AI-driven personalisation: between productivity of experience and Swiss discretion

Note revised on 25 May 2026. Article originally published in March 2026 — full rewrite. This note absorbs the content of the piece previously titled "UX in the AI era: the personalisation revolution", de-indexed and redirected to this page.

AI-driven personalisation has become a general topic of conversation in marketing and customer-experience departments. It promises an adaptation at the individual scale of everything that constitutes a customer journey: the content presented, the recommendations produced, possible price variations, the visual organisation of interfaces. This promise is not empty. In Switzerland, however, it confronts two structural constraints that significantly modify how it deploys.

This note sets out what AI-driven personalisation effectively does, and what the Swiss context — cultural and regulatory — calls for in calibrating its intensity.

What AI-driven personalisation does, compared with classical segmentation

Classical marketing segmentation operates on segments — groups of customers identified by a few common criteria, addressed with a uniform message. AI-driven personalisation operates on the individual, in real time, from a combination of behavioural and declarative signals specific to each user. This difference in granularity produces effects on three distinct action levers.

The content presented to a user adapts to their language, their interaction history, their supposed sector of activity, the type of question they previously asked, the depth they are at in their journey. A consulting firm can thus automatically present the case studies most relevant to the detected profile of the visitor. A B2B SaaS publisher can adapt its value proposition to the segment the visitor probably belongs to.

The recommendations produced — products, contents, resources, contacts — rely on models that have learned preferences from collective and individual behaviours. This capacity exceeds recommendations based on simple similarity — "customers who consulted X also consulted Y" — to integrate a contextual understanding of intentions, sometimes over multiple sessions separated by weeks.

Prices and conditions can also be personalised, within precise frameworks. Bundles adapted to the profile, promotions targeted at high-potential segments, payment terms modulated according to history. This dimension is the most delicate to arbitrate in the Swiss context — a point treated in detail below.

What the UX dimension adds to behavioural personalisation

Beyond content and recommendations, AI-driven personalisation now extends to the very organisation of interfaces. Design and analysis tools enable interfaces that automatically adapt to the identified user: priority navigation on the sections they consult most, guided journey for a new user, visual reorganisation according to terminal, time, or detected habits.

This UX extension also modifies the designer's role in the process. Repetitive tasks — graphic variants, resizings, generation of base mock-ups — are absorbed by the tools. The designer's value shifts towards defining the experience strategy, the quality of the brief given to generative systems, the rigorous curation of produced outputs, and the supervision of brand coherence across all touchpoints. This evolution is not a threat to the design function: it is a shift in its added value.

For operational steering, the integration of this UX dimension with behavioural personalisation calls for a coordination that marketing and design teams had not necessarily structured in the past. This coordination is an internal-organisation subject more than a technological one.

First Swiss constraint: cultural discretion

Swiss users have a relationship to privacy that significantly modifies the tolerance for visible personalisation. Perceptible personalisation — "we know you are interested in X", "based on your previous visits", a too-precise recommendation that reveals extensive tracking — can produce an effect opposite to the one sought. Instead of reinforcing perceived relevance, it triggers a withdrawal reaction from the visitor, who feels watched.

This cultural sensitivity is not negotiable by a marketing decision. It is coherent with the Swiss market's strong expectations regarding discretion and data protection, and it modifies the trade-off between intensity of personalisation and acceptability of the experience. The practical rule that emerges consists in favouring personalisation that is functional rather than demonstrative: relevance is verified in the quality of the experience, not in the staging of customer knowledge.

The high-end Swiss tourism market illustrates this tension well. A hotel chain that subtly recalls the preferences of a recurring client — without formulating it explicitly, without mentioning it in communications — produces a quality experience that the client recognises without having to formalise it. The same chain displaying "we remember you prefer rooms with a balcony" produces an intrusion effect that may suffice to drive away the segment it was trying to retain.

