Key takeaways
- ChatGPT Shopping still pulls its product catalog from Google Merchant Center, but it filters and ranks that data with its own logic, and the overlap with what shows in Google's own AI Mode carousel is tiny (around 1.28% per a Productrise study of 100,000+ SERPs).
- ChatGPT now shows actual paid ads in roughly 20-32% of web-search responses as of mid-to-late 2026, up from zero before May 27, 2026, with a self-serve Ads Manager, conversion-optimized CPC bidding, and product feed campaigns supporting up to 1 million SKUs.
- Google's ad stack inside AI Mode has moved fast too: sponsored tiles, mid-conversation paid listings, and (as of August 31, 2026) ad units embedded directly inside AI-generated text, not just stacked above or below it.
- Tracking tools differ sharply. Google Ads still has the most mature conversion infrastructure (Data Strength Uplift, Meridian GeoX, Customer Match with IP matching). ChatGPT's measurement stack (Pixel plus Conversions API, dynamic URL parameters since August 2026) is catching up fast but still lacks published benchmarks.
- Winning in both now requires separate feed strategies: new Merchant Center attributes built specifically for "conversational experiences" (question-and-answer, related product, popularity rank) matter for AI Mode and, by extension, for ChatGPT Shopping, since both draw from the same Shopping Graph.
Why this comparison keeps changing
If you wrote a guide on this topic in January 2026, half of it would already be wrong. ChatGPT Shopping had zero advertising back then, full stop. Researchers who pulled apart 1.1 million product tracking parameters from ChatGPT shopping sessions found exactly zero sponsored listings, just organic product cards sourced straight from Google's Merchant Center data. By late August, that had flipped: OpenAI launched conversion-optimized CPC bidding for product-feed campaigns, a carousel ad format, and measurement partnerships with Triple Whale, Sonar Optimize, and Hightouch.
Meanwhile Google wasn't sitting still either. It spent 2026 quietly reshaping how ads show up inside AI Mode, culminating in ad units that get woven directly into the AI-generated answer text rather than bolted onto the edges.
So the honest framing for 2026 isn't "ChatGPT Shopping vs Google Shopping ads" as two static products. It's two platforms racing to monetize AI-mediated shopping, built on overlapping but increasingly distinct data layers, with very different measurement maturity.
The shared foundation: Google Merchant Center
Here's the part that surprises a lot of marketers: ChatGPT Shopping doesn't have its own independent product database. It sources listings from Google Shopping's underlying catalog, meaning your Google Merchant Center feed is effectively your ChatGPT Shopping presence too. Every ChatGPT product URL still carries a srsltid=AfmBOo parameter confirming that lineage, alongside a simple utm_source=chatgpt.com tag for attribution.
That shared foundation matters practically. Fix a title, description, or GTIN issue in Merchant Center and it can ripple into both Google Shopping and ChatGPT Shopping visibility at once. But "same data source" does not mean "same results." A Productrise study covering 2 million+ product listings across more than 100,000 SERPs (US/UK, August 9-31, 2026) found only 1.28% of products in a standard Google Shopping carousel also showed up in Google's own AI Mode for the identical query on the identical day. AI Mode surfaced an average of 3.9 products per query versus 27.8 in standard search results, roughly a 7x narrower slate. Even when the same product appeared on both surfaces, the listed seller differed almost half the time (49.6%), and prices differed on 38.1% of matched products, with AI Mode showing the higher price in 68.4% of those mismatches.
That's Google's own AI surface disagreeing with its own Shopping tab on more than a third of prices. ChatGPT, pulling from that same messy underlying graph, inherits a version of the same problem. If your feed data is inconsistent across variants, regions, or pricing tiers, expect inconsistent representation everywhere AI touches it.
What actually shows up, and how often
The surface-level comparison people want is "how often do ads appear." The honest answer requires separating two different things on the ChatGPT side: the shopping feature (organic product cards) and actual paid ads.
Promptwatch's data on ChatGPT ad frequency shows ads went from zero before May 27, 2026 to appearing in roughly 20.1% of web-search-enabled responses over a trailing 90-day window through mid-August, with the daily rate crossing 30% on July 1 and holding there, peaking as high as 43.9% on some days. Most of those ads (73.3% on a 7-day average) show up against organic, non-branded prompts, not brand-name searches. Shopping-feature triggers, the product-card experience itself, remain comparatively rare across all prompts but concentrate heavily on commercial, transactional queries, which is exactly where you'd expect them.
