Pricing: AI in service of margin, not just volume
- 1With costs unchanged, a point of net price typically returns close to twice what a point of volume does, without diluting scarcity.
- 2Margin is lost in the price waterfall (local discounts, distributor terms, currency gaps), rarely in the list price.
- 3AI helps rebuild the waterfall, spot deviations and estimate price sensitivity. Discount rules and positioning remain human decisions.
- 4Track realized net price and margin by customer, not revenue alone, and frame the use of customer data (Swiss FADP, European law).
In many retail and luxury companies, pricing is approached in one direction only: how to sell more. A discount is adjusted to clear a collection, a term is granted to keep a distributor, a promotion is launched to hold the quarter. When AI arrives, it is often wired to the same logic: forecast demand, optimize volumes, speed up turnover.
We think the other side of the equation deserves attention first. For a house whose value rests on desirability, margin is something to protect more than to conquer, and realized net price is the best indicator of how well it is protected.
A point of price is not a point of volume
Consider a worked example, with explicit assumptions. A house with CHF 120 million in revenue, a 60% gross margin, and variable selling costs of about 5% of revenue (commissions, logistics, packaging).
- One extra point of volume represents CHF 1.2 million of revenue. With a contribution margin of around 55%, the additional contribution is about CHF 0.66 million.
- One extra point of net price at constant volume brings CHF 1.2 million, less proportional commissions: in the order of CHF 1.1 to 1.15 million, with almost no additional cost.
The gap is a factor close to two. It still understates the advantage of price, because extra volume requires inventory, communication and sometimes new points of sale, and in luxury it can erode the scarcity that justifies the price in the first place. The figures depend on the assumptions above, but the order of magnitude holds across most cost structures in this sector.
Where margin actually leaks
The list price is rarely the problem. The leak sits in the waterfall that leads from list price to net price collected: discounts granted in boutiques by advisors, terms negotiated with multi-brand distributors, year-end rebates, returns, goodwill gestures, and price gaps between markets amplified by currency movements.
Each item is small and decided locally, often in good faith. Nobody sees the sum. For a brand selling in CHF, EUR, USD and Asian currencies, a strong franc can also open price gaps between regions within a few months, which invites arbitrage and grey-market activity.
What AI does well
Used with discernment, AI helps in four concrete ways:
- Rebuild the waterfall from point-of-sale, ERP and CRM data, and highlight deviations by boutique, advisor, customer and reference.
- Estimate price sensitivity by segment and collection, separating what belongs to season, launch or scarcity. Volumes are low on exceptional pieces, so estimates there should be handled with caution.
- Monitor price corridors across markets and currencies, and alert when a gap exceeds a threshold set by management.
- Recommend, not decide: a discount ceiling depending on context (customer, product, stock, relationship history), subject to human approval.
What it does not do
AI does not define positioning, range architecture or the decision not to sell at a given price. Prices that move in real time, as in air travel, fit poorly with a house whose clients expect consistency over time. In many cases, clear approval rules and a simple discount grid deliver most of the gain before any model is needed. The model then consolidates and monitors.
What is specific to luxury and watchmaking
Three features call for a sector-specific approach.
Global consistency. A client compares prices across Geneva, Paris, New York and Dubai. The price corridor is a brand decision, which data analysis informs but does not replace.
Independent distribution. Many watch houses also sell through multi-brand retailers. The terms granted to these partners weigh heavily in the waterfall and are often negotiated without a consolidated view of the true cost of the relationship. Teams need margin by customer, not only volume purchased. We discuss these issues on our page dedicated to retail, luxury and watchmaking.
Customer data. As soon as pricing relies on individual profiles, the revised Swiss Federal Act on Data Protection (FADP, in force since 1 September 2023) and European law apply: transparency on processing, vigilance on automated decisions, and disclosure to consumers when a price is personalized on the basis of automated processing in the Union. Pricing systems are not, in principle, among the high-risk uses of the EU AI Act, which entered into force on 1 August 2024, but the AI literacy duty, applicable since February 2025, deserves anticipation. It is better to frame these points at the start than to discover them afterwards.
Implications for leaders
1. Measure before optimizing. On a limited perimeter (one brand, two markets, one channel), rebuild the full waterfall. A few weeks is usually enough to reveal the most significant deviations.
2. Set the rules before the model. Who can grant which discount, in which circumstances, with what approval? This grid is a management decision, not a technical setting.
3. Capture the reason behind discounts. A discount without a reason cannot be analyzed. Structured sales call reports and an automatically fed CRM tie each commercial gesture to its context without adding to the workload of field teams. Likewise, qualifying inbound requests, notably from distributors, allows the right terms to be applied from the first exchange.
4. Track the right indicators. Net price realized against list price, discount rate by advisor and channel, margin by customer, price gap between markets. Revenue alone hides the leaks.
5. Tie pricing to strategy. The price level you want to defend is a matter of positioning and value creation, which we address in our Strategy & value creation expertise. AI executes and monitors a line; it does not draw it.
A discipline before a technology
The approach follows the usual order: a clear pricing strategy, deployment in the teams with simple rules, then steering over time. AI is used where it reduces effort or reveals a gap the eye cannot see, and nowhere else. This is the logic of our Growth & commercial performance expertise.
If you would like to know where the main margin leaks sit in your price waterfall and which decisions to take first, our diagnostic is a sensible starting point, with no commitment to a rollout.
A personalised first reading in three minutes.