When taking over a new business/requirement, what is the underlying thinking framework?

How to think systematically and develop effective strategies in the face of new business or needs is a key skill that every product manager and business owner needs to master. This article uses a specific project case to introduce in detail the underlying thinking framework that can be used when taking over a new business.

When you introduce someone to a project you’ve worked on, you can explain it from the following ideas:

1. The business I do is XXXX, and I am responsible for the XXX link (or I am the person in charge of this business) in this business, and I am responsible for the XXX indicator.
2. At that time, the current business situation was XXXX, but we wanted to achieve the goal of XXX, so we can describe the difficulty of achieving this goal.
3. Around the goal, I constructed the formula for XXXXX, around which the formula was for the XXX parameter, hypothesis 123 was proposed, and the final result was XXXX.
4. If your business formula cognition or the project as a whole still has a process of iteration, refinement, and deepening layer by layer, it can also be described.

In actual work, when we take over a new business/project, we can also start from the above framework.

Below, I use an example of business and data desensitization:

Business background: After the growth of “tools + content” has entered a relative bottleneck period, there is an urgent need for new ideas or directions to move growth to a new level (scene words). In other words, with just so many resources and so much time, hurry up and “do whatever it takes” to give me growth. So, analyzing and analyzing, combing and combing, the idea of “social belt growth” was adjusted at the critical point of a certain neuron. If you want to convince the bosses to establish a project, you have to give logical proof, so start to do the paper logical deduction and demonstration of the business feasibility before the project is approved.

In the paper argument stage, two hypotheses:

1. user value assumptions;
2. business growth assumptions;

The two hypotheses are logically valid (including qualitative quantitative-data/research, scientific analogies) in turn. If one does not hold, there is no need to continue, and even if it continues, there is a high probability that it will fail. Of course, some people may ask here, what if it is just a logical argumentation process that is wrong, but may it actually be true? There is such a possibility, but you think that the person in charge of this business can’t even do this step correctly, then even if the follow-up enters the project stage, the corresponding product plan, growth strategy, etc., are based on the same person in charge, so what is the probability of subsequent success?

To achieve these three challenges, product managers will only continue to appreciate
Good product managers are very scarce, and product managers who understand users, business, and data are still in demand when they go out of the Internet. On the contrary, if you only do simple communication, inefficient execution, and shallow thinking, I am afraid that you will not be able to go through the torrent of the next 3-5 years.

View details >

Therefore, how many points a person has in cognition, how many points can he solve, and the overscore is a small probability event, and in the long run, the mean will be reverted. Therefore, doing growth and entrepreneurship are essentially doing probability, and the size of the probability depends on the cognitive level and a little luck of the person doing it.

If there is a God’s perspective (God’s perspective: all objective facts/theories/laws exist objectively, God’s perspective can see the whole picture, the human perspective can only see the part, and everything that humans do is infinitely approaching God’s perspective – assuming that God’s perspective’s cognition is 100 points), then there are two kinds of each needobjectiveStatus: Either holds or doesn’t. God knows thisfact。 And we don’t know. What we can do is to deduce first to see if it is logically true, and if it is not logically true, there is no need to continue. There is a high probability that it will fail if it is launched. If true, then put forward the key hypothesis (Hypothesis driven), (based on scientific analogy/xx model or method/qualitative quantitative analysis) to give the minimum solution, put it in front of the user (practice-iterative cognition), and actually verify.

1. First, demonstrate whether the user value hypothesis is true;

First, clarify what user value is:

That is, what problem users encounter in what scenario, and what solutions we can provide to solve this problem, this is user value. In other words, it is: userswillingChoose your smallest product solution. What the product wants is to “do whatever it takes” to make more users perceive this user value faster, and the probability of the user group staying will be greater.

Demonstration: Clarify the target user group, estimate the size of the group range, bring it into the scene, and judge whether there is a demand.

For example, (strangers) social growth is growing: large-cap users can be potential users, divided into social active initiators and passive recipients. The minimum core is the contact-maintenance-establishment of one-on-one social relationships between strangers. The scenario is based on the “assumption is content” scenario of the main business of the product, so users have a common carrier for the scenario and may have a common topic. Therefore, logically judging that the group is objective and the core is established.

