NIQ and Similarweb are building a joint measurement system that will connect product discovery inside AI assistants to website traffic, conversions, and sales. The first version is scheduled for the fourth quarter of 2026 and aims to solve a problem that has frustrated marketing teams since generative AI entered the shopping journey: most AI-influenced visits never show up as AI referrals in analytics dashboards.
The attribution blind spot
Similarweb published research in June 2026 that quantified the gap. After tracking users who received brand recommendations from ChatGPT, it found they were 2.5 times more likely to visit the recommended brand's website within the following week. But 55.9% of those visits arrived through search rather than a direct click from ChatGPT. Similarweb said it excluded users who had recently visited the brand or had already mentioned it in their prompt.
Adobe has reported a similar pattern. Adobe Digital Insights found that 34% of surveyed consumers used AI assistants for product research before later searching online for deals. The planned NIQ-Similarweb system will extend measurement beyond referral traffic to examine what consumers ask AI assistants, which products appear in recommendations, how brands compare with competitors in those results, and whether AI interactions connect with later website visits and purchases.
What the system will measure
NIQ will contribute product intelligence, consumer behaviour data, and retail sales measurement. Similarweb will provide digital data covering activity inside generative AI platforms and traffic reaching external websites after consumers interact with those services. The companies said the traffic component will include both direct clicks from AI platforms and subsequent visits linked to earlier AI-driven product discovery.
The system will also assess whether product content is complete and structured in a way that allows AI assistants to identify and represent products accurately. It will separately measure where products appear when AI assistants recommend options to shoppers and compare a brand's presence with competitors.
Shopping moves deeper inside AI platforms
The measurement challenge is growing more urgent as AI platforms add commerce functions that keep more of the shopping process inside the assistant. OpenAI introduced Instant Checkout in 2025, allowing eligible ChatGPT users to purchase supported products without leaving the conversation. The system uses OpenAI and Stripe's Agentic Commerce Protocol to pass order information to participating merchants, while merchants remain responsible for payments, fulfilment, and customer service.
Google has introduced its Universal Commerce Protocol, an open standard that allows agents and commerce systems to interact across consumer platforms, businesses, and payment providers. In May 2026, Google announced plans to roll out Universal Cart across Search and the Gemini app in the US during the summer, letting shoppers complete checkout on Google with participating brands or transfer items to a retailer's website.
NIQ's planned system is intended to combine several stages of AI-assisted shopping within the same measurement framework: consumer queries, visibility in AI-generated recommendations, product-content readiness, AI-influenced traffic, and subsequent conversion or purchasing activity. NIQ said existing measurement systems cover areas including retail, ecommerce, media, and search but do not provide a complete view of purchases involving AI-mediated discovery.
The revenue signal is already visible
Separate Adobe Analytics data reported by Reuters in June 2026 found that US shoppers referred to retail websites through AI services such as ChatGPT and Gemini generated 53% more revenue per visit than consumers arriving through non-AI sources. That figure covers identifiable AI referrals, not purchases where an AI interaction was followed by a visit through search or direct traffic. NIQ and Similarweb plan to include those subsequent visits in their measurement system.
Similarweb's US data also showed visits to AI platforms increased 28.6% between January 2025 and January 2026, while referral traffic from those services remained broadly flat. Users were carrying out more product research and comparison within AI services without immediately opening websites referenced in the response. Recent NIQ research released in August 2026 found that nearly 75% of shoppers use AI during product discovery, while 20% of consumers already use AI as part of the shopping process.
Why this matters for marketers
For marketing teams, the NIQ-Similarweb system represents a shift from treating AI as another referral channel to understanding it as a discovery layer that influences purchases across multiple touchpoints. Without measurement that connects AI interactions to later search visits, direct traffic, and sales, brands cannot accurately attribute spend or optimise their presence inside platforms like ChatGPT, Gemini, Perplexity, and Claude. The 53% revenue premium Adobe identified for identifiable AI referrals suggests the stakes are high - and the 55.9% of AI-influenced visits that currently go unattributed means most marketing teams are flying blind on the true impact. Professionals looking to build capability in this area may find structured learning resources such as AI for Marketing Managers useful for understanding how AI is reshaping attribution, content strategy, and campaign measurement.
Your membership also unlocks:
