Ten thousand words of dry goods summary: the most comprehensive interpretation of operational data indicators

Operations and data analysis are inseparable. Business insights determine the upper limit of data analysis results, and data skills only approximate it.

Every product and operation should have insight into business indicators. I hope that this article will give newcomers a general framework for analyzing business indicators. The content of the article will give you a "broad" and "general" feeling. I hope it will be helpful to newcomers, while old users can just laugh it off.

User acquisition

User acquisition is the beginning of operations. User acquisition is close to linear thinking, or a fixed process: user contact-user cognition-user interest-user action/download. Each process involves multiple data indicators.

Channel reach

Commonly known as exposure , it refers to how many users browse the product promotion page. It can be found in app stores , on WeChat Moments , on search engines , and wherever there is traffic , there will be channels for exposure.

Exposure is a rather vain number. Think about how much information modern people are exposed to every day? How many promotions are contained in it, and how many of them can attract users in the end? More often than not, channel reach is linked to marketing promotion costs, but the results are far from satisfactory.

Advertising and marketing also consider the brand value brought by the promotion. Although users do not click or interact with the product, they know that it exists and it will subtly influence their future decisions. However, brand value is difficult to quantify. In advertising calculations, the system only attributes user behavior to the most recent advertising exposure.

The number of ad clicks is called CTR , ad clicks/ad views. In addition to advertising, it is also used in the evaluation of various recommendation systems.

Channel conversion rate

Now that the ad has been exposed, users should take action. Conversion rate is the most widely used indicator. The industry combines conversion rate and cost to derive CPM, CPC, CPS, CPD, CPT, etc.

• CPM (Cost Per Mille) refers to the cost per thousand. It is charged based on how many people see the advertisement, and traditional media tend to adopt it. The effect of CPM promotion depends on impressions. Users may browse or ignore it, so it is suitable to use banners to display branding on various portals or high-traffic platforms.
• CPC (Cost Per Click) refers to the cost per user click, which is calculated on a per-click basis. For advertisers, this is much more rational than the CPM model. Many people also think that CPC is unfair. Although users do not click, the exposure brings invisible value to the brand, which is a loss for the supplier of advertising space .
• CPA (Cost Per Action) refers to the cost per action, which is calculated based on user behavior . The behavior can be a download or an order purchase. CPA has higher returns than the previous two, but also much greater risks. It is beneficial to the demand side but disadvantageous to the supply side.

The above three are common promotion methods. CPT is based on time, CPS and CPS are counted within the range of CPA. Channel promotion is an industry that relies on technology. The more accurate the user portrait is and the more matched the content is to the user, the easier it is to generate revenue.

There is also an indicator called eCPM (effective cost per mille), which measures the revenue earned for every thousand impressions. This is an indicator for advertisers to estimate their own profits.

Channel ROI

ROI is a widely applicable indicator, namely return on investment .

Marketing and operational activities all start from the profit of the enterprise and quantify the goals through profit/investment. The calculation of profit involves finance, and often uses the simpler revenue as the numerator. When the ROI of an operating activity is greater than 1, it means that the activity is successful and can make money.

In addition to revenue, ROI can also be extended to other indicators. For some products whose business models are not clear and cannot make money, revenue will be replaced by other quantitative indicators. For example, the number of registered users, which is the customer acquisition cost.

Daily app downloads

The App needs to be downloaded, which is an intermediate state. If you don’t pay attention to this link, you will lose a lot of users. The product introduction and promotional copy of the app store will be affected. Some products that are hundreds of megabytes in size often leave part of the installation to be completed in the form of a patch when the application is first launched, such as various games , for fear that the long download time will cause player loss.

The third-party platform downloads the user registration app. This step is prone to data errors, mainly because the users do not match. Technically matched via unique device ID.

Number of new users per day

The number of new users is the core indicator of user acquisition.

New users can be further divided into natural growth and promotional growth. Natural growth can be users brought by user invitations, user searches, etc., while promotion is the growth of users under the strong control of operators. The former is an optimization of slow cooking, while the latter is a marketing of frying and stir-frying.

