I chose the website Gilt which recommends products to buy based on your previous purchase history. ### SCENARIO DESIGN
Perform a Scenario Design analysis (as described in assignment spec). Consider whether it makes sense for your selected recommender system to perform scenario design twice, once for the organization (e.g. Amazon.com) and once for the organization’s customers.
Who are your key users?
Key users are those who have signed up for an account on Gilt.com. The account set-up is free and the user would be alerted about flash sales which run for a prescribed time everyday. So, basically anyone interested in fashion would be a user group.
What are their key goals?
Flash sales offer markdowns on designer women’s apparel, shoes and home goods. The website enables competitive prices for interested buyers.
How can you help them accomplish these goals?
Gilt accomplishes these goals by offering a wide selectcion of products at competitve prices, once-lick ordering, the option to save payments information, free delivery when order size exceeds a threshold and layered access based on past purchase history. For example, users who have spent 10,000 and over would be considered Gilt Noir members and offered first dibs on the flash sales before it is opened to everyone.
Does it makes sense for your selected recommender system to perform scenario design twice, once for the organization (e.g. Amazon.com) and once for the organization’s customers?
No, perhaps not. Designing once for the customer is sufficient.
Attempt to reverse engineer what you can about the site, from the site interface and any available information that you can find on the Internet or elsewhere.
The website still seems to focus on past purchase history. For example, if I clicked my favorite designers as Salvatore Ferragamo, it would show me flash sales that correspond to that designer, it doesn’t show me other similar Italian shoe or bag designers, for example. So, the recommendation algorithm is user-to-user filtering.
Move from user-to-user filtering to item-to-item which will broaden product range and sales. Lower the threshold for individual reward tiers.
In researching Gilt as a company, I referred to a number of sources (cited APA style below):
https://aws.amazon.com/partners/success/rue-gilt-groupe-databricks/
https://venturebeat.com/2015/01/22/it-takes-more-than-great-algorithms-to-make-personalization-work/
https://blog.fastforwardlabs.com/2015/12/09/fashion-goes-deep-data-science-at-lyst.html.