KPI for Product Manager

I know, perhaps the ability of Internet product managers to judge the standard should have been many people have put forward, such as "Quality Model", "I am the ego" and so on, these are very reasonable. Excellent, even the great product managers should contain a lot of quantifiable ability to assess, so as to ensure that their products have not copied soul. But what I want to analyze today is: What should a qualified Internet Product manager look like? Neither the good nor the great is involved.

Product manager's passing line

If the author's point of view, then the Internet Product manager of the passing line only one, then is completed kpi--and this is the author's boss has been hanging on the tip of the tongue.

In fact, to complete the KPI said it is very simple, light to do is absolutely every internet person in the heart that "will breathe pain." Of course, if your boss gives you a set of KPIs that don't hurt you, the Internet company is basically the future.

In my opinion, the product manager who is able to sustain (note: Be persistent, who can perform a KPI once in a while) should have at least 3 abilities to complete the KPI:

1, can understand the means of KPI, and decomposition KPI for a number of key points;

2, can reasonably set KPI;

3, in their own responsible products, combined with user needs, through the improvement of products or other means to reach key points.

In my past work, I have been responsible for the SNS community, games, open platform, game data analysis system, marriage dating and other items of products and operations, this article will be combined with these products, from the point of view of data analysis to talk about the above mentioned capabilities.

Second, understand how the KPI to achieve, and reasonable decomposition

As a product manager, it is estimated that most people have accepted the KPI specified by the boss. This goal is generally a simple measurable figure. For example: "Daily active up to 500,000", "daily income of 300,000 yuan." This simple figure is the boss's expectations, usually based on a number of companies in a specific situation to push out a figure (such as coping with corporate finance pressure). If you can't undo this backwards-pushed KPI number, it's really hard to tell if this is a reasonable goal and therefore not a guide to our work.

So here, let's look at how to simply break a KPI into multiple operable modules. First, we define some keywords. Here is a simple disassembly of a product that requires an account system:

New user cost: Refers to the cost of obtaining a registered user;

Daily new users: no need to explain it;

New user survival: The ratio of new registered users to a set condition (there will always be people who will not use our products after registration), so we have an established condition to confirm that the user has been activated. Game can be judged by a level, other products can be registered for a period of time to calculate the degree of activity;

Effective user average life cycle: Generally speaking, users will not always use a product, once more than a threshold is no longer logged in, basically the user is "dead" (the rate of resurrection is generally very low), the average life cycle of different products vary greatly, such as SNS games may only be 1-2 months, and SNS for several years;

User logon frequency: The product that uses once every day with every 10 days, the same number of effective users, bringing the daily life will be 10 times times different, so we need to understand the effective user in the life cycle, on average how often use our products. For the product manager who is active in the KPI, a user log on a number of times per day is meaningless to KPI;

Access Depth: Users each use the product length of time or access to the number of pages, the focus on the flow of products, this is a very important indicator.

In combination with the above 6 sets of definitions, let's look at an example (please don't tangle the numbers in the example to be close to the facts, they're just for illustration). Suppose the boss gives the goal is: "In the Q2 Day login user reached 600,000" (excluding the day registered users), the boss vowed: "In my opinion, through natural growth can achieve this goal." Cunning you think this statement seems not very reliable, so through bargaining to the boss to "4 million market costs", and currently we have "80,000 days active." So intuitively, our goal is to spend 4 million in 90 days, and increase the daily activity by 520,000. So is this goal difficult or difficult? Can it be achieved through the natural growth that the boss says?

This we need to understand the above defined numbers, assuming: no promotion in the case of this product daily new registered users 40,000 people, and the promotion of the single user cost of 10 yuan, the new user survival rate is 30% (in the registration of a week has 2 logon behavior to determine to survive), active frequency of 0.2 (on average every 5 days), The average user life cycle is 90 days.

First we analyze the peak we can reach and the number of days we can achieve in 90 days, if we depend on natural growth alone. (In this article, for ease of calculation, I will use some approximate calculations, such as the user's average life cycle of 90 days, the loss curve should be a hyperbola, but for the sake of simple calculation, we think this is a straight line, that is, the user loss speed is equal, that is, on the 180th day of registration, all registered users died.