2. Demonstrate whether the business growth assumption is valid (i.e., how much room can be achieved for business goals);

Demonstration steps: Based on the assumption of user value>> clarify the business objectives>> disassemble the business objective formula according to the business objectives>> and then measure the size of the business objectives according to the factors in the formula.

After the above two hypotheses are demonstrated, according to the factors of the business goal formula, judge what are the key stuck points that affect the achievement of the business goal (on the left side of the equation), how big or small the influence of each factor on the business goal, what key hypotheses can you put forward for each factor – verification (including the dismantling of the second and third level indicators of the factor), what are the collaborators on which the factors are to be promoted, and what is the development cost, so that the ROI of each assumption is estimated>> the demand pool comes out, and the priority also comes out.

How to disassemble the business goal formula: start with the end~

Business formula definition: The business result indicators are expressed in the form of mathematical formulas with key factors.

1. First, determine the results that the business needs to get the most; (The most important result in the following example is “user social connection”)
2. and then find out the corresponding data indicators, that is, business objectives; (“Social Pairs”)
3. Identify all core process indicators around business goals;
4. then determine the relationship between these indicators, either multiply or add; (At this point, the formula is out)
5. Identify the core factors that affect the life and death of the business
6. Key hypotheses are proposed for each factor (each hypothesis may be a requirement in the business iteration process, it may come from data analysis, it may come from user research, it may come from competitive product analysis, it may come from… ）

Note: DistinguishCharcoal is sent in the snowandIcing on the cake

Detail step 5, which is very critical and important of the above steps:

Based on the disassembled formula, find out the core factors that affect the life and death of the business, based on the core factorsput forthmultipleKey assumptions, find out among themThe most important key assumptions, that is, we need to determine which hypothesis is the one that will die if the business is not done well (Charcoal in the snowThis problem is at different stages of business development, facing the same business formula,The answer is subject to change.Therefore, at every business stage and every important review, always ask this question, so that you and your team can always enter a state of deep thinking, which can enhance the team’s understanding of the business, and also help to reach a business consensus.

Ideally, each core process indicator may be a relatively independent target result, and it can continue to be dismantled from its secondary indicator formula (choose whether to continue to dismantle as needed). At the same time, it is necessary to estimate the significance of the influence of other factors to the left level of the equation, determine which factors are more important, and which factors to do first (Charcoal in the snow), which factors to do later or not to do (icing on the cake）。

Finally, key hypotheses are put forward for the corresponding factors>> verified>> iterative cognition>> and continue to put forward hypotheses.

Step 6What does it mean to come up with key assumptions?？ It refers to a certain factor (indicator) on the right side of the equation, if we want to improve it (because the improvement of the right factor will lead to the improvement of the business goal on the left), what methods/solutions/strategies, etc. can you think of – these methods/methods/strategies are the key assumptions, which need to be verified, may be proven, may be falsified, and are probabilistic events. Therefore, the essence of growth/innovation/entrepreneurship is probability. A few points of cognition, a few points of solving a problem (see the description in the figure above), solving a problem beyond cognition, the probability of failure will be greater. It is important here that it is very similar to “I think you are wrong”, “I think it should be like this”, “I think it is better this way”, etc., but these are “subjective” opinions, and the hypothesis is an objective judgment, a hypothesis, before it is verified, is just an assumption, not a fact. Once you have a subjective tendency, you may not want to verify it (of course, unless you have verified something similar before, or someone else has verified it), and it is likely to be wrong.

The above process,Basically, the formulation of business goals, the assessment of process indicators, the standards of team division of labor, and the priority of core tasks have been completed at the same time. From this time on, the process of “simplifying complex things” and “standardizing simple things” in business management has been completed。

(This is the process of business goal management)

Above, the specific process is as follows:

Chestnut – social business, user value assumption explained earlier. Now based on user value, the phased business goals are determined.

Social business serves growth, that is, it serves retention and revenue growth. That is, we need to establish social relationships between users (“1,First, determine the results that the business needs to get the most“), through “relationships” to increase user activity, and then improve retention (revenue aside, the path is too long). So, in what form do users build social connections? What can be directly thought of is the IM session scene, and apart from this scene, you can also think of a creative user (for example, when the user publishes a content, because since he has a demand to publish content, we assume that he currently has an external social appeal, and the social appeal here may be an IM session, or it may be an interactive response from others after seeing the content I post), we mainly use the IM scene as an example to illustrate, and we will also bring interactive scenes.

