resources.

3. The third category is for open-ended decision-making problems

To answer this kind of open-ended question, which is our analysis topic, is the responsibility of the analyst. The responsibility of the analyst has become a business responsibility. To a certain extent, it is the intelligence system of my business side, of course, it is not all of his intelligence system. To a certain extent, it is the staff system of my business side, but it is not all of his staff systems. However, because of today's globalization and the modernization brought by the Internet, the weight of our intelligence system and staff system has become much heavier than before.

9. The Misconception That Analysis Is a Responsibility

Analysts need to talk about methods as well as judgment. We often see some debates in the industry in the past, which think that as a data analyst, it is important whether you should do business, whether the opportunity is big, and whether you can judge the market value.

If you make these judgments, some people will accuse you of being unprofessional and saying that you are doing operations. I don't quite agree with this view. Analysts are supposed to assist in business decision-making. Analysis should be combined with decision-making scenarios and specific business judgments. If you don't combine business judgments, there is no way to align with the business side in the decision-making process. This is because as a resource, the resources that support decision-making are also very limited. There are a lot of business directions, a lot of market dynamics, and a lot of possible trends of competitors.

In fact, it can expand into countless stories that can be told. If you pursue telling a good story, an interesting story, you can tell countless stories.

You want to do a disassembly, right? If you want to disassemble and find the main influencing factors, you can find countless anomalies, because there are thousands of disassembly methods, but what should be the focus, should be aligned with the goal, and then be close to the possible grasp. These analyses should be prioritized resources to carry out. Then you need to make judgments.

10. Analysts’ long-term analysis types

The most important thing is to answer whether there are new markets or new spaces to be developed. For example, how to develop new markets or how to develop new market segments is a long-term question that analysts need to answer.

The second question that needs to be answered is the current competitive environment, where I am, whether my current business is healthy, whether we are in the best state, and whether we need to make adjustments. Regarding common questions, I found that the current status of my competition is not very good, but it may be caused by a single point problem, or it may be caused by the entire trading market.

If it is caused by a single point problem, for example, today we found that the conversion rate of new customers on the supply side is not high enough, then we can let the new customer acquisition team solve it. But if the problem is that the balance point between the transaction certainty and pricing of competitors is different from ours, we will suffer in the competition because of the difference in this balance point, but this problem cannot be solved by a single team. At this time, the decision support team must escalate the problem and a higher-level organization must coordinate and solve it.

The third category is to improve operational efficiency, that is, how to create more space for gross profit through industrial progress and the solidification of brand mentality. This is the second point, which requires methods and judgment. I do not agree that analysts cannot think about problems in an operational way. The third category is that all decisions are made in an existing information space, and all experimental iterations and their iterative results are also completed in an existing information space.

11. Analysts should conduct qualitative analysis around business propositions

I just mentioned that business analysts need to establish an observation system for the business. Let's expand on this a little bit. The analyst's responsibilities have changed. Observation has become the guardian of evaluation. We need to establish dynamic analysis capabilities. We also need to separate the observation system from the evaluation system from the methodological level. What does it mean to separate the observation system from the evaluation system?

We often see some data reports that synchronize the progress of weekly OKRs on a weekly basis, or synchronize them by email. In fact, the purpose is to refresh the progress of this OKR every week to let everyone know the current status of the business. Do you have some new changes to answer, but there are problems.

This is equivalent to using the evaluation method to achieve observation. Because when you do evaluation, you can focus on OKR and then focus on the differences between the teams. But when doing observation, in fact, its methods should be multi-level and multi-angle. For example, your upstream influences your business team, their business trends, you need to grasp the changes in the market, the changes in various market segments, and the trends of competitors. This is to establish observation.

We require that all analysis outputs, whether you have written them out clearly or not, must clearly state what business questions you are answering. For example, weekly decision reports, weekly email reports, and daily business reports. The commonality is that no matter how simple or complex the report is, it must be very clear what question you are answering, and it must be answered at a qualitative level.