User acquisition cost

User acquisition inevitably involves costs, which is what novice operators are most likely to overlook. Acquiring new users: things operators should know

The customer acquisition cost should be directly linked to the finances of new users, such as field marketing expenses and gifts for new users. However, the operating costs of the entire product should not be included. In terms of customer acquisition cost, WeChat fans cost between 10 and 20 yuan, and the prices of products vary greatly depending on the business model. For financial management products, the cost of an effective user exceeds four figures, which is extremely exaggerated. However, the industry's overall customer acquisition cost is still rising.

Number of users in one session

A one-session user refers to a new user who downloads the App, opens the product only once, and the usage time is less than 2 minutes. This type of user is most likely a black marketeer or a robot, and cannot even be considered a freeloader .

This is a gray area in product promotion, where various techniques are used to increase traffic and obtain false clicks to gain profits. This indicator is a risk control indicator and is used for supervision.

User activity

User activity is the core stage of operation, and there are relevant indicators whether it is mobile, web or WeChat. On the other hand, data analysis now pays more and more attention to user behavior, which is a trend of refinement.

Daily Active Users/Monthly Active Users

The industry's default activity standard is that users have used the product. In a broad sense, browsing web content counts as "use" and placing an order on a public account counts as "use", and it is not limited to opening the APP. For this part, please refer to my previous article " User Operation , How to Conduct Data Analysis of Active Users".

The active indicator is the basis of user operation, and the active rate can be further calculated: the proportion of active users in the total number of users in a certain period of time. By time dimension, there are daily active rate DAU, weekly active rate WAU and monthly active rate MAU. The number of active users measures the market size of the product, and the activity rate reflects the health of the product.

But can simply opening a product be a measure of its health? The answer is no. A mature operating system will further subdivide active users into new users, active users, loyal users, inactive users, churned users, returning users, etc. Churned users are those who have been inactive for a long time, loyal users are those who have been active for a long time, and returning users are active users who were once inactive or churned but later opened the product again.

Product users are divided into several groups based on different active states, and different groups constitute the total number of users of the product. For a healthy product, the proportion of lost users should not be too high, and the number of new users should be greater than the number of lost users.

PV and UV

PV is an indicator of the early days of the Internet's web sites, and can also be understood as the activity of the web page version. PV (PageView) is the number of page views. A user's access request to a web page can be regarded as a PV. If a user views ten web pages, the PV is 10.

UV (Unique Visitor) is the number of people who visit a web page within a certain period of time, and its official name is the number of independent visitors. No matter how many web pages a user visits on the same day, he is only counted as one unique visitor. How to confirm whether the user is the same person? Technically, it is determined by web page cache cookies or IP. If both of these change, the user is counted as a brand new visitor.

PV and UV are very old concepts, but data analysis cannot avoid them. In addition to browsing various pages on the product, on third-party platforms such as WeChat, various marketing activities can only be carried out through web pages, so PV and UV need to shine.

One thing we must note is that the WeChat browser will not retain cookies for a long time, and the IP address of the mobile phone is constantly changing. Therefore, the UV statistics based on this will have errors (not a big problem, but the error of new visitors in the UV is relatively large). Here, the openid provided by WeChat can be used to replace cookies as the UV benchmark, which requires additional technical support.

Number of user sessions

A user session, also called a session, is a collection of all user behaviors within a time window. A user opens the app, searches for products, browses products, places an order and pays, and finally exits. The entire process is counted as one session.

There is no rigid standard for the time window of a session. The conventional standard on the web is 30 minutes. No matter what the user does within 30 minutes, it is considered a session. If it exceeds 30 minutes, you might as well go out for a meal and then come back to operate it, or open it again, which will be considered the second session.

The default time window on the mobile terminal is 5 minutes.

The combination of the number of user sessions and the number of active users can determine user stickiness. If the number of daily active users is 100 and the number of daily sessions is 120, it means that most users only visit the product once and the product has no stickiness.

User sessions rely on point-of-use data collection. Without recording user operations, it is impossible to know where user behavior starts and ends. On the other hand, user sessions are the basis of user behavior analysis .

