Shulou(Shulou.com)06/03 Report--

When choosing user behavior analysis tools, most enterprises do not know how to choose the user behavior analysis tools that are suitable for their own business. The author used Baidu statistics APP to analyze the website of my own company, which is in the e-commerce industry. recently, the company proposed to refine the operation and use data to drive business growth, so in October, I examined the user behavior analysis products of several domestic companies that have done well: Geek (Ali), Shenze data (Baidu) and GrowingIO (LinkedIn).

In the process of type selection, I sorted out the comparative documents of the functions and services of each company, and made an in-depth comparison of the three major user behavior analysis platforms: geek, Shenze and GrowingIO from six main aspects: team background and product positioning, data access mode, quantitative analysis function, qualitative analysis function, secondary development and data application, service project, etc., hoping to be helpful to enterprises with user behavior analysis needs when choosing analysis platform.

I. team background and product positioning

Geek team: from Alibaba Group Taobao (CEO, CTO), Aliyun (Chief architect), CEO is a product, operation and marketing background, has co-founded and raised nearly 10 million US dollars. CTO and architect are senior technical experts in Ali big data.

Product positioning: intelligent analysis platform for user behavior

According to the official website of digital geek, digital geek is the leading third generation Internet data analysis platform, which provides full solution products based on AARRR user life cycle management model. More than ten data analysis methods such as multi-dimensional segmentation, simultaneous group analysis, funnel analysis and comparative analysis are used to provide Internet operators with analytical data such as customer acquisition, activity, retention, transformation and user behavior, and provide comprehensive and open data API. Support the real-time application of Internet platforms in all industries in marketing, operation, Amax B testing and other links, and improve the conversion rate of marketing, products and operations through accurate data analysis, so as to make enterprises operate more scientifically and intelligently.

Shenze team: Baidu Big data Department log analysis (CEO, CTO, etc. 4 people), 4 people are technical background, CEO is Baidu big data technical manager.

Shenze Positioning: flexible customized Multidimensional data Analysis products

Shenze Analysis (Sensors Analytics) is a user behavior analysis product launched by Shenze data, which provides privatized deployment and SaaS deployment, realizes basic data collection and modeling, and can be used as a PaaS platform to support secondary development. Shenze Analysis is mainly aimed at typical business scenarios such as marketing channel effect evaluation, refined operation improvement, product function and user experience optimization, and assisting management decision-making.

GrowingIO team: is a star startup team. CEO is the director of LinkedIN's business analysis department and has rich experience in the implementation of enterprise data analysis systems.

GrowingIO positioning: a new point of non-buried point user behavior data analysis product

GrowingIO is a new generation of data analysis product based on user behavior, which can collect full and real-time user behavior data without burying points, and the data analysis is more refined. It helps managers, product managers, market operations, data analysts, growth * *, etc., to improve the conversion rate, optimize website / APP, and achieve rapid user growth and cash out.

Team comprehensive strength score: GrowingIO > several geeks > Shenze data

2. Data processing (access to data and export)

The access of user behavior data is generally in the way of burying points, which refers to the process of transmitting the behavior indicators and business indicators that need to be collected through SDK to the analysis platform for analysis after adding statistical monitoring code to the website and APP.

Generally speaking, different data acquisition methods are used according to different business scenarios and data acquisition requirements. At present, the mainstream embedding methods include code embedding, visual embedding, full embedding and backend embedding. Let's take a look at the comparison in data access:

Back-end data access: the front-end embedded code mainly collects behavior data, while the back-end embedded code mainly collects business data, which can be implemented according to different scenarios. For example, what you want to see is behavior data, which often does not exist in the data table. It can only be achieved through front-end embedded code.

Import historical data: both geek and Shenze data are supported. Importing historical data can activate past data and tap potential value. However, Shenze supports one-time import of historical data. In addition to user behavior analysis, digital geeks also have business intelligence BI products, which can not only be imported at one time through BI, but also be updated automatically when they are updated. GrowingIO does not support importing historical data.

