“What kind of product operation is it?”

“Is product operations responsible for making products?”

“If product operations is also responsible for product work, then what does the product manager position do?”

“What is the relationship between product operations and product managers? How to distinguish job functions?”‍

Many colleagues who have just started working or are not familiar with operations often have the above questions.

Indeed, in the vertical field, the job responsibilities of content operation, live broadcast operation, channel operation, user operation, event operation... are very clear and intuitive, and can be clearly distinguished just from the job titles.

The responsibilities of product operations vary in different companies, which leads to a vague understanding of product operations.

1. What role does product operation play?

Product operation ≠ product manager.

Product operation is a profession that manages product content and users from three levels: content construction, user maintenance, and event planning. It usually takes on two indicators: company revenue and number of users.

Simply put, product operations are the group of people in a company or department who are primarily responsible for setting KPIs, and are the "moneybags" who determine how much your year-end bonus will be. After a year of hard work, whether you can eat porridge or eat rice in the end depends on how hard the product operation guys give.

Product operations are responsible for the two key indicators of "revenue" and "number of users". If product operations do not understand data analysis, they will not be able to sensitively discover problems from the data and propose solutions, which may seriously lead to the failure to achieve the department's KPI indicators.

2. 3 tips for product operation and data analysis

Not every product operator understands data analysis. Many operators have a liberal arts background and are naturally not sensitive to data, let alone operating complex data formulas.

Faced with data, many product operators are caught in a dilemma: on the one hand, there are blind spots in their own knowledge, and on the other hand, there are the hopes of the entire department. Based on past work experience, the author summarizes three shortcuts for data analysis for those who are just entering the industry or do not understand data analysis.

Taking revenue indicators as an example, how to judge whether the revenue indicators are normal and how to improve them subsequently?

1. Apply the universal formula and check each indicator one by one

The secret to the universal formula is to break down the indicators into the smallest detail and find the fatal problem. The so-called most detailed indicators are those that are infinitely broken down until they are exhausted.

As mentioned above, revenue = number of paying users * average unit price per person. Of course, this formula has not been fully broken down. Because if we break it down further, the number of daily active users can be further broken down into the number of active users in each channel, and the turnover can be further broken down into the turnover of different product categories, etc.

Since the promotion channels and product categories of each product are different, there is little reference value in further splitting them. Product operators can continue to break down the details according to their own business needs.

1) Have you remembered the universal formula?

Next time your boss asks you to analyze data, just apply the formula to export the historical data of four dimensions for comparison by year/month/day (depending on the analysis requirements your boss has given you). When performing comparative analysis, please note that you can only analyze one data variable at a time, and avoid analyzing several variable data at the same time, as this will make it impossible to find the cause of the data.

2) What does it mean to keep a data variable?

As shown below: This is a company’s revenue and user data for the past six months.

There are peaks in the cash flow in February and May. According to the notes, it may be the growth brought by the Spring Festival or the May Day holiday, but what specific actions caused the data growth? Is it accidental or inevitable? Is the growth model replicable? Product operations need to be analyzed one by one.

To analyze the rise and fall of data, two analysis methods are introduced - year-on-year and month-on-month.

• Year-on-year: that is, the comparison of data in the same time period of this year with that of last year, such as Spring Festival with Spring Festival, May Day with May Day.
• Month-on-month: comparison of data for consecutive periods, such as this month compared with last month.

3) Month-on-month analysis

After understanding the concept, let’s review the month-on-month data from January to June 2019.

According to the universal formula:

Turnover = Daily number of users x Daily activity rate x Payment rate x Average price per person

If we list the month-on-month growth rates of various data, we will find that when the turnover increases, some indicators (such as daily active users, etc.) will inevitably increase, and when it decreases, they will also inevitably decrease, forming a positive correlation. When analyzing, you can group the data of positively correlated indicators into one category and ignore them first, and only look at the negatively correlated indicators.

