Deconstructing Taste – The Crayon Way

Product updates   |   
Published May 20, 2014   |   

Crayon’s product manager Ajay Kashyap explains how Crayon’s proprietary taste algorithm helps in understanding customer taste in a holistic way; not bounded by traditional constraints of category, availability, costs and location proximity, in conversation with Crayon’s sales specialist Priya George.

Tell us a bit about Crayon’s choice platform? What does it do?

Ajay: Let me illustrate this with an example: Lets say that you are traveling. You face multiple questions: Where will you stay? What food will you try? What to buy? Which movie or local attractions to see? What book to take along for the long haul, and so on.

There are any number of possibilities, and hundreds of opinions from review sites to friends’ recommendations… you are facing the Paradox of More – more choices than ever before and less time to navigate them.

Research suggests that when provided with 4-6 choices, people are more likely to choose, feel more confident in their decisions and are happier with their choice.

Crayon aims at simplifying the world’s choices. Our SimplerChoices platform helps provide you with the most relevant and personalized choices across categories.

Wow, can you delve into that a bit more … how do you come up with the most relevant options?

Ajay: We believe a person’s eventual choice in any given transaction depends on four main attributes or parameters:

1. Taste – What does she like?
2. Context – What is she looking for now?
3. Influence – Who influences her?
4. Behavior – What has she been doing in the past?

Our SimplerChoices platform attempts to develop mathematical models for all these attributes.

For training the models, we have curated six billion taste connections across 15 million products and 31 million users!

Our true differentiator is the way we have used this data to deconstruct choice: Crayon’s proprietary algorithm helps in understanding the customer in a holistic way, not bounded by traditional constraints of category, availability, costs, and locational proximity.

What’s your approach to deconstructing taste, and tell us how it differs from some of the typical methods that are in play today?

Ajay: Again, let me give you an example:

Say a person likes the movie “Pulp Fiction”. This can be de-constructed into multiple pieces:

– She likes the popular movie Pulp Fiction
– She likes a particular piece in the movie which can be seen in a lot of other niche movies as well
– She likes films featuring actor John Travolta or by director Tarantino
– She like ‘Film Noir’ movies
– She like all ‘things’ that cater to such a taste, and the movie Pulp Fiction was an example in movies. Hence, she would also like products in other categories (say books, restaurants, holiday destinations) that cater to such a taste
– Her taste in Pulp Fiction,’ could tell me about other things she might like, even though she might not have had the opportunity to explore them (say cinema of the same genre in other languages)

The diagram below summarizes this:

We have developed an algorithm that de-constructs taste for every single entity across all categories, led by a configurator. This algorithm captures that genuine taste, which would emerge if such constraints as costs, locational proximity, availability of products and information were removed.

What would the person really want … our algorithm zeroes in on that.

Interesting. Using SimplerChoices, where do you see Crayon a few years from now?

Ajay: I dream of SimplerChoices being used by a billion people, intuitively simplifying choices in their day-to-day life. I wish that when they have to make a choice, they say, “Lets Crayon it!”