Italian consumers are rewriting the rules of search faster than most marketing departments have revised their measurement frameworks. On 18 March 2026, IAB Italia brought together researchers, platform representatives, and agency leaders at Cariplo Factory in Milan for Search Forward, a full-afternoon conference dedicated to mapping the structural shift underway in how Italians discover and buy products. What emerged was a portrait of an ecosystem mid-transition - one where data is scarcer, AI is growing, and the old logic of clicks as the primary currency is losing ground.

The numbers are stark. According to the Osservatorio Search in Italy 2026, a survey of 1,000 online adults conducted between 2 and 9 March, 35.34% of Italian online adults aged 18 and over now use generative search tools to inform shopping decisions. That figure nearly doubled from 19.6% recorded in February 2025. Marco Loguercio, digital marketing and e-commerce advisor and creator of the Osservatorio - which has tracked Italian search behaviour since 2004 - presented those results at the event. The survey used a CAWI methodology with quota sampling by gender, age, and geographic area.

The near-doubling of generative search adoption in roughly 12 months raises immediate questions about channel attribution. According to the data presented at the conference, 88.6% of respondents had already tried a generative search tool for shopping purposes, and 68.6% described them as useful or indispensable. At the same time, the purchase journey remains fragmented across platforms. Physical stores ranked first as information sources, cited by 73.9% of respondents, followed by traditional search engines at 72.1%, marketplaces at 59.7%, generative search at 45.3%, and social media at 42.1%.

Google AI Mode is gaining ground within that generative search category, but from a low base. According to the March 2026 data, 9.4% of Italian online adults use Google AI Mode to research purchases - the first measurement of that figure since the feature became available in Italy in late October 2025. Of those AI Mode users, 44.2% have stopped using traditional Google Search entirely for shopping, relying exclusively on the conversational interface. The remaining 55.8% use both. The combined footprint of Gemini and AI Mode now exceeds ChatGPT's share in Italy when measured against shopping-related use: 25.3% for the Google ecosystem against 23.4% for ChatGPT.

Google expanded AI Overviews to Italy and eight other European countries in March 2025, and subsequently brought AI Mode to more than 40 countries in October 2025. The Italian adoption figures now provide one of the first measurable signals of how that rollout has shaped actual consumer behaviour in a major European market.

The myth-busting session

One of the more structured contributions at the conference was a segment dedicated explicitly to challenging prevailing assumptions. The Osservatorio data was used to test four claims that circulate widely in the Italian marketing community.

The first - that a strong brand offers immunity from AI disruption - did not survive contact with the data. According to the survey, 33% of respondents now make purchasing decisions based on immediate needs rather than brand loyalty, and 14% reported discovering new brands specifically through AI tools. Among consumers aged 18 to 24, 55.6% said they had changed their purchasing habits. That figure declined steadily with age, reaching 29.8% for consumers aged 55 and above. AI tools are actively surfacing alternatives that bypass established brand preferences, particularly for younger cohorts.

The second claim - that generative AI delivers accurate, reliable product information - fared no better. According to the survey results, 67.3% of users encountered at least one problem when relying on AI-generated product recommendations. The most frequent issues were generic advice (cited by 22% of those who experienced problems), incorrect pricing (18%), unavailable products (16%), and imprecise specifications (16%). Fifteen percent received recommendations with no verifiable source. The data suggests AI tools often function as a starting point rather than a definitive reference, particularly for product-specific decisions where accuracy matters.

The third assumption tested was that AI will displace physical retail. The data did not support a sweeping conclusion. According to the survey, 64.4% of respondents who used a generative search tool in a shopping context subsequently visited a physical store - 39.3% purchased there, while 25.1% ultimately bought online after visiting. Only 24.2% completed a purchase online without setting foot in a shop. The data points to generative search functioning as a shortlisting tool rather than a terminus.

The fourth claim - that agentic e-commerce will become the dominant mode of purchase - met the most scepticism. When asked what role they wanted AI to play in future purchases, 28% said none at all. A further 62.2% preferred a reactive role, with AI assisting only when prompted. Just 9.8% favoured a proactive role where AI anticipates needs or manages the entire process autonomously. The data reflects a consumer base that values assistance but retains strong preferences for control.

