1. Will the feed of each day overwrite the earlier day feed? OR the new products in feed will be appended to earlier feed available in Target?
As documented in Feeds , any data uploaded using the .csv file, Google product feed, or Analytics Product classification feed overwrites the existing entity attribute value in our database. So yes the feed of each day will overwrite the products from the previous feed .
2. Will the recommendation only recommend only those products which are available in latest feed in case of daily feed?
Criteria are rules that determine which products to recommend . So depending on when the algorithm runs and when the feed is uploaded there may be some edge cases where products no longer in the feed are still recommended .
3.If we plan to take feed daily and till the next day, before uploading the next day feed. If some of the products/content page got deleted by content/product team, then I believe if recommendation will recommend these content pages/product than 404 error will come
How we can ensure that 404 error page not come for those pages which got deleted by product team and recommendation algorithm is recommending??
In the event that a product is removed from your site; there is no way for Target to know that the product has been removed when delivering recommendations.
In order to not have the product appear in the recommended items, you could try and do the following(Please note that these are just some options that you could try)
- When a product is removed from the site, simultaneously run an API call to remove it from the system: https://marketing.adobe.com/resources/help/en_US/rec/r_Deleting_an_Item_From_the_System.html or you could also remove it manually
This way, when the cache is updated after a few hours- the product in question will not be shown in any recommendations
- Alternatively, you could modify the criteria to check for another parameter such as "outOfStock" which would be set to true if a product is out of stock and not display products that are out of stock (note: this should be done before the product is removed from the site, because without actual visits to the product page, the mbox data collected is not updated in the backend).