Retail measurement firm NIQ and web-analytics vendor Similarweb announced on September 2 that they are jointly building a product to track how AI assistants influence shopping, with an initial rollout set for Q4 2026. Nothing ships yet: the companies’ own FAQ answers the availability question with one word, no.
The stakes are structural, not incremental. Search teams and brands have spent two years guessing at how much AI-driven discovery converts into sales, because a chatbot conversation leaves no referral trail a browser-based analytics stack can follow. NIQ contributes retail data touching roughly 7.4 trillion dollars in worldwide consumer spending each year; Similarweb contributes the clickstream visibility inside AI platforms. Combining the two is meant to close that gap, not just describe it.
The product will track five things: consumer intent (what shoppers ask AI assistants), agentic shelf visibility (where a brand shows up in AI recommendations), product content readiness (whether AI can parse and recommend a listing), plus two conversion-side metrics, AI-driven traffic and AI-driven conversion, both anchored to verified sales. Coverage is planned across ChatGPT, Gemini, Google AI Mode, Perplexity and Claude, a five-platform spread that matters given how fast share has moved: Similarweb’s own June tracking showed ChatGPT’s share of generative AI traffic falling from 76.4 percent to 52.7 percent in a year, while Claude nearly tripled to 8.9 percent.
NIQ frames the opportunity around new purchase protocols, including Google’s Universal Commerce Protocol and OpenAI’s Agentic Commerce Protocol, both pitched as letting a shopper complete a purchase without leaving the AI conversation. The adoption record undercuts that framing. A May scan by Originality.ai counted just 26 live implementations of Google’s protocol among the three-million-plus sites it checked, and none of those adopters helped originally design the standard. OpenAI’s version fared worse in practice: the company shut down its Instant Checkout feature in March, months after Walmart reported that in-chat purchases converted at roughly a third of the rate seen among shoppers who instead completed checkout on Walmart’s own site. The protocol exists. The product it was built to power did not survive a year.
This is where the value of a verified-sales layer becomes clear, and where its limits show. Generative engine optimization vendors have sold visibility metrics for two years without a way to prove those metrics predict revenue. NIQ’s retail data could settle that argument for the first time by attaching real purchase records to AI citation counts. But an August framework from the Interactive Advertising Bureau found that just 16 percent of brands measure AI visibility in any systematic way, largely because no shared definition exists for what counts as a mention or a citation. NIQ has not disclosed whether its shelf-visibility tracking draws on real user prompts or a synthetic query library, a distinction the IAB framework says determines what the resulting numbers actually mean.
NIQ also arrives with an unusually concentrated position: it already owns retail sales measurement, purchase-based ad audiences, a data clean room, programmatic distribution and an enterprise data interface built on the Model Context Protocol. Layering an AI-visibility standard onto that stack puts one supplier at multiple points in a measurement chain that commerce media has historically tried to keep separated from media sales.
Search and e-commerce teams evaluating AI visibility vendors this quarter face a real sequencing choice. Committing now locks in a fragmented, self-defined metric set; waiting for NIQ’s launch means going through peak holiday trading with no verified-conversion view of AI-driven traffic at all. Whichever path they choose, product content readiness (the category that lines up directly with structured data and listing hygiene) is the lever teams can pull now, well before any vendor sets a launch date.
PPC Land reported the NIQ-Similarweb collaboration on September 3, 2026, citing the companies’ September 2 announcement.