Supplementary note: The establishment of “relationship” means that there is conversational communication or other forms of interaction within a certain period of time. Even if they no longer communicate after this cycle, they have already established a relationship with each other – from strangers to othersrelativelyAcquaintance. In terms of data, they are relatively better than users who have not entered this stage in other “interactions”, such as browsing each other’s works and liking and commenting on each other.

Therefore, we can think of the indicator of “stranger social relationships” to measure our business goals, that is, the number of users who have successfully established social relationships, we will change the word, call it “stranger social logarithms” (in the above disassembly steps)2. Then find out the corresponding data indicators, that is, business objectives, according to the results”）。

At this point, the phased business objectives are determined:

Strangers socialize in pairs

• Stranger’s Caliber: There has never been an IM session recorded in history
• The caliber of social pairs: There are m session records within n days of the first session (n and m can be obtained by running the data of the target user in the circle in the early stage, and may be adjusted according to the actual results of the hypothesis test in the later stage)

Next, we need to break down the formula based on business goals (i.e., the right side of the equation), because our product strategy cannot directly affect the “social logarithm” metric, just as it is difficult for the strategy to directly intervene in DAU. In addition, if the business formula is not dismantled, the result is that the subsequent project promotion is scattered and subjective.

Formal disassembly process: (“3. Focus on business goals and find out all the core process indicators; 4. Determine the relationship between these indicators, either multiply or add”)

Stranger Social Pair = DAU * Message page visit rate * Sent session rate * Sent reply rate * Session rate in n days of first chat

Among them, (the above formula is not a good disassembly, just to illustrate why it is not good to give an example):

• DAU is a relatively stable value, and it is difficult for us to intervene in the short term, and as mentioned earlier, it is difficult for strategies to directly intervene in DAU. So, DAU shouldn’t be here.
• The message page access rate, the message page is a first-level tab page of the APP, which is directly associated with a specific functional scenario, and should be attributed upwards and replaced with other indicators that do not contain specific scenarios. (The first-level formula usually avoids the indicators that contain specific scenarios, otherwise the value to the left of the equation will be limited to the traffic size of the current scenario) Visiting the message page is for sending, so the real metric should be “number sent” for up-attribution.
• Sending session rate (how many people in the DAU will send), this is no problem, no specific scenario.
• Sending response rate (how many users who receive the session will reply), this is fine, not including specific scenarios.
• The M-day session rate within n days of the first chat, which is commonly referred to as the “” factor, is determined according to the actual business. It means that after the first chat, there will be m days of chatting in the next n days. The standard for measuring whether the relationship is paired when it works. Only the first chat is not considered a pair (explain, how to count as a pair, just set your own business goals). In addition, this indicator is a lagging indicator, and its difficulty in improving should not be small.

So, the formula becomes:

Social pair = number of sessions sent * sent reply rate * m day session rate within n days of the first chat

At this time, the number of sent sessions is not limited to the message page scene, but other places can also find ways to let users send conversations, and this indicator has some room for improvement. However, according to the current data status of the indicator, even if the factor “number of sessions sent” doubles, the social pair on the left side of the equation will double, but the absolute value is still too small. The absolute value of social logarithms should be at least a hundredfold increase (Difficulty in achieving goals), which will have a significant effect on the retention of the market. Here’s a description:

For example, if the DAU is 1 million, it is assumed that there are 1% of users who are willing to actively socialize, that is, 10,000. If each person sends 1 message a day, then 10,000 people can receive it, assuming that the crowd does not coincide, then 20,000 people will be covered by “social relationships”, and what about the other 980,000? All are “empty”. How to fill this void? Two ways: one is to let more users take the initiative to “send”, and the other is to let more users passively “receive”. Take the initiative: Due to human nature, only a small number of people are willing to take the initiative to socialize (really take action), so even if we can triple 1% (which is very difficult), there are still more than 900,000 “empty”. So, can we get from passive, that is, when a user sends a message to another user, find a way to send the message to more users (with the consent of the user who sent the message). Like whathypothesisEach message can be distributed to 100 people, so if 10,000 users send it, 1 million users will receive it, and the penetration rate of social relationships will go up (This assumption is also the most critical assumption to achieve business objectives –In the disassembly step above“5. Find out the core factors that affect the life and death of the business”(Eh, it’s over, it’s okay, it’s just a hypothesis)Therefore, a key factor is missing from the above formula: the number of senders in a single session。 With this factor, business goals can be improved from spaceLogicallyIt directly increased by 100 times.

