Let me give you an example of the pitfalls I encountered when making event products (data review)

Let’s review the activity process first:

1. Total user data of the activity (2017.1.1—2017.9.7)

Converted into a funnel chart:

If anyone has a better tool or method for making funnel charts, please leave a message below to tell me~~ This is made with an Excel bar chart, which looks rough and can only give a more intuitive feeling. . .

During the entire event, there were approximately 156,000 users, with approximately 300,000 users participating. There are actually 115,000 users who won the prize, of which 92,000 are newly registered users. The ratio of new to old users is about 4:1 . The total cost is around 170,000 yuan , which means that the cost of attracting a new user is less than 2 yuan (about 1.8 yuan) . From this perspective, the activity is relatively successful, attracting as many new users as possible at a lower cost.

The winning rate is 73.8% . Since the winning rate of our event is 100%, the loss mainly comes from two parts: page loss and string code errors. In future activities, you can add a point to record the number of people who click on the “receive button”, so that you can see the page loss rate and string code error rate.

2. Data comparison during the event

The data shows that there is not much difference in user participation during weekends and during the week, but there is a big difference between weekdays and holidays, and there is a big difference between peak and off-peak seasons. Among various festivals, May Day > Dragon Boat Festival > Chinese New Year > Qingming Festival.

Note: The data analysis here does not include a difference test, that is, analyzing the impact of other interference factors ( advertising placement , publicity efforts, number of cans placed, etc.) on the activity data. If these factors can be determined, SPSS data analysis tools can be used to determine whether the differences between factors are significant within a 99% or 95% confidence interval.

3. Comparison of registered users

1. Users registered during the event: Users registered during the non-event period≈1.61 : 1
2. The proportion of registered users during the event to the total registered users is ≈64%

These two sets of data show that this activity has played a certain role in the registration of new users, but as with the data in the previous part, it is also necessary to exclude the impact of other interfering factors on the results. In future articles, I will introduce in detail how to eliminate other interfering factors through statistical principles and obtain the significance test you want.

4. Subsequent conversion behavior of registered users

In addition to looking at the number of new users attracted to an activity, another very important indicator is whether these new users have follow-up behavior in our app, website, etc., that is, converting some dead fans who only registered through mobile phone numbers into our real active users.

Among the more than 92,000 registered users, a total of 95 people made purchases, resulting in a total of 201 orders with a total order amount of 19,670.7 yuan and an average order price of 97 yuan per order. in:

• 91 people placed orders directly through the webapp (no login required), resulting in 193 orders with a total order amount of RMB 17,542.70 and an average order price of RMB 91 per order.
• There were 6 people who logged in for the second time, 4 people placed orders, and 8 orders were generated. The total order amount was 2,128 yuan , and the average order price per order was 266 yuan .

It can be seen that registered users do have certain purchasing behaviors afterwards, but most of them come from the webapp (direct ordering without logging in), and the number of active users is very low. In the future, you can use marketing methods such as SMS, WeChat or app push to enhance user stickiness and increase the conversion rate from registered users to active users.