8 minutes to help you quickly understand data analysis (with book recommendations)
|
With data analysis being so popular in the past two years, if a product/ operation provider doesn’t understand data analysis, how can they make their products stand out and achieve greater user growth, especially when the supply and demand market is tending to be balanced or even over-saturated? How to carry out an operational activity well? … I have been learning some data analysis methods recently. Combining some of my own experience, I have sorted out the following data analysis article. It takes about 8-10 minutes to read this article. 1. What is Data Analysis? Data analysis refers to the use of appropriate statistical methods to analyze collected data in order to maximize the value of the data. Data, like money, does not have much value in itself. It is precisely because of the existence of analytical methods that we can draw certain conclusions and discover problems, thereby tapping into its huge value. 2. Why do we need to do data analysis? Often, VCs need to look at data to make investment decisions; company products/operations need to be iterated upon… Data needs may come from many aspects. Generally speaking, there are four main reasons for data analysis: 1. What is the real trajectory of user usage that drives product iteration? Why do they do this, and is there a simpler process to help us make better decisions? You can also analyze past product data to gain insight into problems and drive targeted product iterations. 2. In-depth demand analysis: Use data to support the needs of users at several levels of Maslow; use data to prove interaction needs; and verify the rationality of company-level needs through data. 3. What is the effect of the new function that drives operational decision-making products after it is launched? Changes in user activity and user retention rate of new features? Which one is better, Plan A or Plan B? For questions like these, the most reliable way to judge whether a problem is good or bad is probably data. Emotional definitions often lead to a lot of unnecessary disputes. 4. To drive operational decisions and evaluate business opportunities, necessary demand research and market research are particularly critical. Is a new market worth entering? Is a new project worth investing in? Is it possible to acquire a company, etc.? 3. How to do data analysis viri.png 1. Data Collection Principle 1: Full volume rather than sampling Collect data from multiple sources, including comprehensive collection of front-end, back-end and business databases. The front end includes web pages and APP clients, and the back end collection is used to supplement the data that cannot be collected by the front end behavioral events. Principle 2: Multi-dimensional segmentation A comprehensive refinement of 5W1H is achieved for customer behavior events, and the behavioral trajectory of who, when, where, why, what, and how is done is fully recorded and refined. Who (who) can be subdivided by registration account, gender, age, personal growth stage, etc.; time (when) can be subdivided by start time, end time, etc.; where (where) can be subdivided by IP, location information, operator, OS, model, IMEI, network access method (2G/3G/4G/WIFI), etc.; reason (why) can be subdivided by hobbies, demand level, etc.; thing (what) can be subdivided by subject, steps, etc. By combining behavioral events with dimensions, we can get the required indicators, such as the region where the user placed the order... There are currently three main methods of data collection (buried points): The first method: Use a third-party statistical analysis standard SDK to access the application. The second method: Use a non-embedded method. The third method: Develop it yourself, refine operations and product decisions. 2. Data Collection The following three factors should be considered when building a data indicator model: a. Integrate behavioral data with business data; b. Regress key data indicators c. Consider data feasibility from multiple dimensions The first key indicator method finds the first key important indicator, and then derives the sub-indicators, such as e-commerce sales If you want to increase sales, you can either increase the number of buyers or increase the average order value. Sales = Number of buyers x Average order value Sales = Traffic x Conversion rate x Average order value When you reach the product details page, this can also be derived as: Sales volume = business details UV x order rate x payment rate x customer unit price Sales volume = activity display x activity conversion rate x order rate x payment rate x customer unit price Indicators at each stage focus on: Stages ● MVP stage (verification): Verify feasibility and target user retention rate , conduct qualitative analysis. There is not much data in this stage itself (except for data-based products). ● Growth stage (crossing the chasm): Most companies fail to reach this