D. Jameson
Published in

D. Jameson

A destination recommendation product for an online travel marketplace

A step by step method which recommends the best possible destinations to boost the confidence of the user

There are two kinds of Holiday seekers. Those who know the destinations clearly and ready to plan their trip, and then there is the other, those who have high motivations to travel but not sure about the destination yet. This recommendation engine is designed for the latter. It explores the basic travel motivations to create a system that is flexible enough to accommodate various facets of traveler needs.

The last thing the holiday seekers want is to get stuck at a place they can’t appreciate, and unable to find the activities of their taste.


To propose solutions that are intuitive and conversations, which will take the travelers(unsure about destination) through a funnel-like journey to have more confidence and faster decision-making towards finalizing a destination and eventually give options for the best packages and deals.


The current solution at TravelTriangle has limitations as it has filters of destination and package mixed in the same bucket- making it unknowingly confusing for the user. It does not help them to decide the best destination or suggest unique rankings that are appropriate to their needs (motivation) articulated through current filters. This makes the holiday search a never-ending loop and a frustrating experience.


To group destinations using a set of parameters that are associated with fundamental ‘inner motivations to travel’ which would help them differentiate one destination type from the other and eliminate the least probable holiday destinations.

The two major factors that make a destination different from the other are Attractions(Nature and Culture) and Activities, the factors like budget and duration do not define a destination from each other as they are more associated with the final holiday package. Design a system that will provide the right filtering strategy at the right time, and mixing the two will only confuse the user.

The basic funnel- framework developed for the recommendation engine.

A two-step solution (user-flow)

The primary CTA filters out ‘Destination not known traveler’ from ‘Destination known’ before reaching the following solution-flow. The following visuals are to understand the system and the logic behind the recommendation feature.

Once the destination is selected, it further shortlists the best packages under the particular destination with sub filters.

The idea is to show the matching percentage of their preferences against each destination/ packages using a systematically tagged data base. Each tag will have a weight/score to represent how much valid/relevant/meaningful a tag is with respect to other destinations. For e.g. A honeymoon tag is more relevant with Kerala as a destination than Gujarat, so weight of this tag for Kerala will be higher than Gujrat. The destination filtering will also have a should have and must have crieterias. It needs to have some must have attractions and activities, and some should have activities, and trip-type. The ordering of destinations is based on primary wight and secondary weight in the backend logic.

The travel themes are basically divided into two parts, ‘Places to visit and Things to do’ — and they all come from clear inner motivations of the travelers, but often not well articulated by amateur travelers.

Like it is true for any e-commerce platform, for any travel marketplace, their online platform will be as good as its inventory recommendation product. This was an attempt to make the Destination recommendation more meaningful for TravelTrinagle.com by humanizing the search experience based on basic travel motivations of a ‘comfort-seeking traveler’.

Read a related story: It explains the thought-process behind the sequencing of the conversational product flow.

Awful to Awesome & All in-betweens in pursuit of better Experiences, Solutions, & Stories…

Recommended from Medium

Do You Struggle With The Quantum Superposition?

K-means Clustering Algorithm and It’s Usecases

Spotify Machine Learning Engineer Interview Case Study

instamentor.com | Spotify Machine Learning Engineer Interview Case Study

Secrets of ThinkOrSwim: how to set up indicators, premarket scanners, watchlists 🔥

Simple, Visual, Geospatial Analysis Towards Pandemic Readiness

Code Review: How I retrieved the test set predictions in the CIFAR10 dataset

How to use Multivariate Adaptive Regressive Splines (MARS) on tabular datasets

Geocoding : Necessary information is sufficient information

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Dhaneesh Jameson

Dhaneesh Jameson

Experience Design Producer

More from Medium

Sprint One: Understanding Physician Burnout

Sprint 2: Officially on the road, fueled by research

Intro to C3 AI design blog series