Actually, although this function was released more than half a year ago, it is a function that can only be viewed via API (or via Query Explorer) instead of the normal Google Analytics screen to view reports.
Furthermore, even if you know the existence of this function, it is difficult to find information by googling. I would like to explain such a “related product” function.
What is the related product function?
The related product function is a function that can only be used with accounts that have e-commerce enabled. It is a function that can make recommendations.
It seems that reports for using this function easily have not yet been prepared, but it will be possible to try using the Google Analytics Core Reporting API or Google Analytics Query Explorer . If you want to run it easily, we recommend Google Analytics Query Explorer . Even if it is necessary to display related products in the navigation of the EC site, etc., by using the Google Analytics Core Reporting API , it is possible to always acquire the latest related products (implementation is , It takes time to call the API, and unless the site has a lot of transactions, it is unlikely that the related products will change frequently, so store data in advance in daily or weekly batches. recommended).
When incorporating a recommendation engine into an EC site, etc.,
- Study the statistical algorithms used to calculate recommendations and implement them yourself
- Make a contract with an external ASP service and let us use it (by paying the monthly ASP usage fee)
However, with the implementation of the related product function of Google Analytics, it can be said that it has become easier to incorporate a recommendation engine than before (although the accuracy of recommendations is Compared to the case of self-implementation or the case of using an external ASP service, it will be inferior at this time).
Dimensions and metrics used in the Related Products feature
The purpose of the related product function is to create a report that people who bought product A also bought product B.
Newly added dimensions and metrics will be used to express the above.
dimension
The dimensions can be broadly divided into the product ID, product name, and product type dimensions for product A and product B , and the statistical model ID (correlationModelId) used to calculate the similarity. At the moment, there seems to be only one type of model “defaults” used for similarity calculation, but it seems that it will increase depending on the expansion of Google Analytics in the future.
- ga: queryProductId
- ga:queryProductName
- ga:queryProductVariation
- ga:relatedProductId
- ga:relatedProductName
- ga:relatedProductVariation
- ga:correlationModelId
metrics
Metrics include a correlation score that indicates the degree of similarity between product A and product B, and an item that indicates the purchase quantity of product A and product B. The correlation score itself is difficult to define, but the closer this value is to 1, the more accurate the result.
- ga:correlationScore
- ga:queryProductQuantity
- ga:relatedProductQuantity
Let’s check the actual data
To see the actual data,
- Google Analytics Collection API
- Google Analytics Query Explorer
You must use either Here, we will introduce how to check the data using the easy- to-use Google Analytics Query Explorer .
Visit the Query Explorer page
Query Explorer is a service that facilitates access to the Google Analytics API. Query Explorer can be accessed from the URL https://ga-dev-tools.appspot.com/explorer/ .
After accessing, authentication to Google is required. After accessing Query Explorer, click the orange band at the top of the screen that says “Click here to authorize” to authenticate. For authentication, log in with the account that uses Google Analytics and approve the authentication.
Specifying and retrieving search conditions
After authentication, fill in the form on the Query Explorer page. Specify per account, property, view, dimension, metric, start date, end date, and maximum number of data. The specified contents are
| item | set value |
|---|---|
| Account | A Google Analytics account for the site that has e-commerce set up |
| Property | Among the properties associated with the account specified above, properties that have e-commerce settings |
| View | Among the views linked to the properties specified above, the view you want to see the actual data (requires e-commerce settings) |
| dimensions | For our Related Products report , select an item in the Related Products at the bottom of the combo box . I think you should select ga:correlationModelId, ga:queryProductId, ga:relatedProductId, ga:queryProductName, ga:relatedProductName |
| metrics | For our Related Products report , select an item in the Related Products at the bottom of the combo box . I think it is good to select ga:correlationScore, ga:queryProductQuantity, ga:relatedProductQuantity |
| start-date | Specify the start date of the period for viewing data |
| end-date | Specify the end date of the period for viewing data |
| max-results (maximum number of data) | In the case of related products, the amount of data is expected to be large, so set a larger number than the default number. You can specify up to 10,000 |
In addition, specify segment divisions, filters, and sort as necessary.
After specifying, click the “Get Data” button to call the Google Analytics API behind the scenes, and after a while, the data that meets the specified conditions will be displayed below. The data can be viewed not only on the screen, but also downloaded in TSV format. You may be unfamiliar with the TSV format, but it is a format that can be opened in Excel as it is, so let’s open it in Excel.
How to read the data
The important columns of retrieved data are ga:queryProductId and ga:relatedProductId and ga:correlationScore . These three pieces of information represent the degree of relationship between product A and product B. Columns such as ga:queryProductName contain information associated with the product (if the product name is easier to understand than the product ID, the product name can be used).
As mentioned above, ga:correlationScore indicates the degree of relationship between products. Unrelated products (pairs of products that are rarely purchased together).
Are general blog sites and news sites compatible with e-commerce?
Is there anyone who thinks “I don’t operate an e-commerce site, so related products are irrelevant”? Even on a blog site unrelated to e-commerce, each article detail page is regarded as a product, and when the page is read (scrolling is XX% or more, etc.), a transaction occurs for that product, which is regarded as e-commerce. You may be able to implement Google Analytics. By implementing a blog site as an e-commerce site, you can apply the powerful analysis reports of Google Analytics for e-commerce to blog sites and news sites, which may lead to the use of related product functions. .
