Post by rahim on Jan 31, 2024 14:47:28 GMT 5.5
user events to the engine with other tag managers or completely without a tag manager. The implementation is often more complex, but in principle there is nothing wrong with it. Display/play out the recommendationsIn order to be able to display product recommendations to users, Google offers an API that returns a list of products that it recommends based on a user ID and/or other data. By the way, this interface can be tested as an example in Google's Cloud Console:However, the query only returns the IDs of the products, which means that it is not possible.
for example, to query the API using JavaScript DB to Data and display recommendations directly in the shop. For this we created a solution that fetches the recommendation from the Recommendations AI and fetches the necessary product information (title, current price, images, etc.) from a mongoDB to provide a query that enables display as a widget:In this case, our “backend” is located entirely in the Google Cloud, while we enable playout via JavaScript, which we control in Tealium. This construct enables us to implement the recommendations completely without technical adjustments to the web shop and to display recommendations to our users. How good is Google's Recommendations AI compared to other solutions? How good are the recommendations? Google's solution goes very far in terms of measurement.
engine is “fed” with page views, shopping cart actions, purchases and even refunds. The events can also be imported from a variety of sources. For example, the purchase history of the last few years can be imported, which means that the engine has a lot of data about the shop's customers right from the start. In addition, each recommendation receives a unique token for identification, through which the recommendation itself can be measured. This means the engine continually learns how good the recommendations are and can optimize them.
for example, to query the API using JavaScript DB to Data and display recommendations directly in the shop. For this we created a solution that fetches the recommendation from the Recommendations AI and fetches the necessary product information (title, current price, images, etc.) from a mongoDB to provide a query that enables display as a widget:In this case, our “backend” is located entirely in the Google Cloud, while we enable playout via JavaScript, which we control in Tealium. This construct enables us to implement the recommendations completely without technical adjustments to the web shop and to display recommendations to our users. How good is Google's Recommendations AI compared to other solutions? How good are the recommendations? Google's solution goes very far in terms of measurement.
engine is “fed” with page views, shopping cart actions, purchases and even refunds. The events can also be imported from a variety of sources. For example, the purchase history of the last few years can be imported, which means that the engine has a lot of data about the shop's customers right from the start. In addition, each recommendation receives a unique token for identification, through which the recommendation itself can be measured. This means the engine continually learns how good the recommendations are and can optimize them.