We are experimenting a new Demand Watch Algorithm, and gave it a try in order to predict Moncler revenues. We analyzed all of our datasets combined and confronted it ex-post with Moncler (MONC:IM) Q1 2018 interim management statements and the results turned out pretty promising.
Since we capture several indicators of how consumers and business actors interact with the brand, we define brand demand as “the propensity of consumers to buy a product of a specific brand”.
This propensity can be measured by how often consumers find brand’s products in stores they love, how often they ask for it, talk about it and visit the brand website, how often they see an ad from that particular brand.
When demand drops, store owners try to boost brands appeal by reducing price with promotions. If this doesn’t work, they’ll eventually reduce the orders to the manufacturer. When demand rises, the opposite happens, and a brand flourishes.
Translating demand into revenue
When demand is rising, it’s the turn of the brand to play their cards correctly. A well-managed distribution network can leverage customers flowing in. In an ideal case, a brand can divert sales from going to the multi brand stores and attract them into their own retail stores, where margin is significantly higher.
To do that, all disturbing elements must be reduced to the minimum: There should be no incentive for the consumer to buy from a multi brand store instead of the mono brand store. Price and promotions play a huge role in this.
When a brand can shift revenues from the wholesale (multi brand ) to the retail (mono brand), revenues can grow more than demand grows, since for the same product sold, the margin is significantly higher.
Alternative Data Steps In
This is where our coverage of alternative data steps in. In a digitalized world where all interactions can be observed, our effort has been to collect all of these information about demand and distribution. Today we are beginning to model a way to forecast this demand, and consequently, have better predictions on revenues.
Obviously the volume of all of this is way too complex to be handled manually, so we turned to artificial intelligence algorithms. The problem had to be broken down into micro components and each component has a dedicated technology to help us solve the big picture.
Moncler Q1 2018 demand forecast
We modeled billions datapoints of our ecommerce and consumer datasets to project the demand rise in the Q1 2018 versus Q1 2017. We are bringing here the final results, in future posts we will enter deeper into how it works and what are its limits.
In the graph we separated the general “demand for Moncler” (dark blue in the graph) from the capacity of Moncler to divert sales from wholesale (grey) to retail (light yellow). The aggregate demand was a weighted average of those above.
You can totally see the “retail pull” as a factor that multiplies attraction to the retail stores” vs the “wholesale pull” weaker. This means it doesn’t benefit the whole potential of the rise in demand of the luxury jackets.
The results were really promising: Our model scoredexactly between the actual Moncler results at current FX and constant FX (which adds a nice input for us to insert a currency factor in our model).
The model scored almost perfect on retail revenues (+26%) and only 2% higher than the consolidated revenue of the brand.
All of this without any use of Moncler internal data, relying 100% on independently collected alternative information and anticipating the publication of Moncler’s results.
The Demand Watch Algorithm is currently in development phase, some of its components are still being assembled today and tested on a multitude of Fashion & Luxury brands.
We wanted to give you a sneak peek of what’s broiling in our kitchen and collect your feedback, which continues to flow in week after week on all of our topics.
Thank you for reading this and interacting, it is great enrichment for us.