Ad UTM support: the UTM parameter is a parameter that the tail of the URL does not affect the URL jump used to mark traffic. You can add UTM parameters to the URL to count the retention and transformation of different promotion media on our products; there are five UTM parameters that can be added to the web page: utm_source,utm_campaign, utm_medium,utm_content,utm_term each parameter needs to match a value, this function is supported by all three products.

Source channel management: through channel management, channels can be tracked. Geeks and GrowingIO both provide direct access and channel identification of common search engines. Such channels do not need to be added manually. Shenshi does not support this feature for the time being.

IOS channel statistics: after integration and initialization in iOS APP by calling trackinstallation, the channel attributes in the device can be written to the attributes by calling trackinstallation when APP is started. Currently, this function is supported by geek and Shenze data.

Mixed development APP access: for mixed development APP data accounting, geek and Shenze data support access, there is a more mature framework, GrowingIO is temporarily in the testing stage.

No burial point collection: the full burial point actually collects all the front-end user behavior data without technical support. Product operator data analysis only needs visual extraction on demand to carry out multi-dimensional cross-analysis, which is supported by both geek and GrowingIO, but not supported by Shenzhi for the time being.

Visual embedding point: on the page with SDK added, students of products, operations, and marketing directly click WYSIWYG to collect the indicators they want to analyze. Visual embedding points do not need technical support, and it is an efficient way to collect relatively simple indicators. Geek and GrowingIO solutions are more mature, and you can also try them out.

Code burying point: it means to record the business indicators generated by the user triggered behavior and the user triggered behavior when the collected code is located in the code location of the user triggered behavior. For example, when the user triggers the order submission behavior through the code, the number of orders submitted, the amount of the order submitted, the order number, payment method and other indicators.

Burying point management: for products with complex data indicators, there are more indicators for different types of burying points, so it is very important to manage buried sites efficiently and avoid leakage and misburying. Geek GrowingIO burial site management supports search, editing, deletion and other management operations.

Event import timeliness: the user's behavior is abstracted into an event. When the user triggers the event, the user will send data to the platform. When the network speed is normal, Shenze and several geeks are within 1 minute, and the GrowingIO is 1 hour. In the era of big data analysis, the real-time performance of data analysis is a more important aspect when choosing an analysis platform.

Export data file: in the aspect of data export, geek supports the most formats (CSV,Excel,TXT,SQL,JSON,XML) to meet the needs of all kinds of data export analysis. Shenze and GrowingIO are currently only exported in a simple format.

Data processing score: several geeks > Shenze data > GrowingIO

III. Quantitative analysis function

As of January 2018, Shenze and Digital Geek GrowingIO can support four functions: index operation, user flow analysis, user flow analysis, basic funnel analysis, user attribute analysis, user behavior sequence. Let's take a closer look at it through the following comparison table:

Advanced subdivision funnel: after creating a custom funnel, you can horizontally deconstruct the funnel according to user attributes, session attributes, custom attributes and user groups to find the core factors that affect the funnel conversion so as to improve the conversion rate.

Four operations of indicators: geeks can fully activate the indicators of default events, custom events and visual buried events through four operations, and can create absolute and relative indicators suitable for more analysis scenarios according to business needs.

Event analysis: in user behavior analysis, behavior is generally abstracted as events. Shenze and geek support code burial points can promote marketing, operation and product behavior indicators and business indicators through flexible multi-dimensional event cross-analysis.

Subdivision funnel comparison: in the business model, the funnel can disassemble the business process into finer processes through horizontal deconstruction to find the transformed funnel in the process. The subdivision funnel is actually a horizontal structure. By dismantling the funnel through the user attribute dimension, for example, you can see the transformation of each city, so as to find the funnel's funnel in the city to which it belongs. The subdivision funnel function is supported by all three, but Shenze and GrowingIO only support pairwise comparison of dimension values, and several geeks have no limit on dimension values in this area.

