Saturday, October 14, 2006

Recommendation engine for Shops !

I am just back from buying some Sarees for my mom and some shirts for my dad..
Its diwali..

I had quite a trouble finding what my mom would actually like..
Firstly.. I dint have any domain knowledge [in sarees :)]
I dint know what type of saree my mom would like [Kanjeevaram,Bangalore Silk, Dharmavaram Silk, Mysore Printed, Crepe silk and some other ones]..
Wouldnt it be a good idea for the shops to record which type of saree is bought by what type of person?
Then we could mine the data to form associations [age, height, complexion, income group, type of saree, colour of the saree, the design on the saree, the time of the year the saree was bought...]
Then you just give the shopkeeper the profile and bingo.. You get the list of sarees that your mom [or girlfriend for that matter] would like..
This would be particularly useful when the subject is not present at the time of buying the saree...
Amazon does similar stuff.. but the recommendation it does it is for different objects..
Netflix does something similar to this..

May be we can have an online mart for sarees..
Where real models display the sarees that you would like [pre-recorded]..
We will have to do an analysis of the saree market in India..

Population on 1st March 2001 was 1027015247
Assuming that one third of them are women who would buy 1 saree each we would have 343 million sarees bought per year..
Even if 1 percent of these women are internet users [our target audience], we have a potential of getting
3.4 lakh customers.. Even if we make a profit of Rs.10 per saree we can earn 3.4 million Rs per annum

Okay Okay.. I know that this is not how it works.. I need to have a bplan, the data for recommendations and improve the above analysis[?] which is to rudimentary.. Maybe this is how I would explain the concept to my kid...

Gotta prepare for CAT.. got to give the Mock tomorrow :)

No comments: