The activity rate is less than 20%. What is more difficult than attracting new users is awakening dormant users. What can we do?

The cost of attracting new users is getting higher and higher, while user activity is getting lower and lower. How to effectively wake up dormant users is a difficult problem. How to solve it? The article shares some analytical ideas, let’s take a look.

Two major challenges of mobile Internet : the cost of attracting new users is getting higher and higher, and the activity is getting lower and lower

The Internet has been developed for more than 20 years since 1995, and the demographic dividend has also been enjoyed for 20 years. Since 2016, the cost of traffic has become increasingly high, and it has become increasingly difficult to acquire new users. Even if you have money, you cannot buy reliable traffic.

In 2012-13, the cost of acquiring a single user was only a few cents, but now it has increased more than tenfold. The period of rapid development of the Internet has passed, which means the fierce competition is over. The traffic entrances are occupied by leading companies, and the user coverage of BAT and unicorn companies ( iQiyi , Youku, Meituan , 58.com, Didi , and JD.com ) has reached its peak. Since 2016, for most advertisers, not only is attracting new users a big problem, but retaining existing users and promoting user activity is an even bigger challenge.

16 years ago, products were king in the mobile Internet . Since then, products have become highly homogeneous, and it has become increasingly difficult to realize user monetization through product differentiation. After 16 years, mobile Internet has gradually become a place where operations are king. Whether a user uses you is not ultimately determined by product design. It may be that a certain activity attracted him or a certain refined operation retained him. Now, whether it is advertisers or investors , they not only look at your user volume, but also at activity, retention , and ROI.

Therefore, everyone tries their best to keep users active on the APP and strive to gain more users' attention resources. However, the extensive way of attracting people and traffic brings a lot of problems, such as attracting a large number of fake users. If one-third of your users are fake users, you will probably be more anxious than anyone else. Internet companies often claim to have hundreds of millions of daily active users, but they dare not say what proportion of the total user pool their active users account for, let alone say that the actual activity rate of their own users is less than 20%.

So, what are the 80% of dormant users like? Take myself as an example. I have installed 4 or 5 e-commerce apps. Some of them I use once a month, and some I don’t use at all in half a year. These users are marginal users and dormant users of the APP. For advertisers, their actual value is very small if they are not activated or awakened.

Compared with attracting new users, awakening 80% of dormant users is the most cost-effective

It is a consensus among everyone that the cost of attracting new APP users is getting higher and higher. In comparison, awakening 80% of dormant users is the most cost-effective thing. Its advantages are:

1. There is no educational cost for this group of users;
2. The awakening cost is much lower than the new customer acquisition cost;
3. As long as there is suitable content to trigger users to become active again, the effect will be far better than attracting a new user.

Awakening dormant users is essentially a part of refined user operations. The operations staff is responsible for the active KPIs, while the promotion staff is responsible for the newly added KPIs. It is said that there is a fierce conflict between operations and promotion in most companies now. If operations want to achieve good results without being blamed, they have to increase user activity, the most important of which is to wake up dormant users.

Generally, effective wake-up is defined as: for users who have uninstalled the APP, the wake-up activity prompts them to click a button to jump directly to the APP download page, allowing users to download it again. For users who have not uninstalled the APP, encourage them to click a button to open the APP directly. Sleeping users need some stimulation to trigger them to become active again, but can anyone wake them up? Some people say that since the user data is in my hands, I can just push activities to him to wake him up. Of course, it’s not that simple.

Why do many companies (especially large and medium-sized companies with a large number of dormant users) have the need to wake up users? Because the awakening effect I did myself was not very good. Let's think back, do you often receive text messages from a certain APP on your mobile phone, or event reminders within the APP? Most of the event content is something you are not interested in at all, such as loan promotions for users who do not need loans, or free travel activities for new mothers who are not convenient to go out. Even the best wake-up campaigns are useless if you don’t understand your users.

In order to tackle the tough problem of "user awakening", we must combine big data technology

Currently, the most effective way to awaken users by some leading Internet companies is to combine big data technology. In actual operations, user portrait analysis , traffic identification and anti-fraud technology are used to make user awakening more accurate, users more real, and feedback more active.

1. Use big data technology to analyze user portraits and achieve accurate awakening

Most advertisers know very little about their users. Although they have device numbers, they have no idea about their users’ interests, hobbies, spending habits, and consumption levels, so they cannot achieve efficient wake-up.

The quality of user awakening is determined not by who has the user, but by who knows the user better. Big data companies, based on their own massive mobile user data, can use big data technology to complete accurate user portrait analysis, and on this basis, locate precise users for efficient awakening.

Taking the JD Shopping Festival as an example, a shopping festival is divided into many different groups, so targeted push notifications must be made based on in-depth analysis of users. For example, mother and baby activities are pushed to mothers and babies, and 3C product activities are pushed to white-collar men. This way, the chance of awakening will be higher. We can even push more detailed content based on gender, region, occupation, consumption level, etc.

Whether it is user awakening or other targeted delivery, user portrait is the most important part. Our user data dimensions include multiple dimensional labels such as demographic attributes, interest maps, product maps, and behavior maps:

• Demographic attributes: gender, age, education, marital status, children, income, and credit.
• Interest map: for example, games , finance, maternal and child care, live streaming , etc.
• Product maps: game product map, maternal and infant product map, financial product map, and APP map.
• Behavioral graph: Behavioral segmentation based on interests, such as whether game users pay.

How accurate can user portraits be? The figure below takes a car user as an example. All the user's social behaviors are analyzed. You can learn that the user has a recent demand for buying a car, as well as his spending power, preferences for car models, performance and prices, etc. Only on this basis can a more efficient user conversion plan be made. Just imagine, if you push mid- to high-end cars to a user who has no need to buy a car or is only interested in a relatively low price range, there will definitely be no conversion.

User portrait analysis is a preparation for refined awakening. Only when you understand your users well enough can you understand what they need most from their perspective. The user portraits created when awakening users of different products are also different, depending on the product.

In addition, about 25% of APP IOS users are zombie users, and they can be effectively eliminated in advance by combining big data anti-cheating technology. In other words, false users have been excluded before conducting user portrait analysis, and ensuring the authenticity of users will also increase user conversion rates .

2. Design and select the activities and copywriting with the best conversion effect based on user scenarios

First, design targeted activities and copywriting. Second, during the test launch phase, select activities and copywriting with higher conversion rates for target users based on data feedback. Generally, awakening activities should meet the elements of convenience of participation, satisfying users' curiosity, and providing reciprocal temptation. The picture below is an activity of a group buying food APP. After the test, it was found that the wheel lottery activity had the highest conversion rate. The lottery satisfied users' curiosity and sense of participation, and the grand prize was extremely tempting to users, so this activity can instantly trigger users' click behavior.

Each time we launch a new campaign, we also need to test the conversion rate of the marketing copy and select the copy with the best conversion rate for subsequent release. We usually select at least five copies from more than ten copies for release, and then select two copies with high conversion rates for secondary release.

• Copywriting 1: Lucky competition, click to draw, enjoy a Mac Book and a trip for two to Dubai. Free food, drinks and fun, click here if you don’t believe me;
• Copywriting 2: As a big thank you, you still have one free chance to win a prize! A mysterious grand prize is waiting for you, come and participate in the wheel draw!

In this case, the conversion rate of the second copy was significantly higher, and it was eventually released on a large scale. When creating copy, try to maximize the user click guidance as much as possible. For example, the "notification + suspense" of Copy 2 can better guide user clicks, while the declarative sentence structure of Copy 1 is much less guiding. In actual operation, evaluate and optimize based on the data conversion results.

3. Maximize data tracking and refined operations during the delivery process

It is not appropriate to directly release a large amount of data in the initial stage of launch. After testing a stable conversion rate with a small amount of release, optimize the release strategy according to the release efficiency, and then increase the release scale appropriately.

During the delivery process, data is updated on a daily and weekly basis. In fact, you will see that the wake-up conversions are different every day of the week (tested under the condition that the delivery volume is the same every day of the week). After the wake-up conversions stabilize, adjust the daily delivery volume appropriately based on the daily conversion situation.

The daily wake-up time should also be optimized in real time based on the conversion data. For example, in the initial trial stage, when it is unclear which specific time point will have the best conversion rate, try to choose as many time periods as possible (such as 10:00-12:00, 17:00-19:30). When it reaches the stable stage, by comparing the awakening conversion rates, we finally select five time points with higher conversions (11:00, 17:30, 18:30, 19:00, 19:30) for subsequent delivery. The last stage is the refinement stage, which is to further select the time points with higher conversion rates every day from several delivery time points (10:00-12:00, 17:00-19:30) and focus on delivery.

Three numbers to see the effect of big data user awakening conversion

For example, we have done a lot of news, food, and tool APP wake-up cases:

1. Eliminate zombie users: Based on big data anti-cheating technology, zombie (fake) users are eliminated from APP users, leaving only real users. About 25% of APP’s iOS users are zombie users and can be effectively eliminated.

2. User click-through rate : Through the "one thousand faces one thousand faces" promotional media/promotional copy, the user click-through rate is as high as 31%. Before promotion, we first determine the user's active media through multiple data mining. Secondly, based on the user's active media, we accurately capture silent users and reach them through user active media DSP and other forms to complete the awakening conversion.

3. Awakening conversion rate: The awakening user conversion rate is as high as 14%, which greatly reduces customer costs. The wake-up conversion rate is mainly based on accurate user portraits based on big data: in-depth mining of search data, competitor data, and mobile content browsing data to create a portrait for each user, including gender, occupation, interests, etc., to ensure that the promotion plan for each category of users is targeted.

The concept of the "second half" of the Internet was proposed by Meituan CEO Wang Xing at the 2016 World Internet Conference in Wuzhen . Some people say that it is really difficult in the second half when all the dividends are lost. It is difficult to attract new users and even more difficult to keep existing users. But on the other hand, big data technology ( artificial intelligence , machine learning) is becoming more and more mature, and it has proposed solutions for refined user operations . Some leading or mature companies are actively carrying out big data user awakening activities, and maintaining the established territory is the most important issue in the second half. If you don't take action today, those 80% of dormant users may become someone else's tomorrow.