Shoppers’ annual Spring Festival Gala "Double Eleven" shopping carnival, how many orders did you place?
Although the editor didn't buy too much this year, it can be regarded as a contribution to the "Double Eleven"!
Double Eleven is the annual sales promotion season for e-commerce retail. All e-commerce companies and brands are gearing up to get full firepower. Brands wanting to win the e-commerce battle are indispensable for consumer insights. , The more thoroughly the brand understands consumers’ buying behavior, the more effective it will be to help companies carry out marketing activities in a more targeted manner.
This article analyzes several key factors of e-commerce through data analysis, explains which possible factors affect sales, and helps e-commerce retail find insights to increase sales.
Today, we use the literature Sakar, CO, Polat, SO, Katircioglu, M. et al. A public sample data in Neural Comput & Applic (2018), combined with JMP software to do some interesting exploratory data analysis of online shopping users' purchasing behavior.
Step.1 Information acquisition
Original source
The original information can be obtained through URL Download the original data in csv format and open it through JMP; you can also directly use the web page reading function of JMP to directly obtain the web-side data.
(JMP operation: file -> open from internet -> webpage)
Information background introduction
The data set contains 12,330 online shopping users' purchase behaviors on the site within one year, as well as the corresponding 17 user behavior records and final transaction results. Among them, 17 behavior records include 10 numerical characteristics and 7 sub-type characteristics.
- Basic information
Management webpages, management stay time, information webpages, information stay time, product webpages, product stay time, indicate the number of users opened on different types of webpages and the total stay time.
- Bounce Rate
Bounce rate indicates the percentage of users who enter a website page from a certain path and leave the website without doing anything. It can be used as a metric to measure the entire website, or as a metric to measure the page.
- Exit rate
The exit rate represents the percentage of all visits to this page that leave the website from this page for a particular page, and is generally used as a measure of the page;
- Page value
The page value represents the average value of the web pages that users have visited before completing the transaction;
- Special day
Special day, which means the interval between website access time and specific special day;
- other
In addition, it also includes the operating system, browser, region, traffic type, visitor type used by the user, whether it is a weekend, and the month of the year.
Step.2 Visual exploratory analysis
The bounce rate is related to the success of online marketing
Customers only view a single page and then exit, making it difficult for brands to have a chance to persuade consumers to buy a product. After all, they only browse one page. Let's take a look at the bounce rate below.
(JMP operation: Analysis -> Distribution)
It can be seen from the figure that 90% of customers have a bounce rate of less than 6%, and the average bounce rate of all users is only 2%. Is it perfect? Please don't be anxious and happy. It may be inaccurate, or there may be a technical problem with how the website’s analytics tracking code is integrated into the website.
Because based on past experience, the "normal" bounce rate is between 40%-60%, less than 40% is very rare, and higher than 70% is worrying and requires quick action.
The current bounce rate is beyond the expected range and looks "unbelievably good", which should be caused by repeated analysis code somewhere in the website.
Multiple exploration of users' numerical characteristics
When doing shopping behavior analysis, the user's numerical features may have many dimensions. With the help of JMP's multi-analysis method, the relationship between each dimension can be quickly discovered, and it is possible to achieve dimensionality reduction operations to reduce the risk of subsequent feature monitoring. The necessary resources are wasted.
(JMP operation: analysis -> multiple methods -> multiple)
Based on current data, the number of users opened on different types of web pages is positively correlated with the stay time, which is well understood. The bounce rate and the exit rate are also obviously positively correlated because of the similar calculation formulas. In addition, no obvious numerical characteristics are found to be correlated.
The impact of the way users go online on sales
Through the statistical results of the chi-square test in the figure below, we can capture the relationship between whether the transaction is completed or not and the customer's operating system, browser type, and traffic type.
( JMP Operation: Analysis -> Take X Fit Y )
As far as the operating system is concerned, different operating systems have different proportions of users completing transactions.