Advanced funnel analysis: the advanced funnel function can see the details of funnel inflow and loss at each step, which is only supported by a few geeks for the time being.

User flow analysis: user flow analysis is also called traditional path analysis, which can be used to analyze and optimize the product experience process by analyzing the overall flow of users after entering the product. This function is supported by all three.

Intelligent path analysis: the intelligent path is used to analyze the key paths in all the paths that the user reaches the target path. By creating a funnel for the critical path, the analysis method of optimizing the critical path is analyzed. Shenshi does not support it for the time being, but the other two support it.

Form analysis: forms can be seen everywhere in Internet products, but in order to optimize the form filling experience, it is necessary to analyze the user's filling behavior at the field level. This function is a feature of the geek platform, which can analyze the form experience from the form conversion rate, form abandon rate, filling time, refilling rate and other indicators, so as to help us optimize the form experience and improve the form conversion rate.

User composition analysis: several geeks and GrowingIO both support user composition analysis, but Shenze data does not support this function for the time being.

Active analysis: the retention of several geeks can create the number of daily active users, weekly active users and monthly active users for all events, and can also be split according to different dimensions.

Retention analysis: the retention of geeks can create daily retention, weekly retention and monthly retention of all event behaviors, and can also be split according to different dimensions.

Magic numbers: magic numbers are key behavior indicators that affect product indicators through data analysis. For example, Facebook found that retention increases significantly the next day when users follow more than 5 users. This feature is an exclusive feature of GrowingIO.

Event distribution analysis: after the user behavior is abstracted into events, through the event distribution analysis, we can see that the number of behavior triggers is analyzed according to different time granularity, and no product optimization decision is provided to support. This function is supported by both geek and Shenze, but not by GrowingIO.

Page and page group analysis: for PC and app, all pages can be analyzed by page, page group, and specified page. The page groups of geeks can basically meet the needs of page group analysis in all scenarios, and support the creation of page groups based on a variety of rules, such as header matching, mismatch, inclusion, regular expression, and so on.

Error analysis: when users visit our products, if there is a request error and the request times out, the error analysis of geek can analyze the error page, error details, error time and other indicators, help technical colleagues to grasp the product performance experience access problems at the first time, and help to improve the product conversion rate in terms of details. This function is also a feature of several geeks.

User clustering: the third generation user clustering through user attributes, time attributes, behavior attributes fine positioning of user clustering, through fine positioning of user clustering, can carry out fine operation, precision marketing and other user operation strategies.

Profile of grouped users: after we filter out the users through the clustering tool

User attribute analysis: after adding the SDK for data collection, the analysis platform can collect user attributes, session attributes, channel attributes, promotion attributes and so on by default. The three platforms all support user-defined upload of user attributes. All business indicators and behavior indicators can be subdivided in multi-dimensions according to user attributes. Is the basis of all subdivision functions.

User behavior sequence: the user behavior analysis platform can view all user behavior sequences and understand the user's

Churn analysis: predict user churn through user behavior model and build user churn model, which can only be supported by magic strategy for the time being.

Behavior prediction: predict the trend of user behavior based on user behavior historical data, which can only be supported by Shenze data for the time being.

Cross-screen analysis: user id can get through the analysis on multiple devices, and this function can only be realized by several geeks.

Intelligent path analysis: intelligent transformation path analysis, according to the selected target event, automatically analyze the path combination of the target event, the good use of intelligent path can simplify the conversion funnel setting and improve the analysis efficiency.

Form analysis: from the time the user enters the form page, there is a micro-funnel, from the total number of people entering to the number of people who finally completed and successfully submitted the form, in this process, how many people begin to fill in the form, when filling in the form, what difficulties encountered in filling out the form can not complete the form, all affect the final conversion effect, the number of geeks can analyze and optimize the above indicators from the field level. Thus increasing the conversion rate of the form.