Second constraint: the FADP framework and European compliance

The Federal Act on Data Protection (FADP), in force in its revised version since 1 September 2023^[1], rigorously frames personal-data processing that feeds personalisation. For a Swiss company also addressing a European clientele, the General Data Protection Regulation (GDPR) applies in parallel.

Five operational principles follow. Information to the visitor on the data collected and its purpose, in accessible terms. Proportionality, which limits collection to the data necessary for the declared purpose. Purpose, which prohibits the use of data for objectives foreign to those for which it was collected. The right of access and deletion, which allows each user to consult and withdraw their data. Impact assessment for high-risk processing, which can, depending on the case, call for such an analysis for automated individual profiling.

The practical consequence for a Swiss company is to build personalisation on first-party data — collected directly from the customer with their explicit consent — rather than on data acquired through opaque channels. This discipline produces two beneficial effects: superior data quality because better qualified, and built-in regulatory compliance rather than retrofitted.

The specific tension of dynamic pricing in Switzerland

Among the levers of personalisation, dynamic pricing driven by predictive models deserves particular attention in the Swiss context. Swiss commercial culture remains attached to price transparency, and the Price Indication Ordinance frames price display to the consumer with a rigour that exceeds the European average^[2].

For a Swiss actor, this constraint does not invalidate dynamic pricing. It limits its amplitude and requires legibility of the pricing policy. Moderate adjustments, communicated clearly, remain compatible with the regulatory framework and with cultural expectations. Brutal and opaque variations penalise trust, which is the commercial asset hardest to rebuild on a market of restricted size.

This tension is not negotiable. It distinguishes calibrated price personalisation — which contributes to profitability without degrading trust — from unbridled algorithmic optimisation that maximises short-term revenue at the price of a lasting reputational effect.

The central trade-off: intensity of personalisation versus acceptability

The operational question facing a Swiss company is therefore not whether it should deploy AI-driven personalisation. The technology exists, it is accessible, and the competitive advantage it produces in certain contexts can be measured. The question is to calibrate the intensity of this personalisation to the equilibrium point where it effectively improves the experience without crossing the cultural and regulatory threshold of acceptability.

This trade-off does not resolve by a general formula. It is built through iterative work. A/B tests on the effect of different levels of perceptible personalisation. Measurement of satisfaction and return rate of customers exposed to different intensities. Documentation of observed thresholds by customer segment. Progressive adjustment of parameters according to real feedback.

Three operational principles distinguish the companies that hold this trade-off from those that miss it. Personalisation is tested before being deployed — not the other way around. Personalisation is measured on its real effects — customer satisfaction, loyalty, returns, and not on the immediate conversion rate alone. Personalisation remains reversible — the customer can always signal that they prefer a less personalised experience, and the company knows how to respect that signal without degrading the service.

The discipline that distinguishes, and the enthusiasm that distracts

AI-driven personalisation is neither a promised revolution nor a fashion. It is a tool whose value strictly depends on the rigour of the framework in which it is deployed. In Switzerland, this framework integrates cultural and regulatory constraints that significantly modify the default settings of systems available on the international market.

The Swiss companies that succeed in their AI-driven personalisation are not those who push intensity to the maximum technically possible. They are those who calibrate finely, measure rigorously, and know when to step back when the expected effect is not observed. This discipline distinguishes, once again, the actors consolidating their position on a demanding market from those merely replicating practices imported from different cultural ecosystems.

Sources

[1] Federal Act on Data Protection (FADP), revision of 25 September 2020, in force since 1 September 2023. www.fedlex.admin.ch/eli/cc/2022/491/en [↩]

[2] Price Indication Ordinance (PIO), SR 942.211. www.fedlex.admin.ch/eli/cc/1978/2057_2057_2057/en [↩]

Jérôme Deshaie is CEO of MCVA Consulting SA, a Swiss firm specialising in strategic consulting on artificial intelligence, based in Valais.