Daily active peak = daily NEW * survival rate * Active frequency * (Cityinn average life cycle)/2 = 40000*0.3*0.2* (2*90)/2 = 216000

The formula is very well understood: 40000 new users a day, but the effective new is 12000 people (will continue to use our products), plus these effective new users every 5 days will log in once, is equivalent to say 40000 new users daily contribution to daily life only 2400 people, And the daily new effective users will lose 1/180 a day, that is, 180 days later, the daily increase equals daily loss.

As for why divide by 2, is to consider the new user in a period of time also has a certain loss, for example, in the case of a product with a life cycle of 90 days, n users registered on the No. 0 day, the number of people who lost on the 90th Day was N (90/180); The number of users who were registered on the 1th day is n ( 89/180). The number of users registered on the 89th day, in the 90th day of the loss of N (1/180), the loss of the number of users for N (1/180+90/180) *90/2=N*1/4, that is, new users will have 1/4 lost, and so on, in the 180-day stability, in fact, there are 1/ 2 new churn (not yet clear to study for yourself).

So we can see that if we rely on natural growth, basically 210,000 days of life is over, the boss gave you a more difficult task, and this 210,000 or 180 days to reach the number.

So let's analyze the daily activity that can be achieved by natural growth after 90 days:

First, we analyze the loss of new users, the previous paragraph has been calculated, on the 90th day, the new user will be 1/4 lost, while the old user (80,000 days live) will be left around 36000 (why 36000, see the next paragraph 2), then 90 days after the active is 40000*0.3* 0.2*90*3/4+36000=198000, that is, natural growth, when Q2 ends, you can only have 200,000 active users.

Here to say a few of the above number of algorithms, as for why do not speak, we can think for themselves.

1, to give a day life is less than the peak day, the calculation to achieve its time: daily new * survival rate * Active frequency (1+ (Cityinn average life cycle-X)/2* average life cycle) *x= known day live *2, the solution x, is the time required;

2, calculate a day to live in a certain period of time (x days) Attenuation: the exact formula of their own research, if more lazy directly according to 1.2* (x/(Cityinn average life cycle) to calculate attenuation, the error is not too large.

In short, do not understand how the people do not care too much, remember daily active peak = daily NEW * survival rate * Active frequency * (Cityinn average life cycle)/2 on the line, if it is estimated, not so detailed.

Now that we know that natural growth is not a goal, we need to improve our products to improve some key figures and make good use of our promotional funds. Now let's go back and push, how much should we improve on these key numbers in order to achieve a 600,000-day active goal? Our product survival rate has changed to 0.4, the landing frequency has become 0.4, the user's average life cycle has become 120 days, we look at the new 40,000 days, our daily peak and the number of Q end:

Daily Live Peak =40000*0.4*0.4*240/2=768000

Active =40000*0.4*0.4*90*39/48+40000=508000 at the end of Q

Some calculations are omitted.

The conclusion is, if by improving the product to greatly improve survival, use frequency and life cycle, give you a little more time, you will be able to achieve the target of 600,000, but at the end of this q, you can only get 510,000 days to live. But don't worry, we still have 4 million promotion funds, we can take out to increase the amount of registration.

Of course, you don't have to spend 4 million a day to become 400,000 users. Let's assume that your channel can bring you up to 40,000 new (very strong channels) each day, some words must be introduced to me), your 4 million enough to deal with 10 days, obviously, you should be in the last two weeks of Q2, use these channels (the early words, at the end of the quarter when the user is lost). Let us analyze, these channels enough to give you how many days active (I do not calculate the 10 days of users lost, so the previous channel to increase the number of daily limit assumptions are meaningless):

400000*0.4*0.4=64000

OK, plus the channel to bring the daily live users, you can finally reach the end of the Q 570,000 active day. First of all, although this number has not yet reached the 600,000-day target, but the boss can basically forgive you, secondly, you can also consider the optimization of each ad registered users to obtain the cost.