User access time

As the name suggests, user visit duration is the duration of a session. The access time of different product types varies. Social products are definitely longer than tool products, and content platforms are definitely longer than financial management products. If analysts find that most users of content products only visit for a few seconds, then it is best to analyze the reasons.

Function utilization

In addition to focusing on activity, operations and data analysts should also focus on important features of the product. Such as collection, likes, comments, etc. These functions are related to the development of the product and the depth of user usage. No one would like a user who opens the product every day but does nothing else.

Function usage rate also has a very broad range. For example, if a user browses an article, how many users will comment on it and how many users will like it? We can use the like rate and comment rate as two indicators, and then see if there are differences in the like rate and comment rate of different articles, and whether the like rate and comment rate are helpful for content operations . These all belong to function usage rate. For example, for video websites, the core function usage rate is the video playback volume and video playback duration.

WeChat public account indicators can be discussed separately or viewed as an extension of product functions. The image and text delivery rate, conversion sharing rate, secondary conversion sharing rate, follower increment, etc. are consistent with other indicators in this article. It’s just that third-party data is inconvenient and more analysis relies on assumptions.

User Retention

If the number of active users and the active rate represent the market size and health of the product, then user retention represents the product's sustainable development.

Retention rate

Users who use a product for a certain period of time and continue to use it after a period of time are called retained users. Retention rate = number of users still using the app / total number of users at the time.

In today's Internet industry, retention is a more important indicator than new users and active mentions. This is because the mobile demographic dividend is gone, users are becoming increasingly difficult to acquire, and competition is becoming increasingly fierce. How to retain users is more important than acquiring users.

Suppose a product gains 1,000 new users one day, and 350 of them are still active on the next day, then the next-day retention rate is 35%. If 100 of them are still active on the seventh day, then the seven-day retention rate is 10%.

Facebook has a famous 40-20-10 rule, which states that the next-day retention rate for new users is 40%, the seven-day retention rate is 20%, and the thirty-day retention rate is 10%. Products with this performance have relatively good data.

The above cases are all centered around new users. There is also another type of retention rate, which is the active user retention rate , or the old user activity rate, which is the ratio of users who were active at a certain time and remain active later. It uses more weekly retention and monthly retention dimensions.

The new user retention rate and active user rate are different. The new user retention rate is related to the product's novice guidance and various benefits, while the active user retention rate is related to the product atmosphere, operation strategy, marketing methods, etc., and pays more attention to the level of product and operation.

User churn rate

User churn rate and retention rate are exactly the opposite. If the next-day retention rate of new users of a product is 30%, then this means that 70% of users have been lost.

The churn rate can predict the development of a product to a certain extent. If a product has 100,000 users at a certain stage and the monthly churn rate is 20%, it can be simply inferred that the product will lose all its users after 5 months. Although this model is crude and does not take into account user return and new additions, it does reflect that the future life cycle of the product is not optimistic.

A formula can be derived here: life cycle = (1/churn rate) * time dimension of churn rate. It is an empirical formula and may not be effective.

Is it a problem if the product’s churn rate is too high? Not necessarily. It depends on the background of the product. For example, if a product focuses on wedding management tools, its retention rate will definitely be low, as most users will not use it after they get married. But there must be a logic for this type of product to survive. The same is true for travel apps. Users only open them a few times a year, but they can still grow.

Exit Rate

Exit rate is an indicator of the web page. The web page pursues the depth of visit, that is, how many pages the user browses in one session. When the user closes the web page, it can be considered that the user has not been "retained". Exit rate formula: number of page visits that exit from this page/number of page visits that enter this page. If a product page enters PV1000 and the number of visits that directly close the page is 300, then the exit rate is 30%.

The bounce rate is a special form of the exit rate, which is the number of times a user exits after browsing only one page/number of visits. Browsing only one page means that this is the first page the user enters the website, commonly known as the landing page .

The exit rate is used for web page structure optimization and content optimization. Bounce rate is often used in the analysis of promotion and operational activities, and the two are easily confused.

marketing

Marketing also has its own data system, and the Internet's data system was derived from it to develop the AARRR framework. There are two product development models. If a product can gain millions of users in a short period of time, the A AR RR framework is more suitable for it; if a product has a clear business model from the first user, you can also try to apply the concept of marketing.

