Home Blog Alternative Data for ...

Alternative Data for the reputational risk of retail companies

Manuela Bazzarelli

Today, the reputation of companies and brands is an extremely valuable asset, perhaps the most important one. Considering that users are increasingly aware and have total access to information, you should remember that the risks associated with reputation do not only result from direct actions of your company.

A company can be extremely virtuous for what it does but still expose itself to reputational risks. For example, in the retail industry this occurs frequently for all those brands that resell their products in third-party stores and networks.

We are aware that relying on resellers, business partners and distributors is certainly an excellent driver for sales, and above all a market necessity. However, it is crucial to protect your company and minimize the risk of the third parties, through dedicated  mathematical models and machine learning algorithms that monitor, analyze and catalog news, reviews and comments.


Using alternative data to assess reputational risk: what does it mean?

Using Alternative Data in data science and AI-based algorithms means using unconventional data sources to provide more accurate analyses and models.

Alternative data are complementary to first-party data (already owned by companies)  and they can include information collected from the web, such as mainstream news and local reviews, or even data from social media and geolocation, data from apps and much more.

Alternative data is a fundamental resource for completing and complementing traditional information sets, in case they are limited or incomplete. If you are trying to detect reputational anomalies in retail stores in the area, which could negatively impact the sale of your branded products, it is essential to consider alternative data such as customer opinions.

Our 3rdEye solution is aimed at limiting the financial risk of the distribution counterparties by anticipating any anomalies, supporting Marketing Managers and/or Risk Managers to better distribute their products through third parties, offline and online, in particular in the field of consumer goods.


The Samsung Case: intelligent monitoring and sentiment analysis for consumer electronics

Samsung Italy, the national division of the  world’s leading tech brand, commissioned Aramix to create an ambitious AI-based project for risk assessment related to its distributors.

A proprietary framework of Natural Language Processing for the analysis of alternative data has allowed Samsung Italy to process high volumes of information relating to the various shops, including online news, social media and customer reviews, to eventually identify risk signals on a daily basis, classifying the news and calculating a “criticality score” for each one. Thanks to a daily alerting system and a data visualization dashboard, it is possible to calculate the risks of each retailer – physical or online. 

In addition to the amount of data to be processed, this project had two main challenges: the NLP approach and issues related to privacy.

Firstly, deciphering the natural language of users that implies colloquial expressions, typos and inconsistencies: the algorithm was trained to even understand grammatical and semantic errors, as well as logical discrepancies between evaluations (for example, the number of stars of a review) and the comments associated.

Secondly, the correct data management for transparency and privacy is crucial today. In this sense, Aramix is a company licensed pursuant to art. 134 of the Consolidated Text of Italian Public Safety Laws, which protects not only the intelligent monitoring provider for the data collection and analysis service, but also the client. This way, we made sure Samsung works with data governance and project management that are fully compliant with the most stringent provisions of the relevant law.