How should you respond when being asked during a job interview how to do user recall?

With the rapid development of the Internet, the cost of acquiring customers is rising, which means competing for limited users. When asked how to recall users during an interview, how should you respond?

Today, with the rapid development of mobile Internet and increasingly fierce market competition, the customer acquisition costs of some Internet companies have shown an upward trend year by year, which means that companies need to invest more resources to compete for limited users.

Conservative estimates: the cost of acquiring customers for tool categories is 1-5 yuan; the cost of acquiring customers for entertainment categories is 5-50 yuan; the cost of acquiring customers for e-commerce categories is 50-200 yuan; and the cost of acquiring customers for finance categories is 100-500 yuan.

For example:

Assuming that there are 1,000 new users every day and the next-day retention rate is 40%, 60% of users will never open our APP the next day and eventually become lost users, with an average daily loss ranging from several hundred to tens of thousands of yuan.

Although compared with the financing of millions to tens of millions, the loss seems small.

But unfortunately, things are obviously not that simple

The next-day retention rate of most products is far less than 40%.

This means more users will be lost!

And for many products, guiding users to download, activate and register is just the beginning. They need to continue to be guided to convert into paying users.

Here we extend a commonly used AARRR model. Because of its explosive growth method, it is often called the pirate model. Its essence consists of five stages: acquisition, activation, retention, monetization, and recommendation.

As a result, the longer a product operates, the greater the amount of silence and loss.

So how should we deal with the large-scale silence and loss of user groups?

I would like to hear your answers. I will give you three seconds. You can leave your answers in the comment section below so I can see them.

3, 2, 1

Ok, this problem may sound difficult, but do we have a better solution?

The answer is: Yes.

In the user life cycle model, the silent period is between the mature period and the churn period. Continuously reaching and keeping this part of users active through operational means is a key stage in determining whether the product can maintain high retention.

Accurate user tags combined with multi-form and diverse activation methods can help us improve user stickiness and thus retain users. Therefore, continuous recall and awakening of silent and churn warning users is imperative.

Now, we seem to have got the answer we wanted.

But is that the end?

No, this is just the beginning.

We know how to awaken silent users and recall lost users.

"How do you wake up a silent user?"

“How to recall lost users?”

OK, our content officially begins. I will share with you my exploration in building a "lost user recall system".

Since it is a lost user recall system.

So the first thing we need to do is to find the lost users.

“So what is the definition of lost users?”

Lost users: What are lost users? For tools and entertainment products, users who have not logged in for 7 consecutive days may be counted as lost users. Each type of product has different forms and needs to be determined based on the product form.

Silent users: Awakening silent users and recalling lost users are two different things, and the operation strategies are different. Therefore, it is also necessary to clearly define the silent period before loss and know what silent users are.

Naturally returning users: Every day, there will be some natural return of lost users, users who return naturally without any recall strategy. This data needs to be calculated clearly. If you do not make a distinction, when doing a recall, you will be blindly optimistic and think that the natural return is your recall effect.

Strategically recalled users: Users who return because of recall incentives. These are the incremental results of the recall strategy.

All definitions must clearly include two parts: time period restrictions and core behavior restrictions.

It needs to be determined based on the specific business and data analysis results.

Found lost users.

What data on churned (silent) users should we focus on ?

Scale of lost users: how many users have been lost during the data collection period, and what percentage of these users are lost in the overall user base.

User classification: percentage of lost (silent) users;

Channel aggregation: whether lost (silent) users have channel aggregation;

Life cycle classification: In which life cycle does user churn (silence) occur, the nascent stage or the decline stage?

By understanding the data scale, determining the priority of recalling user samples, and roughly understanding the difficulty of recall, in principle, silent wake-up is better than churn recall.

How to distinguish user value and recall priority?

In other words, what determines the order in which these users are recalled?

I'll give you three seconds to think about it, and you can leave your answer in the comment section below.

3, 2, 1

OK.

From the perspective of user value, we mainly segment lost (silent) users based on their historical value, difficulty of recall, reasons for loss, etc., and carry out targeted recall.

Behavioral data mining: Segment users by mining key behaviors.

Historical value: Determine the recall cost of various types of users through historical value. The main grouping dimensions include: user type, cumulative recharge amount, historical level, life cycle and other core business indicators.