The measurement problem

Running alongside the consumer data was a distinct thread about what the industry can and cannot actually measure. Emiliano Sammassimo, Head of Search Analytics at Navla - a brand within ByteK, part of the Datrix group, and a participant in defining IAB Tech Lab's CoMP (Content Monetization Protocol) - framed the challenge in his presentation titled "Decodificare l'Invisibile" (Decoding the Invisible).

According to Sammassimo's presentation, the structural problem is this: LLMs do not provide tracking data. Platforms built to monitor brand presence in AI-generated results can only observe from the outside. There are no official user counts, no transparency about when or how brand properties are cited, and no access to proprietary insights from the systems generating the answers. The entities that own the data simply do not share structured information.

The KPI implications are significant. According to Sammassimo's analysis, the transition from traditional search to a world increasingly shaped by AI-generated answers requires a shift in what marketers measure. Traffic alone is insufficient. The metrics that now matter include mentions, citations, source status, presence in AI Overviews, and LLM inclusion - none of which are currently available through standard tools like GA4, Google Search Console, Bing Webmaster Tools, or Google Ads. New categories of tooling - LLM monitoring platforms, sentiment analysis systems, and vertical analyses of Reddit and user-generated content communities - are required to fill the gap.

Zero-click search has been documented as a structural force since well before generative AI arrived. According to Sammassimo, zero-click searches have existed since 2016. What has changed is scale and irreversibility. For publishers, the inability to convert search visibility into traffic is a crisis. For e-commerce operators and lead generation businesses, it is a less acute problem - users who do click tend to arrive with stronger intent.

The Navla presentation proposed a deductive methodology for working around the data gap. The approach starts with SERP analysis in Google, where observable demand signals still exist. People Also Ask questions are extracted and converted into prompts, which are then tested against multiple LLMs through proprietary APIs. The resulting patterns are organised to accelerate analysis of which brand signals influence AI-generated responses. It is a workaround rather than a solution, but one of the more systematic approaches presented at the conference.

WPP Media and the question of AI visibility

The WPP Media session, presented on 18 March 2026 in Milan under the title "Ricerca e visibilità nell'era dell'AI" (Search and visibility in the AI era), approached the same structural shift from an agency perspective. The team - Veronica Benigno (Consulting Director, Choreograph), Luana Giupponi (Project Management Director, Entertainment), Marco Magnaghi (Managing Director, Media Solutions), and Silvia Martella (Director, Search and Social Services) - framed the core question as one of "future readiness." AI Visibility emerged as the new strategic frontier, moving beyond traditional SEO and paid search logic.

According to the WPP Media materials, search has moved from a mechanism that returns answers to a conversational process - one where brands must govern their presence across outputs they do not directly control. The session challenged participants to test prevailing assumptions about SEO and SEA and to interrogate their own readiness for a search environment where AI intermediation determines which content reaches users and which brands are cited.

PPC Land has tracked the growing gap between impressions and clicks that AI Overviews create, with independent research documenting organic clicks falling by 34.5% when AI Overviews appear. That erosion of click-through rates - measured across multiple markets and methodologies - is the backdrop against which Italian marketers are now operating.

Agentic systems: more announcements than applications

A recurring theme across several presentations was the gap between the pace of announcements and the pace of real-world deployment. Sammassimo described the current period as one of structural transition, characterised by more announcements than concrete implementations. New standards including ACP (Agentic Commerce Protocol), UCP (Universal Commerce Protocol), and WebMCP have been announced but are not yet widely operational. Advertising within AI Overviews and on ChatGPT, he noted, remains in an early phase.

The conference's platform panel - which included Giulia Bessi from TikTok Italia, Isotta Enrici from Google Italy, and Samuele Serci from Microsoft Advertising Italy - reflected the same caution: infrastructure is young, protocols are new, and consumer willingness to delegate purchasing decisions fully to AI agents remains limited. The 9.8% who want a proactive AI shopping agent is a segment, not a mass behaviour.

IAB Tech Lab CEO Anthony Katsur expressed similar caution about agentic AI in December 2025, describing the discourse as having reached a fevered pitch while fundamental infrastructure questions remain unresolved.

