Turning Old Stock into Gold: How Roberto Cavalli Boosts +35% Revenue and +50% Volume with Differs
Turning Old Stock into Gold: How Roberto Cavalli Boosts +35% Revenue and +50% Volume with Differs
Success Story
Turning Old Stock into Gold: How Roberto Cavalli Boosts +35% Revenue and +50% Volume with Differs
Turning Old Stock into Gold: How Roberto Cavalli Boosts +35% Revenue and +50% Volume with Differs



What Roberto Cavalli achieved with Differs - in just 8 weeks:


Cr. Vogue, Roberto Cavalli, Fall 2020, Ready-to-wear
The Brand 🤎
Roberto Cavalli, founded in Florence in 1970, is an iconic Italian fashion house known for its bold designs, exotic prints, and sensual style. Celebrated for innovations like printed leather and stretch denim, the brand blends opulence with Italian flair and artisanal craftsmanship.
The Pain 💥: In a luxury house like Roberto Cavalli, there’s no room for unsold merch. Most discounts followed simple, stock-based rules—leading to an inefficient destocking strategy.
Even the most iconic brands such as Roberto Cavalli need to leverage price changes to get rid of unsold merch and increase revenues. But instead of having an optimal pricing strategy to destock while preserving brand image and profit, discounts were selected based merely on simple rules. It was a mixed approach between following competitors, aligning with marketplaces, and assessing the store's readiness to manage price lists. Discounts on the latest collection were usually a fixed percentage (40–50%) applied across the entire range, with no differentiation between lines, colors, regions, etc. For older seasons, the discount percentage increased with the age of the collection. This is not effective for the brand or for customers.
Risk? Brand equity dilution, revenue loss, unpredictable sell-through rate (STR).
The Cure 🧪: Moving towards an advanced predictive algorithm with Differs, where humans interact with the models to decide the right discount for each item.
Differs helps centralize historical data and builds a sophisticated, custom predictive model that truly understands our customers' behavior: a combination of willingness to buy, long-term stock forecasts, and external effects like weather, seasonality & events or competitor pricing. The model gives rise to an insightful and intuitive tool that explains the motivation behind each discount level chosen. This allows the team to determine the best price for each item—clearing stock with maximum profit—along with continuous support from Differs.
Differs' optimal discounts are applied across the globe, both online and offline.

This is a completely new approach based on verifiable numeric evidence. Roberto Cavalli now has the opportunity to compare results across different discount levels for similar products, helping the brand uncover new insights into pricing strategy
Increasing the discount does not always increase the quantity sold per item
Low stock level needs more discount push
Fast corrective action could be taken by monitoring customers' reaction to different discount
Products within the same merchandising class do not need the same discount level as customers' perception on products are different

The Results ✨: The overall uplift was more than +35% in revenue and +50% in quantity.
To measure the financial impact brought by Differs, we conducted a live A/B test by splitting thousands of products from the last three seasons into two equivalent groups (control and test) in terms of number of items, revenue, reference price, and discount distribution over the past 3 months. Because of this equivalence, both groups were expected to generate similar revenue and volume if they had followed the same pricing strategy.
However, we applied our old pricing strategy to the control group and Differs' recommendations to the test group. Both groups served the same objectives:
For slow movers: reduce all stock as quickly as possible
For older and latest collections: maximize revenues

Final thought: The results speak for themselves!
"Before Differs, our price strategy for old stock was quite a mess. We were stuck with a basic approach. With Differs, we can now leverage test & control alongside predictive modeling." said Fabrizio Viacava, Global Digital Director
The average discount difference between the control and test groups ranged from 3% to 12%, depending on the collection. And in just 8 weeks, we achieves a +35% increase in net revenue—equivalent to +100K€ on 570 items and more than +50% in quantity sold.
That means half as much stock was left sitting—twice as much product was sold—without compromising margin. And with more than 3,000 items typically on discount, the financial potential is exponential. The results speak for themselves!
Differs in three words? Efficient. Easy to use. Effective.
What Roberto Cavalli achieved with Differs - in just 8 weeks:


Cr. Vogue, Roberto Cavalli, Fall 2020, Ready-to-wear
The Brand 🤎
Roberto Cavalli, founded in Florence in 1970, is an iconic Italian fashion house known for its bold designs, exotic prints, and sensual style. Celebrated for innovations like printed leather and stretch denim, the brand blends opulence with Italian flair and artisanal craftsmanship.
The Pain 💥: In a luxury house like Roberto Cavalli, there’s no room for unsold merch. Most discounts followed simple, stock-based rules—leading to an inefficient destocking strategy.
Even the most iconic brands such as Roberto Cavalli need to leverage price changes to get rid of unsold merch and increase revenues. But instead of having an optimal pricing strategy to destock while preserving brand image and profit, discounts were selected based merely on simple rules. It was a mixed approach between following competitors, aligning with marketplaces, and assessing the store's readiness to manage price lists. Discounts on the latest collection were usually a fixed percentage (40–50%) applied across the entire range, with no differentiation between lines, colors, regions, etc. For older seasons, the discount percentage increased with the age of the collection. This is not effective for the brand or for customers.
Risk? Brand equity dilution, revenue loss, unpredictable sell-through rate (STR).
The Cure 🧪: Moving towards an advanced predictive algorithm with Differs, where humans interact with the models to decide the right discount for each item.
Differs helps centralize historical data and builds a sophisticated, custom predictive model that truly understands our customers' behavior: a combination of willingness to buy, long-term stock forecasts, and external effects like weather, seasonality & events or competitor pricing. The model gives rise to an insightful and intuitive tool that explains the motivation behind each discount level chosen. This allows the team to determine the best price for each item—clearing stock with maximum profit—along with continuous support from Differs.
Differs' optimal discounts are applied across the globe, both online and offline.