The consideration of the income type product, also need to increase ARPU index;

Traffic-type product considerations, need to increase access depth indicators.

About 3, 5, 7th retention rate and some common indicators, in fact, is to change the angle of view (survival, survival time), can be used in accordance with the general use of PM.

Third, how to set a reasonable KPI

After the breakdown of the KPI, we analyze how to set up a reasonable KPI.

There may be a PM that says, I don't need to know how the KPI is set, because it's the boss's business. I do not agree with this view, because: 1, if the boss set the KPI is unreasonable, you should be able to help the boss analysis, why unreasonable, is the resource constrained, is deviated from the target, or other? In short, in order to own the company, should help the boss to develop a limited resources within the maximum to meet the core demands of KPI indicators 2, do not want to take the team PM is not good pm, can assign reasonable KPI is when the boss of a ability; 3, set KPI in the process, can help pm carding ideas, better understanding of products and user demands, and find the appropriate implementation path.

I accepted the first KPI is "30 days after the day active 20,000 people", and at that time, there is no promotional costs, 1.0 version interface difference, performance instability, product target undecided, daily live 50-200 people. Now look back, this KPI is purely a pat on the head, or fooled me this just graduated students, so it must not be completed, but then the boss did not blame.

In order to avoid our KPI to become an unreal number from the boss to the employee, we should set the medium-term product target according to the resources that can be controlled, and it is better to have a reference competitor. For example, I mentioned in the above survival rate of 0.4, landing frequency 0.4, this is not the head to decide, but in the same type of products, we choose a better value as the target.

If you do SNS, try to understand the Facebook, Qzone, everyone's users survive and use frequency, assuming that their survival rate is 0.85, 0.76, 0.62, and your current product survival rate is only 0.3, that means that your product has a lot of room for improvement, at least to do 0.7 should be no problem. Similarly, if you do is the classification of life, then study net, 58 of the user situation, you can not allow a user to rent a house every day, but at least should be able to reach the more advanced level of similar products.

When you have an intimate understanding of these data, there is a cornerstone that can be relied upon to discuss KPIs with superiors or subordinates. Also take the above goal as an example, assuming that 0.4, 0.4, 120 is already the industry's top level, you should ask the boss to properly reduce the expected or supplementary budget. Of course, I do not think that PM should be the boss's expectations are very low, as a professional ethics, take the boss's salary, it should be full of achievements, the challenge of unreasonable goals and swept, to show the core value of our excellent PM.

Iv. How to combine product improvement key indicators

Finally, we talk a little bit about how to turn our product 0.2 active frequency to 0.4. Do pm, will certainly in the career contact with a variety of types of products, then how to go beyond the specific product form, master the different product design methods? I think that although the product form is different, but as long as the good logical analysis ability and in-depth research user behavior, we always can find the key point of improvement.

For example, when I take over a new product project, first of all, I will combine the target research product positioning, product positioning is its key to success or not, such as whether we want to make friends or marriage, first of all determines the height we can finally reach. Marriage is no more noble than friendship, but if we aim for long-term and steady flow, it seems that love and marriage can help us achieve our goals better.

Secondly, I personally in the product, the more emphasis on the "user's use path." In the core path, grasp the most easy to improve the point cut, one to achieve the desired effect. For example, "fair marriage", by improving the landing page, improve the registration conversion rate (for example, the use of beautiful handsome or real photos; register step or step; registration option rich or simple), through the referral and sorting algorithm, optimize user activity (can push the right person to the right person, is to stimulate the user to find and continue to use a key point) and other practices, a large increase in the registration conversion rate (that is, I just said the optimization of the registration cost of a key point), the user login frequency.

Some of the calculations in this article are based on what I thought last year when I was designing a third party data analysis system. Look forward to discussing with you!

The author of this article, Li, is a social product director at the Net Mobile division. Zeng Ren Papaya, UC excellent as the company product manager. have done cell phone community, mobile games, marriage websites, coupons and other product items, has a very rich product and project management experience, wireless Internet product development has unique insights.

Sina Weibo: Http://weibo.com/liyan777,Gmail Exchange: evan.liyan@gmail.com