For example, when reading the daily report, even if there is no problem today, you must state it clearly that there is no problem after investigation, all indicators are stable, and there is no problem today. This is a judgment. With this judgment, all the decisions we make today based on this data, including abnormal observations, are actually completed within an effective range.

Of course, if there is a problem, it is also better to clearly identify where the problem is. Whether it is a business problem or a data change, it must be written clearly.

I don’t really agree that all analyses must be implemented, because this is very difficult. A good analysis can often be implemented, but on the other hand, does it mean that an analysis that cannot be implemented is a bad analysis? This is not necessarily the case. Especially in the case of more complex problems and higher-level decision support, it is normal for an analysis to temporarily not be able to support implementation.

I have a strict requirement that analysts must always know what they are answering. They cannot just list a bunch of data and say they don’t know what question they are answering. It’s just that this is very important and I think everyone should read it. I think it’s inappropriate not to put it up. If you don’t know what question you are answering, then don’t put it up.

If you really think it is important, you must figure out why it is important and what question it is answering, and then organize the data you want to present according to the context of qualitative business problems. It is best to have this kind of biased judgment and write it down directly. Even if I cannot make a final conclusion for the time being, this is also a judgment, and the answer should be based on the qualitative business proposition.

We require to start from the end in mind. My definition of scenarios is not to solve problems with scenarios. I hope that scenarios can answer which levels of people are different, which consumption levels are different, and which scenarios have different demands for certain experiences.

For example, an analyst needs to be able to tell me in this scenario, whether users have higher requirements for cost-effectiveness or for cheapness. If this scenario cannot answer this tendency and cannot answer this consumer stratification, then this scenario is meaningless to me. I have already done this at the earliest stage of listing the scenarios that need to be collected and defined, and made this prediction. The scenario I want must be able to answer these three questions.

12. The Path to Advanced Analysts

I found that it is not enough to rely on some statistical skills, or some processes, or our neutral supervisory status to provide decision support. The core strategy carrier cannot be a skill process or an independent assessment or supervisory status. The core strategy carrier relies on attracting and cultivating talents. We need comprehensive talents who have the ability to understand the whole process of the solution, have a certain degree of judgment, and have an understanding and certain degree of judgment on the goals and market.

Especially in the selection of analysts, business understanding is a more scarce ability. Of course, this does not mean that data acquisition ability is not important, but it is not as important as business understanding. However, the basics of data analysts still need to be able to obtain data.

If we talk about professional division of labor, it seems that analysts are not professional, but in fact, relative to the business side they serve, the analyst's greatest professionalism is the breadth of information they provide. Diversification and breadth are the analysts' greatest professional responsibilities relative to the business side. Because of the breadth provided by analysts, the business side I serve will not, when making decisions, reduce their decision-making level due to missing some information or using some information incorrectly.

Because in order to have such a comprehensive understanding of business, it actually requires very high concepts and abilities. If these requirements are not met during the recruitment stage, the cost of solving them later will be very huge. Therefore, recruitment is very important. Talents who have both the qualities and potential in business and the ability to understand and control the process of solving problems are scarce. This is also what is most needed.

13. Common advancement routes for analysts

1. Junior Analyst

When doing analysis, you can focus on describing the phenomenon instead of reading data to me. You have to tell me that the supply and demand are balanced and the transaction rate is 80%. You can't just tell me that the transaction rate is 80%, because there is a big difference. There is no specific judgment involved. That is, you tell me that there is no abnormality in supply and demand, which actually contains a lot of information. Then you follow up with a sentence that the transaction rate is 80%. In fact, it just gives me an auxiliary scale to tell me where the balance point is.

If you just tell me what the closing rate is, and then I do the analysis, if I am the business side, and you ask me if my current closing rate is 80%, is it good, is it balanced, and is it good in the current competitive environment, as a business side, I will be very distressed. It means that you don't see the business clearly.