User cluster profile analysis: geek can perform multi-dimensional portrait analysis on the created user groups, and enterprises can conduct multi-dimensional profile analysis on specified user groups to provide data analysis support for operational strategy optimization.

Event distribution analysis: you can analyze the distribution of all the behaviors triggered by users and understand the analysis of user behavior preferences.

Custom active retention analysis: the custom active retention of several geeks can be customized to create a variety of user retention and activity according to events abstracted from user behavior.

Quantitative analysis tool score: several geeks > Shenze data > GrowingIO

IV. Function of qualitative analysis

As an unavailable part of user behavior analysis system, qualitative analysis tool can visualize user behavior and display user behavior data clearly and intuitively. Let's take a look at the functional comparison of the three platforms in the qualitative analysis module.

Heat map analysis: heat map analysis is a visual expression of the user's page visit depth, click, stay, and slide in the form of a thermal map.

Is a kind of qualitative user behavior analysis product. Common heat maps are link click heat map, browse heat map, attention heat map, page click heat map, split screen heat map.

Link click map: the link click map can visually show the user's click interaction and the click volume and click proportion of each page element, and at the same time conduct heat map subdivision analysis according to different access devices, different times and user dimensions. from then on, you can easily control all user click interaction analysis.

Browsing heat map: when users browse the website, the correlation between mouse movement and eye movement is 84% to 88%. Browsing heat map is a heat map formed by collecting the user's mouse sliding track. The more places you slide, the higher the color heat.

Attention heat map: through the attention heat map, we can grasp the hot areas where the user stays for a long time, and the attention heat map can help us understand the thinking and stay time of the user when browsing, for example, when the user stays on the submit order button on the shopping cart page for a long time, the product manager can combine the order analysis function to analyze the reasons for the user's stay and solve the user's doubts, so as to improve the purchase conversion rate.

Click on the heat map: the heat map is drawn according to the number of clicks by the user, and the areas that are clicked more will be brighter, and vice versa.

Split-screen heat map: visually show the number and proportion of browsing users on each screen. Therefore, the display of product function and page content can be optimized according to the split-screen heat map to improve the exposure of important information.

Video playback: video playback is a qualitative analysis tool that reproduces every click, slide, input and other operations of the user's entire session in the form of video playback, so that product operators can clearly understand and measure the interaction of users on app, web and H5, so that enterprises can gain an in-depth understanding of users and optimize product experience.

Summary: in terms of qualitative analysis function, the heat map of several geeks should be the one with the largest number of heat maps on the market, GrowingIO.

Support connection click map and page click heat map.

Video playback function is a highlight of geek behavior analysis system, although it can clearly and intuitively trace the behavior of users, but video can not be aggregated for the time being, how to solve this problem, video playback will be more effective.

Generally speaking, the number of geeks in qualitative analysis > GrowingIO > Divine Strategy

5. Secondary development and data application

Different enterprises have different requirements for user behavior analysis. For example, most enterprises will consider whether to support secondary development after the introduction of third-party analysis systems, and whether they can meet some customized requirements on the basis of the original system.

For the functional comparison between secondary development and extension:

Index early warning: index automatic alarm is a function of automatic monitoring and alarm of abnormal indicators. Through index alarm, not only the registration volume, registration conversion rate, bounce rate and other behavior indicators can be monitored at any time, but also indicators such as transaction volume and transaction amount can also be monitored in real time. This feature is not currently supported by GrowingIO.

Automated operation: automated operation is a tool to automate operation based on collected behavior indicators and set up operation trigger mechanism. Currently, only a few geeks support this feature.

Push push service: this feature is a tool that can be operated by customizing push on the app based on user behavior. This feature is currently supported by all three.

Privatization deployment: privatization deployment is a service model in which the entire system is deployed locally to the customer. Geek and Shenze data support the way privatization is deployed. GrowingIO is not supported at this time.