User Lifecycle

The user life cycle comes from marketing theory and was formerly known as the customer life cycle.

It has two meanings. One is the marketing survival window for individual users/groups. Users will change over time, and this change brings countless marketing opportunities, which are opportunities for the market and businesses. Like a pregnancy lasting ten months, it is a marketing window with a life cycle of ten months. Companies will establish specific marketing around users in this period. Moving, graduating from college, buying a house, etc. all have typical cyclical characteristics.

The other is the life cycle at the user relationship management level, which is more important to operations personnel. The business relationship between a product and its users changes over time. In traditional marketing, customers are divided into potential users, interested users, new customers, old/mature customers, and lost customers. These progressive stages are very similar to user activity.

For a maternal and infant product, I need to know the marketing survival window, that is, how many months of pregnancy, because the marketing focus in the early and late pregnancy is different, and the just-pregnant period is definitely the most appropriate. We also need to know the relationship between the user and the product, such as whether this mother is a new customer or has used the app before but lost interest.

In marketing data analysis, the most critical link is the new customer-lost customer stage. How long a user can interact with the product will determine the vitality of the product. It sounds very similar to retention. The life cycle calculation formula mentioned above is derived from marketing.

User Lifetime Value

Lifetime value is how much benefit a user can provide to a business during their lifetime, and it requires a financial definition. The Internet industry mentions life cycle more often rather than life cycle value because the Internet's business model is not as simple and clear as buying and selling in traditional marketing.

For example, can the lifetime value of a WeChat user be calculated? No, whether it is GuangDianTong , games or WeChat Finance, it is impossible to derive a generalized model. However, for some products, such as finance and e-commerce , the lifetime value is calculable.

Taking Internet finance as an example, an App provides two businesses: financial management and cash loans . The company's revenue from these two businesses is usually at a relatively stable ratio, and the cost expenditure per user is also a fixed constant. Therefore, the profit becomes the amount of user's financial management and loans, as well as the length of the life cycle. Both are estimable.

Lifetime value is more important than life cycle because for a company to survive, it must make more money, not how long users use its products. For more information, see "A Brief Discussion on the Business Logic of Operations: CAC and CLV".

Customer/User Loyalty Index

The loyalty index is a re-quantification of active retention. Activity only refers to whether the product is used or not. User A and user B both open the app every day, but B makes a purchase, so B is more loyal than A. Data often requires more commercial indicators to describe users, and whether or not they consume is a good dimension.

We can express this using a simplified model:

t is a time window, s represents the number of consumptions, and represents the number of consumptions within a certain period of time. If the time window is selected as month, then t=1 is the number of consumptions within the first month from now, and t=2 is the number of consumptions within the second month from now. The data are listed as follows.

Substitute the number of consumption times into s/(s+1) and transform the data. Its purpose is to converge. From the perspective of loyalty, there is not much difference between consumption 10 times and consumption 100 times. Both are very high-loyalty users who are difficult to lose. The relationship between 10/11 and 100/101 is similar, and extreme values ​​are effectively avoided. For users who consume 0, 1, and 2 times, the corresponding values ​​are 0, 0.5, and 0.66, which are also explainable in business terms.

The index obtained by summing up the monthly data can reflect the user's loyalty in consumption. The legend is just an explanation. In actual application, it needs to be normalized and the time weight should be considered: the more recent the consumption, the more loyal the consumer will be. The above model is simple and suitable for early analysis of various business models. For example, financial investment can calculate the number of investments made by users each quarter.

Customer/User Churn Index

The churn index is a re-quantification of churn, which is the opposite of the loyalty index. The churn rate measures all users, but in order to distinguish the subtle differences between different users, a churn index is needed. In the early stages, churn index = 1-loyalty index.

The specific definitions of the churn index and loyalty index can be adjusted according to business needs. For example, loyalty is based on whether there is consumption, and churn is based on whether the user is open and active, as long as the explanation is tenable.

