5 tips for user recall!

This article shares how to recall users from the perspectives of who left, why they left, how to recall them, how to measure the results, and how to review and adjust.

01

User recall is one of the common ways to improve overall daily activity, in addition to new user maintenance, activation of existing users, and churn warning. Through this article, you can master the ideas and methods of user recall.

A dry methodology is soulless. Next, let us explain how to conduct a recall with examples.

Whether it is a high-level combination of model algorithms and automated tools for push, or manually exporting user group data packages, using tables for portrait analysis, and then formulating strategies for recall, the general idea of ​​user recall can be summarized as follows:

1. Give users who once loved you a reason to love you again;
2. There are thousands of users, so we need to prepare more kinds of "sweet words" this time;

02

1. WHY

Before we start doing a recall, we need to figure out why we are doing a recall:

1. The number of people leaving is greater than the number of newcomers? ——The current situation requires backend statistics on churn rate (new additions/losses), which is not enough to cover expenses
2. Is the cost of recalling and awakening old users lower than the cost of attracting new users? According to statistics, the cost of recall is only 1/5 of the cost of attracting new customers in the market - low-cost new customers, indirect new customers
3. The boss decided on a whim? ——Growth is weak, try all the methods you can use

Real example: "My boss said impulsively: Xiao Zhang, our active data has been growing slowly recently, and the loss rate is quite high. Can you help me organize a campaign to recall lost users this month and see if it works?" - Example of a mobile voice platform

1. How

• Who left?
• Why leave?
• How to recall?
• How to measure effectiveness?
• How to review and adjust?

That solves the problem of why and whether to do a recall. Then there is the question of how to do it.

(1) Who left?

Definition of lost users: users who have not performed key operations within a period of time.

For example: users who have logged in 5 times in each calendar month in the past year and have not logged in in the last month; the definition of churn needs to be adjusted based on the product attributes and user characteristics; the definitions of information content and finance are different.

Several angles to define lost users:

• Confirm key product behaviors; pay? Log in? Send a site message?
• Confirm the length of time that no operation has taken place;
• Consider recallability. Here, the actual natural return rate of users at each group level can be used as a criterion.

For example, among users who have not logged in for the past month and whose daily historical active time exceeds 2 hours, the natural return rate will definitely be different from that of users whose active time is only a few minutes;

As the product in this case is a mobile voice content community app, the operations staff used the length of time users spend consuming content as the core judgment criterion;

Different from community platforms such as e-commerce and travel, user consumption has a certain periodicity and there are actual cost expenditures.

Therefore, for mobile voice content platforms, the definition can be as follows : mobile phone registered users who have logged in more than 5 days per calendar month in the past six months and have not logged in in the past month (first circle out the user group in a broad sense, and then segment it based on key path characteristics, recall touchpoints, etc.)

(2) Why did you leave?

What time did you leave? Where did you go? Where did you go?

If conditions permit, you can ask the UE department to conduct a user churn survey to facilitate subsequent product function or operation optimization; if conditions do not permit, you can ignore it and directly conduct strategic recall of lost user groups, and then gradually optimize in the future;

(3) How to recall?

Develop recall strategies based on the characteristics of groups at each level.

In the first half, we have broadly identified the lost user group. The more detailed group division can be based on user behaviors such as the quality of lost users and the PMF model .

The biggest purpose of figuring out the portrait of the lost group is to stimulate them through certain means in a targeted manner based on the group portrait, and do their best to reawaken and recall users.

For mobile voice platforms, combining platform attributes + recallability + emotional resonance of the scene , the operators believe that the users most likely to return will be the loyal fans of the core KOLs on the platform.

The G-spot of the core fans of this platform may be their favorite anchor or content, because it is a content-based community; this will affect the design and direction of subsequent recall copywriting.

The actual operation is as follows:

Assumption: The longer the broadcast time is, the higher the activity is, the more loyal the audience is.

Specific data dimensions, subdivisions:

① Number of days after the broadcast is completed (if a user completes more than 90% of the broadcast of a certain anchor's program, it will be counted as a valid day)

At that time, if a user has broadcast a certain host's program for 3 full days or more, he or she will be considered a loyal fan of the host.

②Playing time

The data can be further subdivided by sorting the total playback time among users who have completed the broadcast for more than 3 days, and dividing the user groups into different levels according to the proportion of the data duration range.

For example: Among the listeners of anchor A, among the users who have finished broadcasting for 3 days (set threshold), 20% of them have a playback time of more than 2 hours and are set as S-level users, and so on.

OK, the groups are clearly divided, and the strategies for each level are also in place. Now let’s design a recall plan (starting from the content, with emotional resonance as the main stimulus). The general triggering methods for recall are: email/SMS/in-site message/push, etc.

Based on effectiveness and feasibility, SMS was chosen in this case, combined with H5 links. It is best to formulate more than two SMS copy for each strategy to facilitate grayscale testing of the effect. As for the effect, well...

(4) How to measure the effect?

The conventional criteria for measuring recall activities may be: if the lost users open the app within a unit of time after the SMS is sent, it is considered a successful recall, and the recall rate = returning users/total lost users; then calculate the SMS cost and that’s it.

This criterion is correct for the activity itself, but it is not accurate enough for the essence of recall:

First, users who have churned within a unit of time may not open the recall text message immediately after seeing it. Does it count as a successful recall if they open it later?

For example, if the activity definition is that if a lapsed user opens the message within 24 hours after it is sent, it is considered a success. Then, if it is opened within 48 hours, is it considered a success?

Second, using recall quality as a metric would be a more accurate metric.

Recall quality = the comprehensive score of returning users after quantification of effective behaviors such as core behaviors and retention.

For example, in this case, after the user returns, within 7 days, the user who has completed the video for more than 30 minutes is considered a high-quality returning user...

(5) Review and adjust?

Analyze from an operational perspective whether user classification, recall costs, choice of contact methods, choice of copywriting, choice of gifts, etc. are reasonable.

In addition to the need to optimize each node in the first half of the recall process, the care after the return is often something that operations colleagues tend to overlook.

For example, Honor of Kings does a great job in this regard - it gives out big gift packages to returning users, thereby increasing the incentives for users to return.

From a product perspective, is there room for optimization in the landing page selection when users first open the APP, the waiting time from clicking the short link to opening the APP, etc. From a technical perspective, can the recall process be automated?

The so-called solidification and toolization of activity processes: through scripts or background functions, periodic recall of lost users is carried out. This requires operations staff to have a certain product thinking and also to consider feasibility, such as the relationship between development costs and the value of the recalled group.

03

Let’s think beyond the recall campaign and think about the entire user life cycle.

If we focus on the entire user life cycle management, we need to formulate different strategies and measures such as prevention/enhancing activity/anti-churn for users at each stage, and continuously follow up and optimize to improve the conversion rate of each node.

For example:

1. Introduction stage: customer acquisition/precise positioning/discovery/optimization

Explore and optimize the volume-carrying capabilities of each channel, leverage activities to attract new customers at low cost, push high-quality content, recommend friends, etc. to enable them to form content consumption and consolidate social relationship chains.

2. Growth/Maturity: Improve activity and promote transactions (check-in/points/wealth system/activities, etc.)

Based on the mobile voice platform product in this case, what would you do if you encountered the following problems during the growth stage?

• How to stimulate those who consume content but don’t follow friends?
• How to stimulate those who consume content and have a certain number of friends but do not like or collect content?
• How to stimulate those who like/collect but don’t share?

3. Dormant/churn period: early warning/recall

I hope the above can give you some inspiration. Pat it lightly

Author: jim2018

Source: jim2018