Data operations: How to analyze data more efficiently and effectively?

Recently, when people consult Zhuge Jun about business, they don’t say they want to know about data analysis products, but instead say “we want to know about your company’s point tracking solution.” Later, after a conversation, I found out that they wanted to do data analysis, and also wanted to know how the data was collected. Data collection is the premise and foundation of data analysis. Today, Zhuge Jun will talk about data collection and data analysis. Data collection

Three key points of data collection

1. Comprehensiveness

The amount of data is sufficient to have analytical value, and the data surface is sufficient to support the analysis needs.

For example, for the behavior of "viewing product details", it is necessary to collect the environmental information, session, and user ID behind the user trigger. Finally, it is necessary to count the number of people, number of times, average number of times per person, and active ratio who trigger this behavior in a certain period of time.

2. Multidimensionality

What is more important is that data can meet the analysis needs. Flexibly and quickly customize multiple attributes and different types of data to meet different analysis goals.

For example, for the action of "view product details", we can know what product the user is viewing, as well as multiple attributes such as price, type, product ID, etc. through tracking. This will help you understand which products users have viewed, what types of products are viewed most, and how many times a particular product has been viewed. It’s not just about knowing that the user has entered the product details page.

3. Efficiency

Efficiency includes the efficiency of technical execution, the efficiency of collaboration among team members, and the efficiency of achieving data analysis needs and goals.

Based on the above three points, let’s see how to make data collection more accurate, analysis more useful, and the team more efficient.

No embedding or code embedding?

Is it better to collect data without embedding or with code embedding? If it is a very mature product and operation, this is not a difficult choice, but for someone who is new to data analysis products, it is really a difficult problem. In fact, how to do the tracking needs to be based on the product and analysis, it will be better if you combine them together!

Here is a brief introduction:

The main process of the business uses code tracking because it can obtain more detailed data;

For some activities or unimportant behaviors, it is recommended to use visual tracking to save the workload of R&D personnel;

For the promotional landing page, it is recommended to use full embedding points, and you can use browsing heatmaps and click heatmaps to optimize the landing page;

For business data stored in the business database, it is recommended to use server-side tracking.

The value and efficiency of data analysis

Step 1: Clarify data-driven goals

Data collection should avoid being too large and comprehensive. Data analysis needs also evolve with the continuous iteration of products. It is important to clarify the analysis needs for the long term and the current stage to make the analysis more purposeful and the technical execution more efficient.

Example scenario:

Xiao Ge is the company's product manager, and Xiao Zhu is a technical person. Recently, both of them have realized the importance of data in product operations and decision-making. After researching several data platforms, they finally chose Zhuge io and have clarified the data needs at the current stage...

Xiao Ge : "Are you busy, Xiao Zhu? The data indicators in the document, such as login process, registration conversion, purchase conversion, and sharing conversion, are the data indicators that need to be paid attention to in the long run. Be sure to include them. As for the discovery function, we will submit a new version in two weeks. I won't include it for now. Thank you for your hard work."

Xiao Zhu : "Xiao Ge, you are awesome. I will bury you in a moment!"

Xiao Ge : "Oh, and there is a recommender option on the registration page, which requires users to enter the recommender's account. Don't collect the account when collecting data. I just want to see whether the registered users have recommenders, and process that attribute into a judgment."

Komoro : "That's easy. Then tonight..."

Seeing Xiao Ge turning around to leave, Xiao Zhu hesitated and continued typing code silently...

Step 2: Collect data on demand

Collecting data with requirements and analysis goals in mind not only avoids the confusion caused by data redundancy, but also avoids the embarrassment of not knowing what to analyze after collecting all the data.

The figure shows an example of a buried point:

Graphic documents can be organized by data analysis requirements personnel, and table organization allows requirements personnel and technical personnel to collaborate more efficiently, which also greatly improves the subsequent analysis value and efficiency.

Step 3: Multi-dimensional cross-positioning problem

Applications to data can be divided into general analysis and exploratory analysis. General analysis includes monitoring and analysis of daily data such as new additions, activity, retention, and core funnels, as well as data monitoring of daily operations of each department. Monitor daily growth and analyze abnormal situations, such as monitoring and timely optimization of registration failures and payment failures.

Exploratory analysis is an advanced application of data. Conduct correlation analysis on core events and discover key points for product improvement, such as correlation analysis to promote user purchases and finding actions to promote retention.

Step 4: Optimize products and operation strategies

Based on the problems reflected by the data, we can carry out real-time monitoring and timely solutions. Based on the growth inspiration obtained from the analysis, we can perform A/B testing, grayscale testing, and MVP practice.

Step 5: Measure

Measurement is the last step from data analysis to practice, but it can also be the first step. Sometimes we seem to have found a growth point, but experiments show that the facts are not as expected. Don’t be discouraged, don’t be disheartened, and don’t skip meals. The understanding of users and in-depth exploration of the business during the analysis process may lead to cumulative value for the next optimization.

Data analysis thinking

Data collection is important, and the methodology of data analysis is also important, but don’t be superstitious about data, because what may be more important is human creativity and imagination! Data analysis is never a one-time thing. Products are constantly iterating and businesses are constantly updating. From cognition to decision-making, data plays a more auxiliary role. From sorting out needs, to collection, to analysis, to practice, and then to measurement, it is always circulating in the entire process of enterprise growth.

Finally, those programmers who change the world always hope to create more value with their technology. In many cases, what they want may be clear data requirements, clear analysis goals, and a set of efficient and collaborative methods. After all, everyone believes that: being able to accurately solve problems and drive business growth is more important! Heavy! want!

Source: Zhuge io Data Coach