Recall probability: The length of churn (silence), whether the client is uninstalled, and whether the user is reachable are used to group the users by recall probability.

Behavior analysis before churn and behavior analysis after return: Behavior analysis before churn determines the cause of churn, and behavior analysis after return determines the focus of user return, such as the activity attenuation trend before churn and the core consumption point after return.

Next, we need to do "user research" .

The main purpose of conducting user research is to solve the following two problems:

1. Loss reasons: Is it due to subjective reasons of users or platform reasons? Focus on the main reasons for loss on the platform
2. Return motivation: Find out the user's return motivation through research, give users a reason to come back, and shorten the trial and error period

Data mining can solve most of the bases for user segmentation, but it cannot determine the subjective reasons for user loss. Therefore, objective data + user research are the two core tools for user segmentation.

"Here is a reminder. Next, we need to learn more about the recall strategy. These are all useful information. If you find them useful, you can save them to avoid losing them when you need them in the future."

Crowd Strategy:

After grouping users according to user value, recall probability, and churn reasons, different recall strategies can be implemented for different groups to achieve refined operations.

Reach strategy:

SMS: The content delivery rate is relatively high. If the volume is large, the cost is only a few cents per message, but the content format is relatively simple, text + link, to achieve refined push, and the effect is more ideal when combined with push push.

Push: It can be reached only with an end, which may cause disturbances. Try to combine it with attractive activities, and judge the appropriate reach time for your users. Mastering the push time can achieve twice the result with half the effort. Here is a little trick. You can refer to the distribution chart of active time of the general users when designing the reach time and trigger frequency.

Email: In the early years of the Internet, emails were very popular for communication. This type of contact method was more commonly used by e-commerce and financial products, but the testing results were not ideal.

Telephone recall: Customer service costs are extremely high. Unless you are a very high-quality user, customer service telephone recall is not recommended.

Reward strategy:

Design attractive recall incentives, such as coupons, points, free trials, etc. What groups of people should be rewarded, what form of rewards should be given, cash, coupons or points? At what stage should the rewards be given, and what tasks should be completed to give rewards to stimulate users' willingness to return, these are the key points of the reward strategy.

“When designing a reward strategy, you need to pay attention to the balance between incentive costs and effects, otherwise it’s easy to go astray.”

Recall copy:

This is the most important part. This short paragraph not only gives him a reason to come back based on the reason for his loss, but also allows him to see the incentives for the future. All the segmentation research and data analysis are condensed in this recall statement. Therefore, experimental testing of copywriting is also a key link in recall operations.

Experiment & Product Perception:

• Experiment implementation: After the population is divided, the contact method is determined, and the reward strategy is completed, a large number of experiments are carried out to verify which population and which strategy has the highest recall efficiency and the lowest cost.
• Product perception: Try to build a landing page for recalled users and make users have a clear perception of the recall reward.

Recall user operation:

Lost user recall cannot stop at getting users to come back. At this point, the user is equivalent to going through the life cycle of a new user again. It is necessary to focus on the 1-, 3-, 7-, 14-, and 30-day retention numbers of the recalled users until the users enter a stable period.

We have done a lot of work from user recall (awakening) to actual implementation. The whole process is time-consuming and labor-intensive. It is very costly to recall a lost user. In order to avoid doing a lot of repetitive work, we should give early warning when the user is about to become silent and leave, and use product guidance and incentives as far as possible in advance to help users through the shock period.

Churn (silent) warning: Through big data machine learning, the risk of user churn can be predicted, and targeted and regular pre-churn intervention can be performed, which will greatly reduce product operating costs.

This is almost the end of this issue, let’s make a summary.

Against the backdrop of increasingly fierce competition in the mobile Internet, the cost of acquiring users has been rising year by year. For Internet companies, it is particularly important to effectively recall silent and lost users.

Any operational means and strategies must be based on a deep understanding and thinking of users and products. This process requires a deep insight into the characteristics of users and products, otherwise all efforts may be wasted.

Recalling silent and lost users is a complex process that requires continuous investment. It requires the comprehensive use of personalized recall content, multi-channel reach, attractive incentives, and continuous optimization of user experience.

At the same time, through continuous data monitoring and analysis, and continuous optimization of recall strategies, we can improve user stickiness, reduce churn rate, and maximize the user life cycle value.

I am classmate Daming.

See you next time.