Thus, the formula becomes (final formula):

Social Pairs = Number of Sessions Sent * Number of Distributors in a Single Session * Reply Rate Sent * Session Rate on M days within n days of the first chat

The above factors are all multiplicative relationships, and how much each factor increases will directly lead to how much the value on the left side increases. After the business goal is disassembled (formula), next, it is necessary to judge the size of the improvement space of each factor and the target factorFormulate key hypotheses (In the disassembly step above“6. Put forward key hypotheses for each factor”), and then estimate the ROI of each hypothesis and give priority.

Key hypotheses are proposed for each factor:

Number of sessions sent, there is an analysis before, assuming that it can be increased by 3 times;

For this factor, a hypothesis is proposed:

• Hypothesis 1: When the user enters the conversation page, the greeting text is exposed to the user in some form, such as tags, which can increase the probability of the user sending it.
• Hypothesis 2: When users enter the conversation page, use the large model to generate greeting copy, which can increase the probability of users sending it.
• Hypothesis 3: Guide users to initiate a conversation on the user’s main path (for example, if there is a function in the main path of the user on the site, when the user publishes a piece of content, at this node, we extract the relevant information of the current content, combine it with the large model to generate relevant speech copy, and then package it, so that the user can decide whether to send this “package” to many people at the same time), which can increase the probability of users sending sessions. Because in the current scenario, since users are ready to publish content to the outside world, we can assume that users have external social demands – sending out content, wanting to receive recognition and interaction from others, etc.
• Hypothesis 4: In the later stage, if the gameplay runs through (the factor of the number of people distributed in a single session below runs through), the user is provided with an active “session” to initiate outwardFunction Ato improve the session sending rate.

There are also branches in the iteration processhypothesisI won’t list them all.

Therefore, this factor, after the above hypothesis verification, gets its own disassembly formula:

Number of Sent Sessions = Number of Message Page Sent Sessions + Number of Sent Sessions of Primary Path + Number of Sent Sessions of Feature A

Send response rate, there is an analysis in the front, assuming that it can be increased by 10 times;

The analysis of the data found that the response rate was strongly correlated with the form of the greeting speech and whether the other party was online when sending the conversation. So:

Hypothesis 1: When a user sends a conversation, guide the user to send a “high response rate (relevance)” phrase, which can improve the response rate.

After the online AB experiment (hypothesis test), it was determined that there was a causal relationship between the content form and the response rate.

Hypothesis 2: Based on the verified hypothesis 1 (which is now a “fact”), using AI to generate high-response rate content in real time based on user scenarios can improve the session response rate.

Hypothesis 3: The algorithm distribution strategy increases the priority to users who are online at the time according to whether the other party is online.

After the online AB experiment (hypothesis testing), it was determined that there was a causal relationship between “online” and the response rate.

Hypothesis 4: Based on the verified assumption 3, distribute part to users who are online at the time, and leave a part to distribute to users who are not online at the time, but if they open the APP before n o’clock on the same day, they will be distributed to them after opening (in the meantime, the hypothesis of “n” of “n points” can be put forward for verification, and more hypotheses can also be put forward to increase the response rate after opening m minutes of distribution)

Hypothesis 5: New users are sent to new users, old users are sent to old users, and new and old users are cross-distributed. (Among them, new users can also be sliced according to the time, such as new users within 7 days, new users within 14 days, etc.) (But here may be because the proportion of new and old users is too different, so the overall effect may be very small when viewed evenly, so the hypothesis value here is not high)

Hypothesis 6: The average number of sessions received by new users is 1/2, and the average number of sessions received by highly active old users is 0, 1 or 2. (For example, suppose before verification, it is better to send 1 message to a new user, because you are afraid that it will disturb new users and cause churn, or send 2 posts, because you may post multiple posts, and the user is interested in one of them, which may improve the response rate.) Because there is a positive causal relationship between the response rate and retention; The same goes for highly active users, for example, if the user is already very active, it is difficult to say that if you send him a conversation, it will increase his retention, or it will have the opposite effect. Or if he is already very active, then the probability of his reply should be greater, which in turn can improve the sender’s retention, etc.) (However, it can be considered in advance that only a small number of users can send 2 messages due to the number of sent sessions, because the priority is to ensure that the user receives 1 message, and the remaining number of sessions will continue to be sent, so the value of the assumption here is not high)

