Adobe Commerce Customer Segmentation. How it doesn’t work.
One well knew Fortune 500 brand had an issue with the Adobe Commerce Performance.
Suddenly website response time jumps to 10 seconds for all pages nad one in hour sites went down.
After a bit of debugging, it became clear that Adobe Commerce's shitty code causing this issue.
This observer :
Magento\CustomerSegment\Observer\ProcessEventObserver
Produces harmful queries that killing the RDS MySQL database.
So we removed this junkie Magento 2 module Customer Segmentation any way. All these fancy Magento features are just a promise and don’t work well.
Adobe Commerce tries to add more fake features to win that sales competition with Salesforce Commerce. Implementation of that modules is terrible. So if you are considering using Adobe Commerce, uses its Magento Comunity fork :
Customer segments allow you to dynamically display content and promotions to specific customers based on properties such as customer address, order history, shopping cart contents, and so on. You can optimize marketing initiatives based on targeted segments with shopping cart price rules. You can also generate reports and export the list of targeted customers. Because customer segment information is constantly refreshed, customers can become associated and disassociated from a segment as they shop in your store.
This feature produces tons of junky queries like this :
SELECT
?,
sales_order.customer_id
FROM
sales_order_item AS item
INNER JOIN sales_order ON item.order_id = sales_order.entity_id
WHERE
(sales_order.store_id IN ([..]))
AND (
item.product_id IN (
SELECT
main.entity_id
FROM
catalog_product_entity AS main
WHERE
(
(
main.entity_id IN (
[..]With rally pure performance and subqueries.
After removing this module, response time improved x1000 times. You can see that in a picture:
The pink color on a graph is SQL query time.
So Segmentation Magento solution doesn’t work it is another fake feature Adobe sales team will sell for you as fully working. Magento 2 works for stores well with 100–200 products because of the low data volume, but if you have a huge 1000+ items catalog, you will have many performance issues.
