On this page
For most of 2025, asking a chatbot to help you shop got you a short list: three or four product cards, an image, a price, a star rating. Through 2026 that changed on every major engine. ChatGPT, Perplexity and Gemini each shipped a version of the same upgrade: a comparison table that sits inside the answer, lines products up against each other on the attributes that actually decide a purchase, and tells you which one wins on which point. That is not a cosmetic redesign. It is the engine doing the comparison shopping itself, in the same place it used to just list options, and it changes what a product needs to show up with in order to win.
The three engines moved at different speeds
ChatGPT went first on the underlying feature and then came back for the comparison layer. OpenAI’s Shopping Research tool, which builds a personalized buyer’s guide from a short back-and-forth, launched in November 2025. On March 24, 2026, OpenAI replaced the old widget of three or four product images with an explicit table layout for head-to-head comparisons, and it was already common within days: Evertune reviewed roughly 21,000 ChatGPT shopping-style responses on March 26 and found nearly 6,400 of them, about 1 in 3, used the new table format across all 11 product categories it checked (Evertune).
Perplexity had comparison baked in earlier, through product cards that pair images, prices and ratings with a side-by-side view and a buy-now button, extending the “Buy with Pro” checkout it introduced for Pro subscribers in November 2024 and later opened to PayPal-based checkout for all users.
Google moved last but broadest. Its Shopping Graph, which the company describes as more than 50 billion product listings with 2 billion updating every hour, now powers comparison tables in both the Gemini app and AI Mode in Search, so a shopper can describe what they want in plain language and get a table of considerations specific to that product category rather than a list of links (Google).
The table is the real change, not the checkout button
The checkout integrations get the headlines, but the table is the bigger shift for how a product gets chosen. A checkout button changes where a purchase completes. A comparison table changes who does the comparing. Instead of a shopper reading three retailer pages or a review roundup and forming their own judgment, the engine states the judgment directly, in a format built to look authoritative: rows, columns, a verdict per attribute. Evertune’s review of the ChatGPT tables found that 88% of them included an explicit “best for” or “best use” row, the great majority using language like “best for budget” or “best overall” (Evertune). That is the engine doing the work a comparison article or an in-store salesperson used to do, and it is doing it from whatever data it can read about each product, which is exactly why the underlying product data now matters more than it did when the output was a plain list.
Why engines are racing to own this moment
The incentive behind all three launches is the same, and it shows up in how fast shopping behavior is actually moving. Salesforce’s 2026 Holiday Predictions report found that half of shoppers now use an AI assistant at some point in their buying journey, and that 74% say they trust the product recommendations they get from an AI chat assistant (Salesforce’s report, via Communicate Online). That combination, more people starting there and most of them trusting what they are told, is why none of the three engines is treating shopping as a side feature anymore.
The analytics layer caught up within weeks
Brand-level visibility tracking stopped being enough to answer the question these tables raise, which is not “do I appear” but “which of my products wins the comparison.” Two AI-visibility platforms built for exactly that launched product-level tracking within the same stretch of 2026. Profound’s Shopping Analysis, launched November 13, 2025, captures the full response whenever a tracked prompt triggers shopping mode in ChatGPT, including the product tiles, images, prices and merchant links, and groups results by product and competitor rather than just by brand (Profound). Peec AI’s AI Shopping Analytics, launched June 17, 2026, goes SKU by SKU: visibility, win rate and ranking position for each product, the price the engine quotes checked against the actual catalog price, and which competitor takes the slot when a given product doesn’t make the shortlist (Peec AI). Both tools exist because a single “visibility score” for a multi-product brand hides exactly the split that matters, the same gap covered in why an AI shopping assistant picks one of your products over another.
What this means for how products get evaluated
A table forces an engine to commit to specific attributes for every row, not just a general impression of the brand. That raises the cost of the kind of data gaps that used to be invisible in a plain-list answer: a missing price, an inconsistent spec, a product page that doesn’t state plainly who it’s for. It also means the comparison is now happening per product, across however many SKUs a catalog has, which is a different measurement problem than tracking whether “the brand” shows up. Tracking visibility across multiple product lines covers how to set that measurement up so a losing SKU doesn’t get averaged into a healthy-looking brand number. It is also worth remembering that this shift rode in on an ordinary model and feature update cycle, the kind covered in what actually changes when a model updates: none of the three companies framed these launches as a visibility change, but a comparison table is exactly the sort of feature that reshuffles which sources and which products get used in an answer.
Track the comparison, not just the mention
If you sell more than one product in a category, the question worth answering now is not whether your brand gets named, it’s which of your products the table seats at the top and which one it leaves off entirely. Our guide to the engines worth tracking in 2026 covers how to prioritize ChatGPT, Perplexity and Google’s surfaces for your buyers specifically. A free Visibility Check against two or three of your most comparable products will show you whether a table is already being built around you, and Monitor tracking turns that single look into a trend you can act on before a line loses the comparison entirely.