After having enough behavioral data, regression can be used to predict the probability of churn and output a value between [0,1]. At this time, the probability of churn is the churn index.

Customer/User Value Index

The user value index measures the benefits contributed by users from history to the present (the life cycle value is the entire cycle, including the future). It is the premise of refined operations. Different strategies are adopted for users of different values ​​to maximize the effect.

There are two mainstream calculation methods for the user value index. One is the RMF model, which divides users into multiple groups using R (the time of the most recent consumption), M (the total consumption amount), and F (the frequency of consumption). Different groups represent different value indexes.

The second is principal component analysis (PCA), which converts multiple indicators into a few comprehensive indicators (i.e., principal components), each of which can reflect most of the information of the original variable and the information contained is not repeated.

Suppose there is a travel guide website, how do you define high-quality content contributors? How many articles does the user publish? How many likes did the article receive? How many followers does the user have? How many positive reviews does the article have? How often are articles updated? Each indicator is very important. Principal component analysis can include all of the above indicators and process them into two or three indicators (usually linearly related indicators are merged). At this point it is not difficult to process it into a value index.

The above-mentioned indexes are all detailed data for user marketing. How to apply it? The most classic is the matrix method, which divides indicators into multiple quadrants, such as user value index and user churn index.

For users with high user value and high churn index, an active recall strategy should be adopted. For users with low user value and high churn index, appropriate operations can be carried out considering the cost balance... This is an example of refined operation, and it is also an effective method summarized by marketing over the years.

Communications/Activities

Putting communication and activities together, they are two sides of the same coin.

K Factor

A concept widely used abroad: how many users each user invites on average, and how many effective conversion rates the invitations have, that is, how many new users each user can bring. When the K factor is greater than one, each user can bring at least one new user, and the number of users will grow like a snowball, eventually achieving self-propagation. When the K factor is large enough, it becomes viral marketing through word of mouth.

The main channel for domestic invitation dissemination is naturally WeChat Moments . WeChat sharing function and web pages can both increase parameter statistics, which is not difficult to quantify.

Virus transmission cycle

Any form of communication, such as activities, advertising, marketing, etc., will have a communication cycle. Viral marketing may be strong, but unless there is a follow-up, its peak often only lasts for two or three days. This is also the golden period for attracting new customers .

Another dissemination cycle is the invitation mechanism around the product, which refers to the users that seed users can invite after a certain period of time. Because most users will lose the motivation to invite again after inviting others, the propagation cycle can be greatly simplified as follows: Assuming that 1,000 seed users invite 1,500 users in 10 days, the propagation cycle is 10 days, the K factor is 1.5, and these 1,500 users will invite another 2,250 users in the next 10 days.

Theoretically, the K factor and the communication cycle can be used to predict the number of users brought by communication, but this has little practical significance and is more used for the interpretation and analysis of various activities and operational reports.

User sharing rate

Nowadays, all products have built-in sharing functions. For content-based platforms or products that rely on dissemination, sharing rate is a more important indicator. It can be further divided into WeChat friends/groups, WeChat Moments, Weibo and other channels.

One thing worth noting is that the data only shows whether the user forwarded the message or not, but it is impossible to track to whom the message was forwarded. Therefore, when products use material incentives for users to share, be careful about being fleeced. Anyway, I forwarded it to "File Transfer Assistant"...

Activity exposure/views

Communication and online activities are closely related, and there is not much difference between the two. If you want to run a successful event, it is not enough to simply know the number of views of the event. A good event must be the result of data analysis. Take the most common red envelope marketing in the circle of friends as an example. It analyzes the web page parameters as follows:

aaa.com/activity/bigsales/?source=weixin&content=h9j76g&inviter=00001&timestamp=1495286598

The part after the question mark is the web page parameter. source=weixin means the web page is shared to WeChat. content=h9j76g is the specific content of the page, and here is the type of marketing red envelope. inviter=00001 indicates which user shared the message, and timestamp is the specific timestamp of sharing. Different users' sharing pages have different parameters, which are distinguished accordingly.

