Reading an opinion about a brand and how it compares to others in a Web forum is information. Analyzing hundreds of opinions and comparisons of products and brands is becoming an extremely precious business intelligence tool.

It is being offered by Proxem*, founded three years ago by an engineer from ENSIIE (Ecole Nationale Supérieure d’Informatique pour l’Industrie et l’Entreprise), who was so fascinated by the computational handling of language that he made it a subject of research for his master’s in computational linguistics. After 12 years of IT consulting, François-Régis Chaumartin brought in partners such as the Université Paris 7 and the INRIA (Institut National de Recherche en Informatique et Automatique), to set up a global method of operation and notably design a semantics analysis tool usable in French and English. Whom and what was it to analyze? Proxem is aiming at the market for analyzing consumer redemption and business intelligence to enable companies to be attentive to their customers and assess their image compared to that of their competitors. Of course, it is always possible for a research firm to scan the Internet from time to time and pick out useful information for marketing or communications purposes, but no one is fooled by the approximate nature of the findings. Times have changed, as have the stakes, and the Internet is a fabulous flacon that can contain poison. “Before, when a customer was unhappy with a product or service, he talked about it with his wife and colleagues and influenced about 10 people. With the Web 2.0, the same customer can share his opinion with thousands of people, instantly and throughout the world,” explains Mr. Chaumartin. So it is essential today to use semantics analyses of consumers’ verbatim reports and to enable major accounts to know consumers’ degree of satisfaction and detect weaknesses, specifically about the quality of their products. The raw material is quite simply the snail mail and email sent to customer relations departments, as well as blogs and Web forums, with a cross-referencing of sources to refine what is said. Proxem analyzes these thousands of documents to detect future trends and needs. “Our tool does very large-scale marketing studies, with the aid of semantics experts in the analysis phase. This is our added value.” What’s more, Proxem has refined its approach, enabling it to better understand the verbatim reports when they concern products and brands but also when levels of satisfaction and emotion are involved. “We are ‘the academic world champions,’ in analyzing feelings,” adds Mr. Chaumartin, and in proof of this, Proxem was the most precise in this field in 2007 when research was internationally assessed. It’s that the difficulty lies here, especially in a world as sensitive as the beauty world. But it also lies in the ambiguity of words and names, like when a brand has a polysemous name, such as “Alice,” both a first name and a set of services; or “orange,” which is a color, a fruit, and a corporate name; or “total,” a noun, an adjective, and a brand.

So it is preferable, from the outset, to choose a differentiating name for your brand or product line so as to fully benefit from this type of analysis throughout the world, without being subjected to too much loss of information or to too much “noise.” The way it works is very simple.

An “SaaS” solution, or “Software as a Service,” is offered: Ubiq, remote software made available through monthly subscription. Proxem has thus been chosen by a leading brand in supermarket distribution to analyze redemptions in the field to adjust its policy of supply, pricing and quality control, and by a major French daily newspaper to know what its readers are deeply concerned about. Being attentive to customers through their written words to modify one’s image – this is also multimedia.

Sabine Chabbert - Beyond Beauty MAG#30