Hypothesis 7: The user who receives the conversation, after entering the conversation page, the large model generates a good reply for him, and he can reply directly by clicking to reply, which can improve the response rate; (However, this assumption may make users feel fake and affect the continuation of follow-up social relationships after the same day, so this assumption needs to be cautious)

Other hypotheses: will be “The number of sessions sent by the message page“This factor is dismantled by one layer:

Number of message page sent sessions = Number of message page visits * Stranger session exposure on the message page * Session list click-through rate * IM session page message response rate (If necessary, you can continue to dismantle)

Relevant hypotheses can also be proposed for each factor of this sub-formula, but this hypothesis is a low priority after evaluating the ROI. Don’t consider it for now.

Why this factor (Send response rate) is superfluous to the above factor (Number of sessions sent）？ Because confidence in this factor is greater – 10 times; And the implementation strategy is simpler than the previous factor, many of them are algorithm adjustment distribution rules, and the previous one relies on functional adjustments, relying on clients, development costs and cycles are greater, in the final analysis, it is a comparison of ROI.

The first chat is the m-day conversation rate within n days, there is an analysis in the front, so it will not be considered for the time being;

Number of people distributed in a single session, this status quo value is 1 (each user sends an average of 1 session to a few strange users), and after the algorithm distributes the access, how many users are distributed – mainly determined by 3 numbers, one is DAU, the second is the number of sessions sent, and the third is a user receives a maximum of several sessions without being disturbed. Because, assuming that every session is distributed to the full DAU, if there are 100 sessions to be distributed, each person will receive 100 sessions, it’s scary, uninstall. Finally passedSeveral roundsIn the test, the value of the number of people distributed in a single session is about 100 (desensitization data), and the average person receives a maximum of 2 sessions (desensitization data) within 15 days. (Here.)Several rounds: refers to multiple sub-hypotheses proposed and tested in the process, assuming that the same user receives up to M sessions in n days, and n, m and m have corresponding hypotheses）

IfDAU is 400,000 (hypothetical), the session sending rate is 1%, then 4,000 people will send 1 message per person per day (note, the crowd in the daily slice is dynamic, because DAU is dynamic, not that today is this group of people, tomorrow is still this group of people, there are newcomers, there are returners, there are yesterday and today) Each message is sent to 100 people, then there are 4,000 * 100 = 400,000 people in DAU will receive the session, that is, everyone will receive it, which is 100% compared to the session penetration. The actual situation is that whether the user receives the session and the factors that affect the session response rate will determine the timing of distribution, which means that in fact, the distribution strategy here is very complex to distribute 100 people in a single session, and it is difficult to distribute according to the theoretical and logical rules.

Why is there a 100-fold space for this factor, but there are not many “hypotheses” proposed. Because this key hypothesis is easy to verify, there is not much room for subsequent improvement after verification (if the verification is not valid, there will be a follow-up). In other words, compared to this 100 times, it is pulled up by one hypothesis (ignoring the testing process of how many people it is distributed to in the process), but this hypothesis is crucial, and if the hypothesis is not validated, the value of the whole project is not great.

From a quantitative point of view, explain why “The first chat is the m-day conversation rate within n days“This factor: If this factor is removed, then:

Social Pairs = Number of Sessions Sent * Number of Distributors per Session * Sent Response Rate That is, the formula means that as long as a round of back and forth sessions are started between strangers, it is considered a pair relationship. However, there is no significant causal relationship between this indicator and the first-level indicator of growth (retention), that is, even if the user completes this behavior, there is only a certain improvement in retention, and there is almost no change in retention after retention, such as 3, 7, 15, etc. So I think from the perspective of higher-level business target growth, this is not a high-ROI indicator breakdown.

Summary: Business goal management

• validating user value assumptions and growth assumptions;
• Define business objectives;
• dismantling business formulas based on business objectives;
• Identify the life-and-death factors that affect the success or failure of business formulas and put forward key hypotheses.
• According to the situation, the formula of the lower indicator is disassembled;
• It is estimated that the weight of each factor on the improvement space of business goals is small;
• According to the factor weights, key hypotheses are proposed>> MVP validation >> iterative cognition.