It can also be seen from the graph that the operating system is slightly higher when the operating system is "2", while "1" and "3" are lower. This may mean that the website page is not friendly enough to support these operating systems. If you want to improve this part Conversion of revenue, you need to make corresponding improvements. Similarly, for browser types and traffic types, we have also seen their statistical impact on the proportion of users completing transactions, indicating that the website also has room for improvement in this regard.
New and old customers and working days / Impact of weekends on sales
With the help of the chi-square test method in section 2.3, we can also quickly discover some new and old customers and the regularity of sales on weekdays/weekends, but here we try another data sheet summary method.
(JMP operation: analysis -> consumer research -> classification)
Combining the above graphics and information, you can clearly see: Old customers are the main force of website visits , Indicating that the website is doing a good job in customer retention; but we also see that whether it is on weekdays (13.2% vs 26.1%) or on weekends (16.5% vs 21.9%), the percentage of new customers completing transactions is higher For old customers, this shows that the website can make some improvements in the conversion rate of old customers.
For example, old customers can enjoy larger discounts by introducing new customers when buying goods, which not only mobilizes the enthusiasm of old customers to buy, but also adds more new customers to the website.
Centralized display of various analysis results
If you want to centrally display various analysis charts in the form of reports, you can use the script function of JMP to realize real-time update of report link data with one click, saving a lot of repetitive manual operations.
(JMP operation: File -> New -> Application)
The above are the results of some exploratory data analysis. Below, let us further try to use the method of data acquisition to establish a quantitative statistical model for the main characteristics of users and transaction results mentioned above.
Step.3 Data predictive modeling
Through the decision tree, filter the key characteristics that affect sales
Decision tree is a supervised learning method. It can summarize decision rules from a series of data with characteristics and labels, and present these rules in a tree structure. The results are easy to interpret, and it has a wide range of applications in various industries and fields. .
(JMP operation: Analysis -> Predictive Modeling -> Segmentation)
From the figure, we can observe the various stages of the decision tree, from top to bottom showing the most important characteristics that affect the results of the transaction.
The most important of these is the value of the webpage. The percentages of transactions with a webpage value lower than 0.067 and higher than 0.067 are 3.85% and 56.4%, respectively. The gap is obvious. Some more important characteristic parameters are also listed later, such as bounce rate, month, and product-related pages. These information are all important factors driving transaction results. Now they can be quickly selected and displayed, so that brands can have A clearer focus of improvement and optimization.
Choose the best among the best, try more data collection methods
In addition to decision trees, JMP also provides a variety of data acquisition methods such as neural networks, random forests, boosted trees, and support vector machines, and can easily compare model algorithms to achieve the best among the best.
Through the new model screening of JMP Pro 16 to compare multiple data acquisition methods at one time. Before that, in order to prevent the built model from overfitting, you can first follow the 6:2:2 ratio of the training set, test set, and verification set. Split the original data to generate a verification column.
(JMP operation: Analysis -> Predictive Modeling -> Generate Verification Columns)
As shown in the figure below, once input feature parameters and transaction results, the platform will construct multiple algorithm models at the same time, and automatically screen out the current best modeling method as random forest. The performance of the model on the test set is the future The prediction ability of the new data R-square reached 0.6, and the prediction accuracy reached 90.7%.
An accurate predictive model can show that the brand understands the possible transaction results of each user as early as possible, especially predicting the failure of the transaction, and responds and remedies early, such as discounts, such as popping up the retention interface before the customer exits the page.
(JMP operation: Analysis -> Predictive Modeling -> Model Screening)
How about it? After reading today's analysis, does it make you more confident in the "e-commerce war"?
For brands, providing personalized product, service and product recommendations, and making one-to-one marketing for individual needs is a huge advantage of online marketing over traditional marketing.
Combined with JMP data analysis software, you can help companies design product collection pages and promotional activities that better meet the needs of target customer groups through in-depth analysis of online shopping users' consumption behavior, and promptly respond to potential problems discovered. Improve, thereby bringing greater benefits to the enterprise.
>> Download JMP 16 now and enjoy a 30-day free trial <<