When these pages are shared by users to their Moments, the data collection system will record all pages opened and browsed. Page parameters are the prerequisite for detailed analysis of activities. Through source=weixin, data analysts learned the number of views of the red envelope activity on WeChat, and the corresponding numbers are also on QQ and Weibo. The content can show which type of red envelopes users like, which type of red envelopes are received most, and how much they cost. The inviter can see how many views each sharer's sharing page can bring on average.

The more parameters there are, the finer the analysis dimensions can be, and the more room there is for activity optimization. If you are interested, you can look at the web page parameters of various activities in the circle of friends (including web pages) and observe the analysis dimensions of other products. It is a good habit to learn from others' experience.

Activity participation rate

The activity participation rate measures the overall situation of the activity and can be applied to analytical indicators of user activity.

How many people participated in this activity (how many active people)? How many old users participated in this activity? How many new users came because of this activity? What are the sharing data of communication activities? How are the various processes in the activity converted? How many new orders the activity brought. In fact, operational activities can be regarded as a product with a short life cycle, and all product indicators can be applied to it.

Good activities should be institutionalized and integrated into the functional mechanism of the product. For example, the red envelopes of Didi Taxi and the red envelopes of Meitu and Ele.me have gradually evolved from activities into a way of playing and a means of attracting people. Various online games in the earlier period were also promoted through activities and have become normalized game features today.

The mechanism of activities means that data must be analyzed to analyze activity indicators, identify advantages for improvement, and then normalize them into reports: how many red envelopes were used today, how many users were added today due to the activity, and so on.

Revenue

Product, operations or marketing personnel are never responsible for activity or retention, but rather for business, the fundamental finances of the enterprise. Data analysis is not intended to increase activity and retention, but rather to act like a giant funnel that ultimately drives the business, that is, to return to the essence of business.

Number of active trading users

From product exposure to user downloads, and from active opening to revenue generation, product indicators are moving closer to commercialization step by step, and active trading users are the core indicators. The whole process is funnel-shaped.

The transaction here refers to both the buyer's consumption and the seller's supply. If the platform includes B-end and C-end, both ends are equally important and need to be included in the data system.

Like active users, active transaction users can also be divided into first-time users (first time consumption), loyal consumer users, churned consumer users, etc. Segmenting transaction data and indicators is related to the progress of product commercialization, so it is necessary. In fact, at this stage, various indicators have become more inclined towards user portraits rather than report statistics.

Active user transaction ratio, which counts the proportion of transaction users among active users. When a product has enough active users but few transacting users, commercialization is problematic, commonly known as difficulty in monetization. Many companies fail at this step.

GMV

The total transaction amount can be counted in GMV as long as the user places an order and generates an order number, regardless of whether the user actually purchases the product. Internet e-commerce prefers this indicator.

The transaction amount corresponds to the actual cash flow, which is the amount of consumption by the user after the purchase. Sales revenue is the transaction amount minus refunds. As for profits and net profit margins, they involve financial costs, and it is difficult to obtain such data through data analysis, so they are rarely used.

If we regard the above three indicators as the dynamic links of user payment, we can generate two new indicators, which is also one of the thinking of data analysis. The ratio of transaction amount to GMV can actually be converted into order payment rate. Sales revenue and transaction amount also involve refund rate. When the analysis is stuck, it may be helpful to observe these two indicators.

Average order value

In traditional industries, the average order value is the average amount a consumer spends each time they visit the site. On the Internet, it is the revenue from each user order, total revenue/number of orders.

Many gaming or live streaming platforms do not focus on average order value, because due to the nature of the industry they are more concerned with the direct value a user brings. In supermarket shopping, user purchases are long-term and the average order value can be used to adjust the supermarket's business strategy. However, in industries such as gaming, the user churn rate is extremely high, and operators are more concerned about the average user payment, which is the ARPU indicator, total revenue/number of users.

ARPU can be further subdivided. When ordinary users account for too large a proportion, the average revenue per paying user (ARPPU), total revenue/number of paying users, is often used.

Repurchase rate

If you say that repurchase rate is the retention rate in the revenue world, you will know how important it is. Just like adding new users, the cost of acquiring a new paying customer is already higher than the cost of maintaining an existing customer.