Shopping categories and decision velocity

The Osservatorio data also mapped how AI use varies by product category. Generative search performs different roles depending on the type of purchase. Among the products most frequently acquired with AI assistance: clothing and footwear accounted for 33% of AI-assisted purchases, over-the-counter pharmaceuticals and supplements 31.4%, and sports clothing and equipment 31%. In the luxury segment, 18.3% of buyers of branded clothing, jewellery, or watches used generative search before finalising their decision.

Decision timelines vary significantly by category. Fast-moving categories - where decisions are made within two days - include grocery (70.3%), pharmaceuticals (68.9%), pet supplies (68.4%), books (59.8%), and beauty products (57.3%). Slow categories - where decisions extend beyond a week - include furniture and home furnishings (38.8%), travel (37.4%), utilities (35.1%), financial products (34.5%), and insurance (33.4%). The implication for marketers is that the relevance of AI-assisted shortlisting differs substantially across verticals: a furniture brand has a longer window in which to be cited and considered; a pharmacy brand competes for mention in a much tighter decision cycle.

The demographic picture complicates simple generalisations about AI adoption being a youth phenomenon. While younger cohorts lead in discovery behaviour - with Gen Z respondents in the data more likely to start with TikTok, Instagram, or YouTube than with Google - AI use for shopping cuts across age groups. Men currently use generative search for shopping at higher overall rates than women, but the gap narrows or reverses in specific categories such as beauty and personal care.

What the industry is adapting to

The conference's overarching message was not that search is dying. The data from multiple presentations showed traditional search engines retaining high use rates. According to the Osservatorio, traditional search engines remain the second most important information source for Italian online shoppers, cited by 72.1% of respondents. The argument being made was more specific: the value generated by search presence now flows through mechanisms that existing measurement systems cannot adequately capture.

The growth of ChatGPT as a referral source alongside declining organic clicks illustrates the recomposition rather than the death of search. Traffic volumes may compress; the quality and intent of remaining traffic may increase. According to conversion data cited in the broader industry literature, AI search visitors convert at significantly higher rates than traditional organic traffic - a finding that partially offsets the volume decline but does not resolve the attribution challenge.

The practical priorities outlined by Loguercio at the close of his presentation were structured as five actions. First, analyse how customers actually use generative search to reach specific products and services - the data exists and can be found. Second, map and monitor all points where demand forms, not only the dominant channels. Third, recognise that AI tools cite sources that help them respond well, meaning content that is structured, authoritative, and specific creates better citation opportunities. Fourth, make websites and other touch points function as tools for reducing perceived risk: according to the Osservatorio data, the content most effective at supporting decisions includes product images (cited by 50.7%), text and product detail pages (42.4%), in-depth video (41%), and maps or store availability information (40.3%). Fifth, measure how decision timelines vary by category, since changes in those timelines are an early signal of whether marketing actions are actually working.

None of those priorities require abandoning existing search marketing infrastructure. They do require extending it, supplementing it with new measurement approaches, and reconsidering the logic by which brand visibility is defined and tracked.

Timeline

Summary

Who: IAB Italia, together with researchers, agency leaders from WPP Media, Mediaplus, and Navla, and platform representatives from Google Italy, Microsoft Advertising Italy, TikTok Italia, and Pinterest Italy, convened at Cariplo Factory in Milan. The Osservatorio data was presented by Marco Loguercio; the brand visibility measurement session was led by Emiliano Sammassimo of Navla.

What: The Search Forward conference presented data from the Osservatorio Search in Italy 2026 showing that 35.34% of Italian online adults now use generative AI for shopping - nearly double the February 2025 figure of 19.6% - while 9.4% specifically use Google AI Mode, available in Italy only since late October 2025. Presentations addressed brand visibility measurement, the limitations of existing analytics tools for tracking AI-generated citations, and the structural gap between current KPIs and the new landscape.

When: The event took place on 18 March 2026. The survey data was collected between 2 and 9 March 2026.

Where: Cariplo Factory, Milan, Italy.

Why: The advertising and marketing industry in Italy is navigating a transition in which AI-generated search results alter how brands are discovered, cited, and measured. Existing analytics tools - GA4, Google Search Console, Google Ads, and equivalents - do not capture the new forms of visibility that determine whether a brand appears in AI-generated answers. The conference sought to map what is actually changing, quantify those changes with primary research, and identify frameworks for adapting measurement and content strategies accordingly.

Share this article
The link has been copied!