stage. Data analysis in the growth stage needs to be properly matched with corresponding data product analysts, or use more in-depth data tools to make corresponding decisions. Can be divided into two levels 1. In the retention stage, the main indicator is the retention rate (next-day retention, 7-day retention, 15-day retention, 30-day retention rate, etc.); 2. In the referral stage, the viral coefficient and viral cycle are mainly considered: how many new users one user can bring on average. So when the viral coefficient is greater than 1, the information will continue to spread, and the total number of people spreading the information will diverge. On the contrary, when the viral coefficient is less than 1, the total number of people spreading is convergent, as well as the NPS (Net Promoter Score). ● Platform stage (activation and conversion): There are dedicated data analysts and engineers, and the team conducts more in-depth data analysis. The main focus is on the activity and conversion rate of platform users, so as to pass the plateau period and usher in the next growth period. ● Monetization period: revenue cost, user activation and recall, LTV, CAC, channel analysis and other indicators 3. Data Analysis Analysis methods: Effective data analysis methods can deeply explore the value of data. Common data analysis methods and models include user segmentation, A/B testing, multi-dimensional event analysis, funnel analysis, A AR RR analysis, etc. Here we mainly use funnel analysis, AARRR analysis model, A/B testing, and multi-dimensional event analysis as examples: ● Funnel analysis Analyze the changing trend of the number of users from potential users to final converted users in order to find the best optimization space. This method is widely used in the analysis of various key processes in product operations . What is a user conversion funnel? It refers to how your business converts a user step by step. for example: Activities: Activity display -> Click details -> Transform to date girls: Chat up -> Date -> Hold hands -> ... Each stage you go through can be broken down into several sub-stages. At every stage, there will be user loss and user retention. Accurately recording data at every link in the funnel in order to analyze and optimize the conversion rate of each link is the infrastructure of data-driven operations. For example, the funnel model of an e-commerce activity page should be like this: Order rate The conversion rate from browsing the activity page to the details page is 50%, the order rate on the details page is 10%, and the final conversion rate from order to payment is 40%. With such a funnel, we can analyze what each link represents and how to improve it: Activity page -> Details page UV: Whether the content on the page is eye-catching and whether the products are liked by users. Products with poor click effects need to be replaced in time according to the page click situation. Details page UV—>Number of orders: whether the details page is attractive, whether it affects the page loading speed, and whether the products need to be reordered. Number of people placing orders -> Number of people making payments: whether the payment guidance is poor, whether the payment tool is faulty, whether it is lower than the industry average. In addition, within the same system, conversion rates also need to be compared, such as this month versus last month, this week versus last week, to see if they have increased or decreased, so that more accurate conclusions can be drawn and problems discovered. ● AARRR model AARRR (Acquisition, Activation, Retention, Revenue, Refer) was created by David McClure, a venture capitalist in Silicon Valley in 2008, and stands for acquisition, activation, retention, revenue, and referral. AARRR In the following example, which one is better, channel A or channel B? For example, indicators at each stage of game AARRR 1.A (How do users find us?) DNU (daily number of newly registered and logged-in users), promotion channel monitoring (cost, traffic) 2.A (Do users have a great first experience?) DAU (number of users who log in to the game every day), average daily usage time, and item association analysis model 3.R (Do users come back?) Retention rate (next day retention, 7-day retention, 21-day retention, 30-day retention), churn rate, churn warning analysis model 4. R (How do you make money?) PR (payment rate), ARPU (average revenue per user), ARPPU (average revenue per paying user), LTV (lifetime value) 5.R (Do users tell others?) K-factor, NPS, etc. ● A/B testing A/B testing is to use data support to select the final solution through different channels and different groups of people. A/B testing requires certain data support to establish a framework with high accuracy and efficiency, such as targeting different channels, user group release, grayscale release, etc. to come up with a suitable solution, which will not be expanded here. AB test ● Multidimensional event analysis Multi-dimensional