Boost products by recently created

[superdav42]

I would like to add add a boost so recently created products would rank before older products when they would otherwise have relevancy similar scores.

There is the "Product creation date (X days ago)" optimizer rule I could apply but it would take many rules like:
Product creation date (X days ago) less than 30 boost 10
Product creation date (X days ago) less than 90 boost 20
Product creation date (X days ago) less than 180 boost 30
Product creation date (X days ago) less than 360 boost 40
this would get us close to to what I want but it wouldn't quite be sorting them by recency.

Also it look like I would need to create this rule list for each store of which we have many and that would mean over 100 optimizers.

It seems best if there is way to use the "based on attribute value" model with a dynamic attribute showing the freshness or newness and a low boost. Is it possible to make such a dynamic attribute that would be related to how new the product is?

[rbayet]

Hello @superdav42,

I thought I had given an answer but looks like that's not the case.
The way I would do that would be to

• create a decimal attribute "creation date penalty" or "product age penalty"
• create a new Datasource that will work with a LUT (look-up table) initialized by the constructor for instance with 366 slots
  • whose keys would be dates (YYYY-mm-dd) from today to today-365 days
  • whose values would be 1, 1/2, 1/3, 1/4, ..., 1/365

When the datasource loops on products to index

• if the product creation date is earlier than today (just in case) the attribute value is 1
• if the product creation date exists in the LUT, it gets the value : 1 if the product was created today, 1/2 if it was yesterday, etc
• if the product creation date does not exist in the LUT, assign it a dramatic low value like 1/9999

Then your "based on attribute value" boost will actually act as linear penalty by default, unless you change the boost impact (which applies a function like square root or logarithm).
You can also decide to built a LUT more akin to your desire with "blocks" of same penalty value

• value of 1 for all dates within the last 10 days
• value of 1/2 for all dates between 11 and 30 days
• value of 1/3 for all dates between 31 and ...

Regards,