Swiss e-commerce tested by generative environments

Note revised on 25 May 2026. Article originally published in January 2026 — full rewrite.

The "revolution" discourse on AI in e-commerce saturated the professional space for two years. It produced much noise and few operational trade-offs. The observable reality is more precise: Swiss e-commerce is not being revolutionised, it is undergoing a specific shift in the upstream phase of the purchase journey, where the consumer searches and compares. This shift imposes an editorial and structural discipline on merchants who want to remain visible. This note sets out that shift and what it calls for in practice.

The observable shift

Before 2024, online product search came down to two regimes: a direct search on Google that pointed to product pages or marketplaces, and an exploration on those marketplaces themselves. The e-merchant who wanted to be found optimised their product pages for Google and tended to their presence on marketplaces.

Since 2024, a third regime has settled in: conversational search. A buyer who hesitates between several technical solutions, who compares brands on precise criteria, or who wants contextualised advice without having the time to browse twenty product pages, formulates their query in a generative environment. Perplexity has deployed a dedicated Shopping feature. ChatGPT and other generative environments now offer search experiences that can include product comparisons. Google has integrated AI Mode and AI Overviews directly into its search interface, in line with its official doctrine published on 15 May 2026^[1].

This third regime does not replace the previous two. It is added to them, captures part of the upstream journey, and calls for specific optimisations.

What generative environments read in a product page

When a generative environment synthesises several sources to answer a buyer's question, it operates a selection on the product pages it consults. This selection relies on criteria that e-commerce already knows, but which it tightens.

Structured data in JSON-LD format (Schema.org Product, Offer, AggregateRating, Review) becomes the language through which a product is correctly understood by models^[2]. A product page without structured mark-up may be misinterpreted, or set aside in favour of a better-described competing page. This requirement is not new in 2026; what is new is the potential loss of visibility when it is not held.

The text of the page must provide factual elements that are usable out of context. A marketing-slogan-oriented description — "the ideal solution for your needs" — provides no citable information. A description that specifies technical characteristics, typical uses, the product's limitations, return conditions, the effective delivery time, provides the substance that a generative environment can synthesise to answer a buyer.

Verified reviews count to the extent of their quality, not only their volume. A product with two hundred generic five-star reviews is less drawn upon by models than a product with forty detailed reviews describing actual use. This inversion changes the economics of review collection: the chase for volume gives way to the incentive for quality.

Swiss sensitivity to price variations

Dynamic pricing driven by predictive models has spread in several sectors of global e-commerce. In Switzerland, its deployment calls for particular restraint. Swiss commercial culture remains attached to price transparency, and the Price Indication Ordinance (PIO) regulates the display of prices to the consumer with a rigour that exceeds the European average^[3].

For a Swiss e-merchant, 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.

Multilingualism and trilingual coverage

The Swiss market structurally operates in three languages — French, German, Italian — to which English is added for a significant share of B2B transactions and international purchases. A product page available only in French is invisible to German-speaking buyers who formulate their query in German in a generative environment. A page available in German but not correctly translated into French returns a degraded image of the brand on the French-language query.

This requirement is not a 2026 discovery, it has long been carried by serious actors in Swiss e-commerce. What changes is the penalty: generative environments handle each language independently and produce different recommendations depending on the language of the query. A brand that sacrifices a language sacrifices a share of market that does not appear in its classical SEO dashboards.

FADP compliance and AI-driven personalisation

Any personalisation driven by a model — contextual recommendations, conversational chatbots, behavioural profiling — falls within the scope of the Federal Act on Data Protection (FADP), whose revised version came into force in September 2023^[4]. The obligations it imposes on an e-merchant deploying AI in its customer journey are precise: clear information to the consumer, legal basis for the processing, right of explanation, caution on international transfers of processed data.

An e-merchant who automates their personalisation without integrating these obligations into the initial framing exposes themselves to a reputational and legal risk disproportionate to the expected conversion gain. Conversely, a disciplined, transparent, compliant deployment becomes in itself a signal of trust that Swiss consumers recognise.

What a Swiss e-merchant arbitrates in practice

The shift described above does not call for a major technical rebuild for most merchants. It calls for three operational trade-offs.

The first is editorial: reviewing the product pages so that they provide factual information usable out of context, in each language covered. The work is progressive, by product category, and it immediately benefits the three search regimes — classical Google, marketplaces, generative environments.

The second is structural: auditing the quality of existing JSON-LD mark-up, identifying pages that do not carry Product, Offer, AggregateRating and Review schemas, and closing the gaps. This work falls under technical integration and does not require a long project.

The third is measurement: assessing the citability of the brand and its products in generative environments, on the queries that matter — those buyers actually formulate when looking for a product. This measurement did not exist two years ago; it now exists according to codified protocols, such as the one set out in the Cahier MCVA n°1.

The discipline that distinguishes, and the agitation that distracts

The shift to generative environments rewards investment in the fundamentals: honest and complete product pages, clean mark-up, sustained multilingualism, FADP compliance built in, measurement of gaps. This discipline is not spectacular. Nor is it optional. It distinguishes, in the Swiss market, the e-merchants consolidating their position from those who let themselves be distracted by the next AI feature announcement.

Swiss e-commerce is not in peril. It is under reinforced demand.

Sources

[1] Google Search Central, Optimizing your website for generative AI features on Google Search, published 15 May 2026. developers.google.com/search/docs/fundamentals/ai-optimization-guide [↩]

[2] Google Search Central, Product structured data. developers.google.com/search/docs/appearance/structured-data/product [↩]

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

[4] 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 [↩]

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