Recommending personalized touristic paths in Paris through an app


Paris, city of lights, steeped in history, filled with atypical places and incredible monuments, picturesque markets and places of culture on every street corner, attracts nearly 38M visitors each year. Despite this great diversity of possible experiences in the four corners of the capital, tourists mainly concentrate in the few iconic places not to be missed and stay on these marked routes. This concentration of tourists harms the experience of visiting tourists in Paris - who then stay shorter than before in the capital: linearization of the experience, degradation of the most frequented places, endless queues, etc. This concentration also marginalizes the less known nuggets of the capital, while weakening the shop owners who are not located directly on the itineraries of visit. To cope with this finding, the Paris Convention and Visitors Bureau, Paris Habitat and APUR have decided to join the Datacity program, organized by NUMA and in partnership with Mastercard, to find a solution to this problem by tapping in all the richness of available data. MFG Labs is the winner of this 2019 edition of Datacity and therefore helped these partners to design and test a solution: an application that allows tourists to generate and personalize tour itineraries in Paris, to plan a trip that will allow them to get off the beaten track for a memorable experience.


This project being carried out as part of the Datacity challenge, we had to carry out all the steps until the validation of an MVP in less than 4 months: framing the problem, exploring the available data, finding an adequate solution to the project, develop and test it in real conditions.

A multi-disciplinary team was mobilized to lead this project from start to finish.

The team is made up of a consultant, a designer, user research specialists, a data team (business analyst, data scientist and data engineer) and application developers.

Framing via user research and data exploration

The purpose of this challenge was to design an experience that validates a hypothesis: it is possible to improve the experience of visitors in Paris and unclog the places of visit thanks to an application that uses partner data.

A scoping phase was therefore carried out with different objectives: to define a segment of users to be addressed, to better understand the needs of this target, to understand the current tools at their disposal and their limits.

This scoping phase was carried out by mixing qualitative studies with the chosen target (American tourists in Paris, possibly repeaters), and other secondary targets (hotel managers), with studies of several data sets, from open data or provided by partners. In particular, transaction data from American customers, anonymized and aggregated in compliance with GDPR standards, were provided by Mastercard, and made it possible to complete the qualitative studies and orient the questions administrated on the field to American tourists.

At the same time, we have identified the main applications for tourists, in order to study the essential functionalities and question their limits with real tourists. An advanced search on available and usable datasets was also carried out, making it possible to obtain a qualified cartography of usable data.

Design, an alliance between designer, engineer and data scientist to maximize use and utility

The design phase was carried out jointly by the Service Designers and the Data team: a user-centric design along with the presence of data specialists that could validate the technical feasibility of the interface design choices.

In parallel, the technical teams worked on the engineering of the data sets, and the feasibility of the recommendation algorithm, central element of the device.

The complementarity of expertises has made it possible to highlight a strong choice for the structuring of the solution: mini-districts with marked DNA (culture, jewelry, crafts, flea market, museums, etc.) will be identified, and integrated into routes generated by a recommendation algorithm offered to tourists. It was on this strong and differentiating choice that the design and data teams relied to design the end-to-end user experience on the one hand, and the technical design of databases and data recommendation services.

Agile and iterative development


The development was done in an agile way with several streams in parallel in order to have an MVP service for the solution:

  • an Experience stream with the development of a graphic identity, main and secondary screens, and integration within the OTCP website. Storyboards were notably created, making it possible to align all the actors of the project on the experience to develop;
  • a Data stream with the creation of databases and data infrastructure on GCP, the labeling of certain data (public and from partners Mastercard, APUR, OTCP) to have the most complete database on the City of Paris to feed the algorithms developed;
  • an Algo stream to develop the algorithms for creating mini-districts and creating / recommending routes within these districts, ensuring that they meet the desired decongestion objective, and producing it on the application.

Testing in real conditions

In order to validate the experiment, tests were carried out with users in real conditions and the MVP solution developed.

These tests made it possible to correct certain bugs. Above all, they made it possible to collect precious feedbacks to improve the service and to release successive versions.


In just 4 months, this Datacity challenge have resulted in:


  • a clear vision of the competitive ecosystem of travel planning applications and the needs of foreign tourists
  • a unique database of tourist places and activities in Paris, mixing open, private data, and expert labelling
  • a functional application allowing a tourist to create personalized and authentic itineraries according to his desires of discovery
  • valuable user feedback to feed the product roadmap to continue the development of the application
  • a successful collaboration between large and heterogeneous organizations: Mastercard, OTCP, APUR, Ville de Paris, NUMA and MFG Labs.
section image