Secondary development: both geek and Shenze data can support secondary development on the original system.

Authority management: the authority can be subdivided according to different business modules and different functional departments of the data platform.

Geek supports custom configuration of permissions according to functions, applications, data and other levels. You can choose to allow some users to view only certain features and some applications, including only certain metrics.

GrowingIO supports subdivision of permissions by function and data, as well as grouping of users. Provides a highly customized way of permissions, convenient for users to manage all the data within the user project, as well as users. While ensuring a high degree of data security, it also helps users and their organizations to achieve better cooperation.

Magic strategy: the user management provided by Shenze can be divided into three different types, namely, administrators, analysts and ordinary accounts, while supporting the authorization of data and functions for roles.

In terms of rights management: several geeks > GrowingIO > Shenze data

ABG B test:

Athumb B test is a unique function of three behavior analysis platforms, several geeks. A Big B test is a method of scientifically optimizing products through data analysis, making two or more plans for the same optimization goal (such as two different styles of shared button), randomly selecting two parts of users, letting some users use scheme A, and another part of users using scheme B. count and compare the click-through rate, conversion rate, active retention and other indicators of different schemes to find the best product decision-making scheme.

Pain points of the product:

"a posteriori" product verification, if it does not meet expectations, the rollback leads to high development costs and high risk of customer loss.

Most product managers rely on intuition to make decisions, but the reality is that what we think is not necessarily what users think.

It is better to listen to what the user says than to see what the user does. Through the rigorous analysis of user behavior data, it is a reliable product-driven method; no matter how good PM is, it can't run half of the Amob B test.

In the traditional development process, the online scheduling is needed, the efficiency of development iteration is low, and the AB test does not need to be released, so the scheme can be verified quickly.

Using scientific data analysis to guide our product decision is the core of product growth.

Data application score: several geeks > Shenze data > GrowingIO

VI. Service items

In terms of service items, they all provide product description documents, frequently asked questions (FAQ), etc.

Online customer service:

Shenze data: Shenze data service process: scheme consultation, requirements carding, event design, data access verification, product use

Geek: if you have any questions, you can consult the official Wechat customer service at any time. The reply speed is usually within 3 minutes. In addition to the normal carding demand, demand carding, geek provides paying users with growth index system, data analysis on-site training, one-to-one service, data index diagnosis service, growth solution.

GrowingIO: provide on-site product training, online data analysis, growth * college.

Service item score: geek > Shenze data > GrowingIO

Summary

Through the comparison of the above products, services, secondary development and other aspects, the following conclusions can be drawn:

The geek user behavior analysis system mainly analyzes the behavior data of the whole user life cycle based on the AARRR model from three aspects: marketing, operation and product. The data granularity is finer (the data can be subdivided according to any dimension), and the analysis model and methodology are more scientific and perfect. It not only has data analysis function, but also has application data optimization tools such as automatic operation and Aamp B test. Have mature products and successful operation experience GrowingIO's non-buried point technology makes data collection more efficient, while there are relatively new concepts such as magic numbers. The threshold for use is relatively high in these three companies, and the real-time and subdivision of data is slightly weaker than that of geeks and geeks. The biggest advantage is that the content of data analysis is rich, and the content quality of official account is very high. Shenze data is a kind of data-driven product with partial technical style, with relatively few data analysis models, but rich data import tools, focusing on user behavior analysis tools with data warehouse as the core. it's a good choice for companies that want to build a data warehouse.

As our company pays more attention to the application of data in product operation, we do not want to invest too much manpower in technical data access, and hope to produce practical results through data analysis. After careful comparison and selection, we finally adopted a number of geeks. Currently in use for two months, several geeks provide one-to-one analyst services, using improving the order conversion rate as a pilot. Hand-in-hand to help us to improve the conversion rate of nearly 1 stroke 3 boss is also more satisfied with the results (millions of new levels of sales revenue every day).

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.