2. Intermediate Analyst

The requirement is to be able to comprehensively weigh your goals for the business side, and you must understand the decision-making process and the levers. So when analyzing, there are thousands of choices and thousands of dimensions that can be broken down. You have to choose which analyses can serve the goals and which analyses can have implementation plans. These analyses should be given priority. Anything that is useless, unless you have enough energy or some other inputs, you have to reduce investment and make choices, because understanding the process, understanding the goals, and being able to make choices are the requirements for mid-level analysts.

3. Mid- to Senior-level Analysts

It is required that you can agree with the business side on the business level and have the same depth of understanding of the business as the business side. This kind of business knowledge must be sufficient to have basic concepts about the market, competitors, growth, and finance. You must also have some basic judgment on which issues are important and which are not feasible, so as to avoid being completely out of sync with your own business in terms of focus trends.

4. Senior Analyst

You need to have deep insight that can ensure that key business parties have to call you over when making certain key decisions. I would not feel at ease if I did not discuss this with you because you have such deep insight. You must truly become a member of the first team, rather than just helping me manage 20 analysts.

You can form a business analysis team, but it means that you are a member of my first team and my core decision-making requires your presence. It’s not that there is a statistical problem that I don’t understand and I ask you to help me check it. In this case, the value of being a member of the first team is not high enough and you need to have a certain depth.

As I am using concepts and cognitive abilities to guide everyone, I have to some extent downplayed the management capabilities of systematic thinking and execution. Of course, these two aspects are also very important, especially for many jobs. Having the ability to think systematically is basically the basis for determining a person's rank.

To a certain extent, I have downplayed the focus on these two aspects. Of course, the company's job level requirements will talk about the scope, that is, whether your responsibility boundaries and the business parties you serve are important, and whether you can connect multiple fields. But that may be a result. When you achieve a high enough key level and can actually run a team well, decision-making knowledge will naturally be acquired.

I actually made a trade-off. This kind of alignment in the level of cognition is the most difficult to achieve, and it may also be decisive for doing a good analysis. This is how we guide and train people, and the focus is on improving key levels.

14. Requirements for analysts in different data-driven models

If you are a product manager or a business person in a company and want to work better with data analysts, what should you pay attention to? In fact, the most important thing is what successful practices data-driven has, and then what kind of resources should be allocated under what circumstances. In fact, there are several clearly defined models. We have also talked about the data-driven model above. Let’s look at the analyst portraits corresponding to these models below.

1. Silicon Valley Model Analyst

If your product is very suitable for pursuing a fast growth through experimental iteration, and your business framework is relatively complete and reliable, then the model you need to adopt may be the growth hacker model of Silicon Valley. In this case, the support that the data team can provide you is achieved through more professional methods.

That is to say, a small sample can accurately answer the iteration direction. You don’t need too much data. You can answer it in a very local space and in a very short time. It can answer which solutions are better or need to be corrected. It is actually fast.

At this time, the requirement for the data team is that you have to find some analysts who are experienced in experimental design and process control of point placement, and who have a comprehensive grasp of statistical methods. In addition, he should have good personal integrity and neutrality in terms of neutrality, so it can be done. The analyst you need is this type, and there are actually quite a lot of resources for this in the industry. First of all, if you make this choice, you need someone who is relatively skilled and professional, and who has a good grasp of the inherent processes.

2. Behavioral Metrics Analyst

This model is to do user research and traffic research. Just like I said, when I want to enter a niche field, I now have a lipstick called Palace Wall Red, the Forbidden City lipstick. This lipstick may be a niche brand, but I want to run it well, so I have to keep trying what kind of slogan and what kind of color my users like the most, and then combine it with offline judgment.

This model is the interpretation of traffic selection, which is actually completed to a large extent by industry experts. If it is placed in an Internet company, this kind of function replacement or business adjustment is more likely to be completed by the business side or product managers.