In many analysis scenarios, first-time users will be singled out as a label, and users who have made purchases more than twice will be considered regular customers. The reason for doing this is that the meaning of going from one to two is far more than just adding one.

The first time a user makes a purchase, it may be to experience the product, it may be to get a discount, or it may be due to strong promotion from the operation. Various factors contribute to the first order. The proportion of their second consumption will drop drastically (corresponding to a drop in the next-day retention rate), because consumption at this time gradually depends on the user's trust in the product, and their liking for the model or the formation of habits.

Many times, the longer it takes for users to make a decision, the higher the average order value, such as in investment and travel. At this time, the first-order repurchase rate is an indicator that needs more attention, as it means more profit.

The repurchase rate is more often used in the overall statistics of repeated purchases: the number of users who consume more than twice in a unit of time accounts for the total number of purchasing users.

The repurchase rate is another indicator, which refers to the ratio of users who transacted in the previous time window and continue to consume in the next time window. For example, if an e-commerce company has 1,000 consumers in April, and 600 of them continue to consume in May, the repurchase rate is 60%. If 300 out of 600 customers have made purchases more than twice, the repurchase rate is 50%.

Return rate

The return rate is a risk indicator. The lower the return rate, the better. It not only directly reflects the financial level, but also affects the user experience and the maintenance of user relationships.

commodity

Here we are discussing data analysis mainly based on commodities. Commodities are not limited to the retail industry. Knowledge markets, virtual services, and value-added services are all types of commodities. It has many common analysis templates, such as shopping cart, purchase and sales.

Shopping basket analysis

Shopping basket analysis should not be limited to e-commerce analysis, but user consumption behavior analysis.

The correlation rate is an indicator of shopping basket analysis, specifically the ratio of the number of sales to the number of transactions. In large-scale shopping malls and shopping centers, joint consumption is the core of business operations, and multiple purchases by users are considered joint consumption. In e-commerce, it is the depth of shopping that is the prerequisite for increasing profits from a single consumption.

Product popularity is a quick and effective analysis. The products can be divided into the top 20 most popular products, the top 20 most profitable products, etc. It relies on the 80/20 rule to find the profit lever. Many marketing will combine it with the linkage rate. For example, e-commerce focuses on promoting multiple popular products that can bring traffic. The hot products do not make money, but rely on the joint sales of other profitable products. This strategy of combining traffic products with profit products is not uncommon.

The most well-known aspect of shopping basket analysis is probably relevance, which can be simply understood as what other things users who have bought a certain type of product are more likely to buy. Beer and diapers is probably the most famous example, and although it is wrong, it reveals that there is indeed a connection between the products.

There are two core indicators in association analysis: confidence and support. The support degree indicates the proportion of a certain product A and a certain product B in the shopping basket at the same time, and the confidence degree indicates how many people who bought product A also bought B, expressed as A→B. Every time Lao Wang goes to the market to buy vegetables, he likes to buy a bunch of onions. In the analysis of Lao Wang’s vegetable basket (shopping basket), onions and other vegetables have high support. But does it mean that Lao Wang will definitely buy other vegetables after buying onions (onions → other vegetables)? No, we can only say that after buying vegetables, Lao Wang will buy onions (other vegetables → onions). In addition, there is also improvement. The most famous is the A priori algorithm.

Association analysis is not only applicable to shopping carts, it is also used in many marketing scenarios for upselling and cross-selling. Common ones include large consumption + cash loans, medical care + insurance, etc. The purpose is to increase revenue.

Purchase, Sales and Inventory

Purchasing, sales and inventory management is a classic management model in the traditional retail industry. It splits a company's merchandise operations into three links: purchasing, warehousing and sales, and establishes a full-link data system. In actual business, many scenarios are closely related to purchasing, sales and inventory.

There are several basic concepts in e-commerce: products, SKU, and SPU. A commodity is a single product as understood by consumers. On any mainstream e-commerce website, a product details page corresponds to one commodity, also known as SPU. On the product details page, there are also options for size, color, and style. These attributes form the SKU, the smallest unit of inventory. Each attribute corresponds to a different SKU. For example, if a piece of clothing has three sizes: SML, then this piece of clothing is an SPU, and the three sizes correspond to three SKUs.