This is a completely new approach based on verifiable numeric evidence. Roberto Cavalli now has the opportunity to compare results across different discount levels for similar products, helping the brand uncover new insights into pricing strategy
Increasing the discount does not always increase the quantity sold per item
Low stock level needs more discount push
Fast corrective action could be taken by monitoring customers' reaction to different discount
Products within the same merchandising class do not need the same discount level as customers' perception on products are different

The Results ✨: The overall uplift was more than +35% in revenue and +50% in quantity.
To measure the financial impact brought by Differs, we conducted a live A/B test by splitting thousands of products from the last three seasons into two equivalent groups (control and test) in terms of number of items, revenue, reference price, and discount distribution over the past 3 months. Because of this equivalence, both groups were expected to generate similar revenue and volume if they had followed the same pricing strategy.
However, we applied our old pricing strategy to the control group and Differs' recommendations to the test group. Both groups served the same objectives:
For slow movers: reduce all stock as quickly as possible
For older and latest collections: maximize revenues

Final thought: The results speak for themselves!
"Before Differs, our price strategy for old stock was quite a mess. We were stuck with a basic approach. With Differs, we can now leverage test & control alongside predictive modeling." said Fabrizio Viacava, Global Digital Director
The average discount difference between the control and test groups ranged from 3% to 12%, depending on the collection. And in just 8 weeks, we achieves a +35% increase in net revenue—equivalent to +100K€ on 570 items and more than +50% in quantity sold.
That means half as much stock was left sitting—twice as much product was sold—without compromising margin. And with more than 3,000 items typically on discount, the financial potential is exponential. The results speak for themselves!
Differs in three words? Efficient. Easy to use. Effective.
What Roberto Cavalli achieved with Differs - in just 8 weeks:


Cr. Vogue, Roberto Cavalli, Fall 2020, Ready-to-wear
The Brand 🤎
Roberto Cavalli, founded in Florence in 1970, is an iconic Italian fashion house known for its bold designs, exotic prints, and sensual style. Celebrated for innovations like printed leather and stretch denim, the brand blends opulence with Italian flair and artisanal craftsmanship.
The Pain 💥: In a luxury house like Roberto Cavalli, there’s no room for unsold merch. Most discounts followed simple, stock-based rules—leading to an inefficient destocking strategy.
Even the most iconic brands such as Roberto Cavalli need to leverage price changes to get rid of unsold merch and increase revenues. But instead of having an optimal pricing strategy to destock while preserving brand image and profit, discounts were selected based merely on simple rules. It was a mixed approach between following competitors, aligning with marketplaces, and assessing the store's readiness to manage price lists. Discounts on the latest collection were usually a fixed percentage (40–50%) applied across the entire range, with no differentiation between lines, colors, regions, etc. For older seasons, the discount percentage increased with the age of the collection. This is not effective for the brand or for customers.
Risk? Brand equity dilution, revenue loss, unpredictable sell-through rate (STR).
The Cure 🧪: Moving towards an advanced predictive algorithm with Differs, where humans interact with the models to decide the right discount for each item.
Differs helps centralize historical data and builds a sophisticated, custom predictive model that truly understands our customers' behavior: a combination of willingness to buy, long-term stock forecasts, and external effects like weather, seasonality & events or competitor pricing. The model gives rise to an insightful and intuitive tool that explains the motivation behind each discount level chosen. This allows the team to determine the best price for each item—clearing stock with maximum profit—along with continuous support from Differs.
Differs' optimal discounts are applied across the globe, both online and offline.

This is a completely new approach based on verifiable numeric evidence. Roberto Cavalli now has the opportunity to compare results across different discount levels for similar products, helping the brand uncover new insights into pricing strategy
Increasing the discount does not always increase the quantity sold per item
Low stock level needs more discount push
Fast corrective action could be taken by monitoring customers' reaction to different discount
Products within the same merchandising class do not need the same discount level as customers' perception on products are different

The Results ✨: The overall uplift was more than +35% in revenue and +50% in quantity.
To measure the financial impact brought by Differs, we conducted a live A/B test by splitting thousands of products from the last three seasons into two equivalent groups (control and test) in terms of number of items, revenue, reference price, and discount distribution over the past 3 months. Because of this equivalence, both groups were expected to generate similar revenue and volume if they had followed the same pricing strategy.
However, we applied our old pricing strategy to the control group and Differs' recommendations to the test group. Both groups served the same objectives:
For slow movers: reduce all stock as quickly as possible
For older and latest collections: maximize revenues

Final thought: The results speak for themselves!
"Before Differs, our price strategy for old stock was quite a mess. We were stuck with a basic approach. With Differs, we can now leverage test & control alongside predictive modeling." said Fabrizio Viacava, Global Digital Director
The average discount difference between the control and test groups ranged from 3% to 12%, depending on the collection. And in just 8 weeks, we achieves a +35% increase in net revenue—equivalent to +100K€ on 570 items and more than +50% in quantity sold.
That means half as much stock was left sitting—twice as much product was sold—without compromising margin. And with more than 3,000 items typically on discount, the financial potential is exponential. The results speak for themselves!
Differs in three words? Efficient. Easy to use. Effective.