Pinterest is opening a new bidding lane inside its search results: one built around what a user is visually browsing, not the words they typed to get there. The company announced Visual Search Ads on September 17, a performance format that lets advertisers bid for prominent placement as shoppers search, compare products and inch closer to buying, with the ad slots surfacing inside both search results pages and individual Pin detail views. Search advertising has run on keyword auctions for two decades. Pinterest is betting its visual-recognition technology can open a second axis of intent alongside them.
The format pairs ordinary keyword relevance with Pinterest’s own machine reading of a photo: what product it shows, what a shopper likely wants from it, and how urgently. Advertisers who win the bid can send users directly to their own websites. That gives brands a placement mid-funnel, while a shopper is still comparing options rather than after they have already decided what to search for by name.
Pinterest says the opportunity is large by its own measure: the company reports its users run over 80 billion queries a month, that most of those searches start from an image rather than typed text, and that better than half of them carry commercial intent. Those are Pinterest’s own figures, drawn from internal platform data rather than independent measurement, and the company has not published a methodology for how it classifies a “visual” or “commercial intent” search.
The ads run on Pinterest Intelligence, the company’s AI system for parsing images, products, tastes and user intent into advertiser-facing targeting signals. CEO Bill Ready framed the launch around a broader shift in how people express intent, saying shoppers now signal what they want through richer cues than keywords alone, and arguing that gives marketers a chance to reach them while people “see, save and decide.” That is a company describing its own product. Pinterest has not disclosed independent data on how Visual Search Ads perform against keyword-only placements.
Visual Search Ads enters beta for eligible advertisers across all of Pinterest’s ad markets in the coming weeks, built to support lower-funnel goals alongside Pinterest Performance+, the platform’s existing AI-driven ad suite. Pinterest paired the launch with two other advertiser tools: Priority Products, which lets brands flag specific items for the system to favor around seasonal moments or promotions while Performance+ keeps optimizing automatically, and a new self-serve option for testing creative and targeting variants against one another, including inside Performance+ campaigns. A new App Promotion objective is also entering global beta, aimed at driving app installs. Pinterest says one advertiser using it saw as much as an 85% improvement in cost per install compared with prior approaches, a company-provided result the company itself says individual advertisers may not match.
For search teams, the shift matters less as a Pinterest feature and more as a preview of where bidding logic is headed. A visual-intent auction has no keyword string to target, no negative match, no rank tracker line item; the unit being bought is a moment of attention on a specific product image, so the standard playbook built around query volume and match types does not transfer cleanly. Teams running Pinterest as a discovery channel will need clean image-level product data and visual metadata in place before the beta opens, not a keyword list.
Pinterest is also extending the same AI stack to shoppers directly. A consumer feature called Restyle, launching in beta in the U.S. and Canada, lets users upload a photo of a room and generate AI redesign ideas to compare, though that is a product feature rather than an advertising one, and it is a smaller part of this week’s announcement than the ad format itself.
Search Engine Land reported the launch on September 17, drawing on Pinterest’s own announcement and the search-volume, visual-share and cost-per-install figures Pinterest supplied. None of those numbers carry independent verification, and advertisers weighing early access to the beta should treat them as marketing claims until Pinterest publishes performance data drawn from outside its own systems.
Search Engine Land reported this story, by Anu Adegbola, on September 17, 2026.