Product management is not as simple as we think. Some users like rose gold iPhones, and some users prefer 128G iPhones. How to better sell these products starts from the procurement stage.

Procurement includes three dimensions: breadth, width and depth. Breadth refers to the product categories. The more abundant the categories, the better they can satisfy consumers' needs. However, it also brings the disadvantages of difficult management and sales. There are 50 categories of mobile phones on the market, and a certain mobile phone store sells 30 of them, with a category ratio of 60%.

Procurement breadth is the SKU ratio, which represents the richness of product selection. The iPhone comes in three colors: black, silver, and rose gold, and in three capacities: 16G, 64G, and 128G, for a total of 9 SKUs. If a mobile phone store only sells rose gold, the SKU ratio is 0.33. Procurement depth is the average number of items per SKU.

Inventory is an intermediate state, purchasing is input and sales is output. Inventory is a dynamic and rolling process of change. We often use the inventory consumption rate within the past time window to measure the consumption of existing inventory. A shopping mall consumes 1,000 pieces of inventory every day in April, and the inventory at the end of April is 50,000 pieces. It takes 50 days to consume these 50,000 pieces. 50 days are called inventory days. Although the formula is an ideal situation, it is okay to use it to judge out-of-stock conditions.

Everyone is more familiar with the sales link, and the indicators focus on two aspects: sales speed and sales quality. Sales velocity is often expressed as sell-out rate, which is expressed as sales quantity within the time window/inventory quantity within the time window. This is a ratio, so the cumulative sell-out rate can be used. The cumulative sold-out rate of a certain product was 50% in March, 60% in April, and 80% in May, which means that the product is gradually sold out and should be restocked. Conversely, if the sell-out rate remains low, promotions should be conducted or purchases should be reduced.

The quality of sales is linked to the discount rate, which is the ratio of the actual amount collected to the standard amount. There are many red envelope discount promotions in China, so a statistician of discount rates is very necessary. A typical application of discount rate is the price elasticity index: the percentage change in product sales when the price changes by 1%. This index will directly affect profits.

There is a lot of content in purchasing, sales and inventory analysis, so people who are familiar with retention and activity analysis may find it a little uncomfortable. However, the mainstream model of Internet monetization is e-commerce or its variants, and knowledge in this area is indispensable. Taking Internet finance as an example, the investment targets have typical purchasing and inventory characteristics. The target's investment amount, risk level and type, target's remaining quantity and estimated inventory days can all be directly applied to purchase, sales and inventory indicators. When analysts discover that the inventory days of a financial product are too long, they need to analyze the reasons, whether it is too many SKUs or weak growth.

at last

At this point, everyone must be dizzy. Business is a complex system, and data analysis is never simple. The combination of the two is full of challenges. My content does not cover everything. For example, e-commerce also has indicators of search effectiveness. When users search in the search box, how many of them are empty searches? And among the non-empty searches, how many valid searches generate clicks? A small search box can also have many hidden meanings.

The more important abilities are insight and discovery. All the indicators in the article were not invented by me. They were summarized by predecessors in marketing and data analysis. However, in my personal learning, I did not swallow them whole. For each indicator, I would stop and think about how to use it. What kind of past experience can he relate to? Data analysis cannot quickly help you gain business experience in the short term, but thinking more is a skill that is easier to master.

Of course, not so many indicators are needed in analysis. Often two or three key indicators are enough. From the business perspective, these indicators are not necessarily part of the job, so don't feel pressured by KPIs. A better way to drive and analyze is to set a big goal for the department, such as revenue, and split the revenue into two or three logically related secondary indicators. For example, more paying users can bring in revenue, a longer life cycle can bring in revenue, and a higher average order value can bring in revenue. The secondary indicators can be divided into three levels by allocating them into multiple small groups or scheduling them according to time.

Mobile application product promotion service: APP promotion service Qinggua Media advertising

The author of this article @秦路 compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting! Site Map