Data analysts monitor the existence and flow of traffic. Analysts can complete technical work, but the interpretation of each traffic change requires people with industry knowledge to complete the interpretation and the next round of iteration. At this time, the cooperation between the two parties must be very close, and the initiative to utilize the traffic actually lies with the industry experts. What analysts need to do may be to have a very good and automated traffic dashboard, and then constantly let industry experts use data for heuristic analysis, and then adjust business strategies to interact with users.

3. The third model analyst

It is what is called business analysis in Meituan, which helps senior leaders align the goals of various business teams and track progress. It sometimes helps predict goals to a certain extent.

At this time, the analyst actually needs to have strong business knowledge, communication, and the so-called economic model, that is, not the unit economic model, but the business economic model. The requirements in these aspects will be very high. It is best to have an MBA or a student who has studied business-related majors. He can help the big boss to reasonably estimate the goals, communicate these goals, and understand the company's business and operating strategies.

At this time, you actually need talents who are more business oriented and relatively more communication oriented. Sometimes they will use algorithms, but using algorithms at this time does not mean that the skills themselves are very high. It requires a neutral party to convince everyone and eliminate disputes in the name of science. This type of person needs this kind of comprehensive decision-making support. In fact, to a certain extent, the decision maker himself cannot escape.

In the third model, there is also a type of analyst, or we cannot call him an analyst, who uses business data to clearly see the data and help the company go public.

Of course, this group of people must first clearly understand the company's operating data and business data, but the most important thing is to find some kind of logic or business correlation and be able to support their own views through current data. The difference between them and business analysts is that business analysts are more based on data or deduce conclusions about business views.

This kind of analyst is more based on the result goals. I want to do such an argument. Therefore, his portrait is more inclined to look for consulting companies, investment companies or industry research companies. He needs to have a deep understanding of valuation logic, and at the same time he can understand the business, have a business background, and interpret the data to establish an argumentation relationship between the current business data and the target valuation.

So first of all, if you want to use data analysts well, you need to divide the model, which business model do you use to run your team, and which one do you want to be efficient? Efficiency is the most important thing.

15. How to cooperate from the perspective of analyst ranking

In fact, analysts can generally divide into several levels, and there are various obvious shortcomings from bottom to top, and the methods of cooperation are also different:

The first type is that you have strong data acquisition ability but weak business analysis ability and have comprehensive skills. As a business party, you can tell the information you want very well, and then you can obtain information from the skill level, which is what technical accuracy can guarantee you.

But its disadvantage is that it may not be able to understand your problem from a business level. You have to explain it very clearly, what I want, and from which perspective, what signs are better compared to whom, and what is bad, you must explain this direction, reference system, this measurement and judgment basis very clearly in order to make a very good cooperation. That is, students who may be more junior or particularly skilled are the way of cooperation they need.

Then, there will be a very broad understanding of this kind. He will have the ability to feel business problems. At the same time, it can also provide some methods and methods in other fields for you to refer to. At this time, you need to communicate more at the goal level, at the trend logic level, and at the solution level. This can have some opportunities for communication and efficiency improvement.

What's better is that the analyst who can participate in the decision-making process is that in the process of your discussion of goals and the discussion of this plan, there may be some potential sources of information that can itself affect the directional decision, so it can be introduced in the early stage.

Therefore, we should look at specific issues in a specific way, and how to cooperate is the most important thing. Generally speaking, we have a clear understanding of which model we are in, but in general, we cannot get direct services from particularly high-level analysts, so we often need to find the right person and explain the reference system and measurement standards particularly clearly in order to have a relatively high cooperation efficiency. If there is a slight difference here, it may become that although each of us is outputting independently, the knowledge effect of decision-making will be poor.

16. Which communication and task analysts think are valuable

The first one is when you keep talking about my conceptual goals, which is what is my qualitative goal?