event analysis breaks down data from multiple angles to discover the specific reasons for data changes. The combination of behavioral events and dimensions can generate data indicators, such as in e-commerce applications: Behavioral events (1H): Searching for products, clicking on product details, submitting orders, paying orders, after-sales service, etc. are all a series of events Dimensions (5W): Who (who) can be broken down by name, gender, and age; When (when) can be broken down by length of stay, order events, payment events, and arrival time; Where (where) can be broken down by IP, city, operator, OS, model, IMEI, and network access method (2G/3G/4G/WIFI); Why (why) can be broken down by hobbies, demand levels, and so on; What (what) can be broken down by subject, steps, and so on. Combining the two can produce multi-dimensional indicators, such as the region where the user placed the order, the channel from which it came, how much the order amount was paid in the past period of time, etc. 4. Data is not omnipotent Although data is indispensable, it is not omnipotent. For example, it is difficult to get drive in the direction of product innovation, and it is difficult to get sufficient data to judge long-term user feedback. So what methods can truly drive rapid growth in the number of users of a product? verify – Create truly meaningful products A large proportion of products on the market are meaningless, especially in this oversaturated market. To achieve rapid user growth, we should return to the core of the product, create something truly valuable, and support it with data-driven development. Such a combination may lead to greater growth. – Building brand value It is especially important to focus on shaping the brand concept for core users. In traditional industries, many companies pay more attention to branding than those in the Internet industry, but on the Internet, it is often not taken seriously, which has a lot to do with its own popularity. Perhaps we can learn more from the slogan and brand building cases of the fitness app Keep . – Utilize growth hacking techniques In the book "Growth Hacker", many cases of growth hacking technology growth are discussed, such as how to achieve user growth at low cost and so on. 5. Finally, I recommend some books on data analysis • "Head First Statistics": Statistics in Simple Terms • "Lean Data Analysis" • "The Magic of Data - Data Analysis Based on Open Source Tools" • "Data Mining - Applications in Marketing , Sales and Customer Relationship Management" • "R Language Practice" Mobile application product promotion service: APP promotion service Qinggua Media information flow The author of this article @冯海平is compiled and published by (APP Top Promotion). Please indicate the author information and source when reprinting! |
<<: How to do holiday marketing most effectively?
>>: How much does it cost to create a children’s clothing mini program in Zhangjiakou?
Recommend
Why do we put so much effort into shooting short videos, but people who watch them don’t feel anything?
People who make short videos are always eager to ...
Xiaohongshu KOL promotion: the secret of popular notes!
As self-media and internet celebrities became mor...
How to use the Tik Tok mini program? How to activate the Douyin mini program?
How to use the Tik Tok mini program? How to activ...
Changsha’s tea tasting studio is built in. Where is the most reliable tea drinking place?
Changsha Tea Tasting Studio has its own arrangeme...
Ad landing page conversion is poor? 3 reasons!
The effectiveness of the landing page is related ...
Facebook/YouTube Video Advertising Guide
In recent years, the growth of digital video adve...
Jiuzhaigou earthquake, brands must stop issuing blessing posters!
Yesterday (August 8) at 9:19 p.m., a magnitude 7....
Si Nian Tarot Beginner + Advanced Course Video Amy Tarot System Course Video Total 33 Episodes
Si Nian Tarot Beginner + Advanced Course Video Am...
3 steps to analyze user churn
What is user churn ? This can be judged by the TA...
The latest news on Beijing’s epidemic lockdown in 2022: When will things return to normal? Attached is the unblocking standard!
Recently, four clustered epidemics have occurred ...
Tips for creating landing pages for advertising!
This article will share with you the techniques f...
What does Baidu's bidding display volume mean? What is the relationship between the display volume in Baidu promotion and the online name search volume?
(1). Impression volume refers to the number of ti...
New trends in marketing observation in 2020
In the current environment, no matter whether the...
A guide to filling out the application form for the 2020 College Entrance Examination. How to fill out the application form for the 2020 College Entrance Examination?
As the old saying goes, "30% of the test and...
Taking the 98 Shopping Festival as an example, how to use product thinking to promote sales
The Purchase Festival on September 8, 2016 was th...