When it comes to data analysis, we may think of a lot of big and difficult data analysis methods, such as SQL queries, Spark code, building databases, etc., but growth hackers are not programmers, maybe only with Excel.
In fact, it is enough to use Excel. The method of data analysis does not require sophisticated algorithms and charts, the focus is to see the problems behind the data.
Problem resolution
Data analysis is one of the essential skills of growth hackers, and more and more Internet operations job recruitment also requires candidates to have data analysis capabilities and experience. The concept of growth hacking is to guide product iteration, marketing and execution strategies of operational activities through real feedback of data, and use data to gain insight into performance growth opportunities.
The basis for mastering growth hacking thinking lies in mastering the ideas of data analysis, data analysis models and tools, but how to find problems and opportunities from data changes is the key skill of growth hacker thinking.
What are the data analysis methods for growth hacking?
1. Summarize the trend through the data curve.
Line charts, scatter charts, and histograms are the basic data charts necessary for data analysis, and these charts feedback the changes in the amount of data in the stage through the basic curve. From the observation of the basic curve, growth hackers are good at summarizing the problems reflected in the shape of different curves and restoring the data performance to the scene where the data occurs.
For example, the user growth curve is the most common data analysis chart. When you plot a chart of the user growth data of the product for the last 1 year or more, you can see different growth characteristics, as shown in the figure.
Growth hackers summarized different characteristics such as hilly curves, roller coaster curves and cheese slice curves from the user growth curve, reflecting actual operational defects, such as the cheese slice curve reflecting the rapid growth of products in the early stage, but the lack of customer acquisition channels in the later stage. Regarding the characteristics of the user growth curve, it is specified in the 005 question.
2. Disassemble indicator insight problems.
From a single curve, the formation process of historical data can be restored. If you want to find problems in the data, you must take a single indicator as the core and disassemble the relevant factors that affect the indicator.
For example, in the analysis of data indicators related to new user growth, the following 3 steps can be disassembled.
The first step is to disassemble the new user data of each source channel and the subsequent retention and activity data of the new users of the channel.
The second step is to disassemble the data indicators that affect the conversion of channels, including the click rate, promotion cost, and landing page conversion of each channel.
The third step is to disassemble the user's behavior data in the landing page. For example, the loss data from entering the landing page to registration, receiving coupons, etc., find the links with high loss, and analyze the problem.
In daily data analysis, growth hackers should understand the differences between different app versions, different mobile phone models, different operating systems, different screen sizes, users in different regions and different access sources, and be very familiar with conventional data indicators.
3. Build user portraits through user behavior.
User portraits of growth hackers are divided into two categories: individual user portraits and group user portraits. The individual user profile records the user's full life cycle data. Generally, from the download, activation, registration login, access to the page, browsing frequency, use time and other behavioral data, as well as gender, age, region, browser brand, system version, display screen height and other attribute characteristics to form a user portrait.
User access behavior tends to be multi-device, multi-state, and multi-terminal. When collecting data, we collect all the behavior data of the same user on different devices (mobile phones/computers), different states (logins/non-logins), and different terminals (multiple mobile phone logins) to form a complete record of user behavior.
The group user portrait is based on the user group mechanism established based on certain common behavior characteristics. For example, user grouping is established in the dimension of users whose access time is longer than 60 minutes in 30 days. From the other behaviors of this group of users, insights are provided on personalized attribute characteristics such as geographical distribution, model distribution, and usage time distribution, as well as core behavioral data characteristics such as shopping frequency and customer unit price, and the refined operation activities are launched for group users to increase the number of users in a targeted manner.
4. Use funnel analysis to gain insight into the causes of loss.
Each product provides a "master path" to the user.
The main path of e-commerce products is "Home - Search - Search Results Page - Product Detail Page - Add to Cart - Place Order - Confirm Order Information - Payment - Confirm Receipt".
The main path of information products is "Home - Content List Page - Article Page - Comments / Forwards / Favorites - Related Readings - Back to List Page - Back to Home Page".
The ability of funnel analysis is to discover the churn of users in various links under a set path. Find the node where the user churns through funnel analysis, and then return to the relevant page to find the cause.
For example, in news products, users churn a lot in the process of converting from the content list page to the article page, then it is possible that the user is not interested in the content of the current list. Further, by comparing the labels of the content that the user reads daily with the labels of the list contents, it is confirmed that the speculation is correct.
5. Multi-dimensional user hierarchical analysis strategy.
The refined operation should meet the common needs of group users based on the characteristics of group users. Growth hacking needs to find the specific needs of the user group through multi-dimensional user screening in user groups.
If the growth goal is to increase product sales, then you can plan activities for two types of user groups, one is the user group with high purchase frequency, and the other is the user group with high purchase amount.
In user management, we can use RFM (see question 36 for details) for user grouping. By filtering users based on the three criteria of recent purchase time, consumption amount, and consumption frequency, the following different user groups can be filtered.
(1) Users who have made a purchase in the last 30 days and have not made a purchase in the last 7 days.
(2) Users who have ordered more than 3 times every 30 days.
(3) Users whose consumption amount is greater than 1,000 yuan and less than 2,000 yuan in the last 30 days.
Based on these 3 criteria, we screened out users who had a high consumption frequency in the last 30 days and had not consumed in the last 7 days. After establishing the group, the next step is to analyze the distribution of daily shopping categories, the distribution of centralized consumption intervals, and the distribution of use time of this group of users, and through the relevant characteristic data, insight into the characteristics of users to design targeted activities.
6. Discover the behavior of users by restoring the user's use path.
In terms of insight into user behavior, growth hackers also discover the behavior patterns of users through behavior path analysis. Behavioral path analysis refers to observing the user's subsequent behavior path after specifying a node.
For example, if we want to analyze what users of online education websites do after visiting the homepage, we can see some possible data through behavioral path statistics: 74% of users enter the search course page, 23.9% of users go to see the course details, 2% of users start to register, and 0.1% of users log in.
These data feedback is that the user is starting from the search for the course, and the user's initiative in the registration process is low, so it is necessary to improve the click-through rate of registration by adjusting the registration entrance or optimizing the registration guidance mechanism.
Through the analysis of the user behavior path, you can also observe whether the user's active behavior is consistent with the "main path" of the product design, if the frequency of the user's behavior outside the main path is relatively high, it means that the main path design does not conform to the user's behavior habits.
7. Predict churn through retention analytics.
User retention is a key factor in predicting user lifetime value (Ltv). By analyzing the user's 7-day, 14-day, and 30-day retention rate, and then using the formula, growth hackers can predict the churn rate of users. Further, growth hackers can improve overall user retention by observing the portrait of each retained user and discovering the user's behavior habits.
8. Use click heat map analysis to do product optimization.
Click heat map analysis is an effective means to optimize product interaction design. By recording the number of clicks, dwell duration, browsing completion and other data of users in different positions of the page, a click heat map is formed, which can observe the location and related information of the user staying for a long time, so as to optimize the position and size of important buttons and guide users to click.
summary
Growth hackers often need a mature and feasible data analysis system to do a good job in data collection, cleaning and calculation formulas, etc., to improve the efficiency of analysis. The real purpose of data analysis is to find out the motivation and demand characteristics of user behavior behind the data through different forms of analysis methods.