For example, according to the data from March and April, while the turnover increased, the daily active users, daily active rate, and average unit price all increased. Although the increases were not exactly the same, only the payment rate decreased. At this time, we should focus on the payment rate indicator.

The next step is to make an assumption about the payment rate: Assuming the payment rate remains unchanged from last month, what is the turnover? The payment rate increased by 5%, so how much is the turnover? The data gap in between may be where you need to improve.

In the end, you may find that the reason why the payment rate dropped slightly is that the operation did not release new coupons in time in the last two days of April, which led to a decrease in the payment rate of active users and the repurchase rate dropped from the original 46% to 39%, thereby lowering the payment rate for the entire month.

This is the reason for the decline in data, and coupons need to be replenished in a timely manner in subsequent operations.

Year-on-year comparison:

When analyzing year-on-year data, it is best to group the same type of data into a table and plot it into a chart for easy viewing and discovery of patterns. as follows:

In order to meet their KPIs, many companies will have a revenue peak at the end of the year and then the data will fall back in January. The data will surge in February during the Spring Festival, then fall back in March and April until rising again in May.

The data growth patterns in the two years in the chart are similar. The only difference is that the data growth in 2019 is more variable, which should be related to some operation or product action. At this time, we need to go back to the business level to find the reason.

For example, four events were launched during the Spring Festival in 2019, and the turnover of each event was over 200,000, while only one event was launched in 2018, with a turnover of only 180,000.

Data analysis shows that this may be the main reason for the sudden increase in turnover. Therefore, in the future, before the peak traffic period arrives, operators need to be prepared for the battle and plan revenue-increasing activities in advance to avoid missing out on the traffic bonus period.

2. The absolute value is not important, the increase in data is important

As shown in the figure below, this is the Tab page data of a certain APP and a newly launched APP. The main evaluation indicators of the new Tab are turnover and new users.

Looking at the user data of this Tab alone, the number of new users is increasing every month. It seems that the guy in charge of the business Tab has done a good job in attracting new users. The boss should give him a bonus. But those who have data experience don’t think so. Why do they say that? Check out the increase data:

From the half-year data, it can be seen that although the total number of new Tab users is increasing every month, the increase in new users in four months is far lower than the overall new user data of the app (that is, the effect of attracting new users is lower than that of the entire app).

It seems that the operation editor did not attract many new users to the new tab, so the growth in the total number of tab users is only natural growth and has nothing to do with the operation strategy. If the increase in Tab is higher than that in APP, it means that the operator has done a good job in operation and attracted more users.

It can be seen that a high absolute value of data does not mean that the business is operating well, and product operations should pay more attention to the increase in data. When the growth rate exceeds the market, even if the total data does not seem high, you still operate the Tab superbly because you are competing with larger indicators.

3. If you want to retain users well, you should pay attention to CTR data

CTR refers to click-through rate, which is a commonly used term in Internet advertising. CTR = actual number of clicks/number of impressions.

As shown in the figure below, this is the data performance before and after the new algorithm is used. In March, the app launched a new algorithm recommendation, and the client switched 50% of its users to the new algorithm.

Looking at the data marked in red, the total number of users on the homepage has increased significantly since the switch to the new algorithm, with an increase of 300,000 users compared to January, and an increase of 500,000 users in May.

Many people who are just starting to be responsible for product operations easily fall into a misunderstanding - only looking at absolute value data. The increase in the number of users proves that the algorithm is well optimized, otherwise how could there be a growth of 300,000?

But in fact, the effectiveness of the algorithm should not be tested by the absolute number of users, but by CTR. Moreover, we cannot simply look at the total CTR results of the page. We need to split the data into two parts - the CTR performance of the cut-off data and the original algorithm data. Only in this way can the comparison be valuable and we can know which algorithm recommendation we prefer.

Have you learned it? Write down the above 3 tips in your notebook, and you will easily learn basic data analysis and be able to handle a KPI of 100 million in revenue.

Author: Kaka

Source: Kaka's Product Notes