For example, if I do a business today, I call hotel reservations. What is my qualitative goal? Should I open a path for this segment, or should I see what the next level of market is? The original fast hotel users started to use my products, and I was equivalent to being independent of the original market, and some users who were not in the original segment came in.

Or is it that I hope to serve the original users through business travel hotels, so that the overall average price will fall by 10%, and build a competitiveness compared to my competitors, or is it that I will specifically focus on this peak period of business travel. Because I don’t have enough hotels, I will improve our peak experience by increasing supply.

Of course, it may also be both three, which means talking about business goals at the qualitative level. This is what analysts like to listen to the most, because it is very meaningful input for him to judge how he chooses in the analysis process, how he should help the other party establish upstream observations, and how he helps the other party to evaluate.

At the same time, it means to jointly confirm the boundaries of this information. Within this goal, how much can we talk about how to do this goal? For example, I am now going to make a business hotel reservation, and I chose to improve the operational efficiency during peak periods. So what are the paths?

I need to continuously improve the efficiency of the project from the perspective of engineering, or to cooperate with some hotel supply chains and switch during peak periods. What paths are possible to explore? These paths are exactly what they love to listen to.

Then the second article is about decision-making ideas, which is how to distinguish priorities. In fact, every team has priorities. This routine is different, and the decision-making ideas behind it are different.

When some teams make decisions, they are actually very biased at what is not feasible. No matter how much you discuss with me, it is useless. The thing that is most annoying when we make products is absolutely infeasible. You keep telling him the commercial significance to me, that is, I will tell you how much it is, or I am telling you that there is a specific difficulty in this, which requires resources to overcome. You repeatedly emphasize to me that this project has been decided by the senior management, and it is discussed at different levels.

Each team has its own different ways of priorities. Typically, it is based on the perspective of feasibility, priority from the perspective of feasibility, priority from the perspective of existing resources, priority from the perspective of importance of business goals, and priority from the perspective of coordination and cooperation, which is about decision-making ideas. How our business method priorities is also very meaningful to analysts.

17. What are the needs of analysts who think they have no value and don’t like to listen to them

What analysts don’t like to hear the most is that I don’t care. What I want is accurate data , which means data about these issues. No matter how I get accurate data, it must be accurate.

This is fine in some cases, but it is really hard to deal with it in some cases. Because once the boundary is too large, it is extremely difficult to just get the data correct. It is impossible to have unlimited resources, and it is impossible to converge all risks in an online data source under unknown risks. This is something analysts are more afraid of and don’t like to listen to.

The second thing I don’t like to listen is whether you can look for it again . Why does my new feature have no benefits? Can you observe other directions? We originally planned to evaluate these two result indicators, but we didn’t see it. So can you look at other indicators? Let’s look for it again. Maybe my iteration is valuable. When we do experiments, we will actually encounter this. This is because this observation indicator is not good. I want to find profits within a larger range. I don’t like to listen to it.

When making target predictions, some things have historical laws and can be predicted, while some problems have no historical laws and are difficult to predict. What does analysts like to hear the least? Can you find a method from the perspective of statistics and see if you can estimate an accurate valuation for me?

It's a bit inaccurate. It's really inaccurate. It doesn't mean that you can eliminate interference by optimizing the statistical method and make estimates for things that have no history and no historical samples. This is more difficult to do, that is, once a thing transcends its model, it may provide the ability to make decisions. In fact, you can't solve it by saying that I will conduct in-depth research, spend some time, or find other methods, or it's difficult to solve it.

The third type is that I think the data is not enough. You need to establish more observation dimensions for me , which means talking about data completely separated from the business and qualitative goals. It should be to say that to what extent I think the current data is, I cannot make decisions, or what basis for you to make judgments, rather than to say that I am separated from the business and constantly ask analysts to give more analysis dimensions. This is also something analysts don’t like.

Author: Arun's Growth Research Institute

Source: WeChat official account "Arun's Growth Study Society (ID: arungrowth365)"