The Proof of Concept. Unlocking hidden value in visual data.
2025
What’s the difference between your visual contents and the visual contents created by influencers and consumers who talk about you?
This case study demonstrates how our semiotic AI can be used to analyze massive visual corpora and to give strategic advice to a company that manages
a luxury brand.
METHODOLOGY
Methodology: the Semiotician’s glance at scale
From Big Data to Strategic Insight
Traditional research is limited by scope. We processed over 35,000 images and 16,600 posts on Instagram related to two competing luxury brands, a massive big data volume that our technology helped us analyze qualitatively and quantitatively.
The role of AI: tags, graphs, clustering, quantifications and mapping
Trained with specific semiotic categories, our AI was able to recognize the contents of the images, telling which were the most meaningful from many points of view: figurative (objects, persons, places, etc.), plastic (colours, shapes, spatial disposition), enunciational (style, perspective, etc.), and narrative (roles, kinds of actions, values at stake, themes and topics).
Doing so, the AI was able to construct graphs that showed the most statistically recurring features of such images and the most statistically meaningful relationships between such features.
It was also able to differentiate them in clusters and to quantify the weight of each cluster in the corpus.
Finally, it automatically built semantic mappings to position each image depending on their communicative goals.
STRATEGIC FINDINGS
The brands vs influencers and consumers
The true value of Sense-Maiking emerges in the gap. We precisely identified the differences between the codes used by the brand to communicate on Instagram and the codes used by influencers and consumers to talk about them and their products.
Key Strategic Insight: the Data revealed
Practicality vs Symbolism
While the luxury brands aimed to frame their products as high value status symbols, the data showed that consumer and influencer self-produced content often focussed on a practical function, vouching for the product’s quality and for the experience provided by its material components (3200 self-produced images talking about the products of Brand 1 said so)
Self-centered vs Other-centered people
The graphs, which displayed the most recurrent contents in the images produced by brands, influencers, and consumers—as well as the statistically most significant relationships among those contents—clearly showed that while one of those brands sought to construct the image of a self-centred, individualistic male consumer performing to display his masculinity, the influencers and ordinary people preferred to portray themselves with the brand’s products in social contexts, where they could build rewarding relationships with others.
Fashionable Classic Women vs Femmes Fatales
By clustering the 35,000 images in our corpus, we immediately identified a few anomalous groups, in which influencers and ordinary people associated the brands with portrayals of women very different from those constructed by the brands themselves in their communication campaigns. What stood out most was a strong presence of femmes fatales—sexy and sensual—but markedly distant from the elegant, classy women that the brands, despite their differences, sought to represent. We were thus able to flag these anomalies, which, especially when the images were produced by influencers, could be corrected.
ACTIONABLE ADVICE & CONCLUSION
Turning Insight into Market Mastery
Sense-Maiking final recommendations are not assumptions: they are derived from a robust quali-quantitative method, merging human expertise with the machine’s objective precision.
The goal of this study was to compare the thousands of images produced and posted by two competing luxury brands on their respective Instagram channels during a specific period, and to analyze the visual content of their communication campaigns in relation to the tens of thousands of images created by influencers and ordinary users to talk about the same brands and certain types of their products over that same timeframe. In this way, we aimed to test the extent of control these brands were able to exercise over their image, as well as their ability to capture the codes and mindsets of their consumers.