The Blind Bartender Effect and Virtual Profiles of Children

What is the difference between children and adults
Children are very different from adults as consumers. What are the differences?

Children have only one constant characteristic – gender. Everything else: height, foot size (those same physiological parameters) change quickly, it will not be possible to remember the size once and recommend it for the rest of your life. The system should not only remember these parameters, but also accurately predict their change.

One adult can have several children

 

– therefore, the system should recognize this and maintain not one, but two or more virtual profiles.
There are certain categories of children’s goods “for the little ones” (for example, diapers), which are bought with a certain frequency. It would be nice to sell them regularly, too, right? By the way, there is a laos phone number data special size grid for the same diapers: from 0 to 24 months.

mobile phone number library

There are children’s holidays, on the eve of which sales of children’s goods increase. The system should have advertising tools: to remind adults in time that it is time to return to the store.
All these features should be taken into account in the recommendation algorithms. Otherwise, we will get the effect of a typical blind bartender.

How do “children’s” recommendations work
A virtual profile is created for a child, just like for an adult. It is based on the history of purchases and views of children’s products. Gender and age are determined. Age up to 2 years is automatically updated every month, after 2 – every year.

The age in the virtual profile is each time correlated with views and purchases of products for a certain age. A new profile is created in two cases:

If a purchase was made or enough views were collected for a child whose gender is not among the calculated ones.

If a purchase was made or enough views were collected for a size and age less than the minimum calculated or greater than the maximum calculated.

An example of how this

 

affects recommendations: products of the wrong size are excluded from the blocks:

“Interesting” – if the system knows the size of a certain type of clothing, then this blog meeteffective all-in-one on this basis, recommendations of products of the wrong size are hidden.
“Similar” and “popular” – even if the system does not know the sizes, preliminary data can be obtained from the currently viewed products.
Information about the child’s gender also significantly affects the recommendations issued by the system.

An example of how this affects recommendations: products of the opposite gender are excluded from the blocks:

“Popular products” on the main page.
“This is interesting.”
“Buying now.”
“Also buying / also recommend viewing” – are excluded if the viewed bw lists product matches the gender of the child.
This saves the buyer from “garbage” recommendations and increases their overall quality. More precisely, recommendations – more clicks, it’s that simple.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top