On the Google side, AI Mode's ad rollout moved through clear stages across 2026: sponsored tiles in main search product grids landed in April, ads mixed into free listing grids in the Shopping tab in late April, mid-conversation paid listings in late May, and by August 31, Google was testing ad carousels placed inline, mid-response, inside AI Mode answers themselves, with a second sponsored block at the page footer. One observed example even had AI-generated descriptive copy written in the same tone as the surrounding organic answer, with a disclosure line noting the AI response about the ad was generated independently.
| Dimension | ChatGPT Shopping/Ads | Google Shopping/AI Mode ads |
|---|---|---|
| Ad presence | Zero before May 2026; now in ~20-32% of web-search responses | Present in standard Shopping for years; now also inline inside AI Mode answers since Aug 31, 2026 |
| Product catalog source | Google Merchant Center (shared) | Google Merchant Center / Shopping Graph (native) |
| Minimum spend | Eliminated entirely as of May 5, 2026 | No minimum; standard Google Ads auction |
| Bidding | CPC (recommended $3-5 start) plus oCPC since Aug 24, 2026 | Mature CPC/CPA/ROAS bidding, AI Max auto-migration |
| Feed scale | Up to 1,000,000 SKUs per advertiser | No published hard cap, enterprise-scale feeds standard |
| Conversion tracking | Measurement Pixel + Conversions API, dynamic URL params since Aug 2026 | Data Strength Uplift, Meridian GeoX, Customer Match, self-serve Conversion Lift |
| Published benchmarks | None official yet; third-party CPM estimates $16-60 | Extensive, years of published case studies and industry benchmarks |
Tracking: mature vs. catching up fast
This is where the two platforms genuinely diverge, and it's not close yet. Google's conversion-tracking stack had another full year of upgrades in 2026. Data Strength Uplift launched September 10, 2026, quantifying how much additional conversion volume you recover by strengthening first-party data signals, with brands connecting offline and app data seeing roughly a 26% average increase in incremental ROAS. Meridian GeoX, Google's geo-incrementality testing tool, moved from beta to full global availability the same day, at 31% lower cost than before. Multi-Source Conversions, a beta that launched September 21, lets CRM offline uploads backfill missing user data in tag-based events. Customer Match started accepting IP addresses and timestamps on September 25. Self-serve Conversion Lift testing opened to any advertiser with 1,000+ conversions and a $5,000 budget on October 7, no longer requiring a Google rep to set it up.
ChatGPT's measurement stack is real but younger. The OpenAI Measurement Pixel (a JavaScript snippet, oaiq.js) handles browser-side conversion events, and the Conversions API adds server-side reporting that's more resilient to browser blocking. OpenAI captures an "oppref" click reference in a first-party cookie to stitch clicks to later conversions, and it does modeled measurement when direct signal is missing. The August 2026 update added dynamic URL parameters, so campaign, ad group, and ad IDs auto-insert into destination links, which removed a real pain point: early advertisers had to manually tag UTMs and frequently got attribution gaps because of it.
What's still missing on ChatGPT's side: published benchmarks. There's no official CPM or CPC rate card from OpenAI, and third-party observed ranges swing from roughly $16 to $60 CPM depending on who's reporting and which vertical. Compare that to Google, where CTR, CPC, and conversion-rate benchmarks by industry have existed for over a decade. One widely cited comparison put average Google Ads CTR at 4-6% (top position closer to 7.94%), against ChatGPT ads averaging a 0.68% CTR overall, with top-quartile advertisers reaching 1.0% and best-in-class brands hitting 1.57%. Those numbers will move. They're not apples-to-apples yet because the ad formats, placements, and user intent are different, but they tell you where the maturity gap currently sits.
Optimization: the feed is no longer one thing
The single biggest practical shift in 2026 is that a Merchant Center feed now effectively needs to serve multiple distinct selection logics instead of one. Google added a batch of new Merchant Center attributes starting January 11, 2026, several explicitly flagged as "primarily intended for use in conversational experiences such as AI Mode": product highlight, product detail, variant option, item group title, related product, question and answer, document link, and popularity rank. The question-and-answer attribute alone supports up to 30 Q&A pairs per product, 1,000 characters each. Google reported that conversational attributes showed up in 50% of lululemon's AI Mode product recommendations during internal testing it announced in September 2026.
For ChatGPT, there's no separate feed to submit; it's still riding on the same Google Merchant Center data. But given the shopping-feature trigger concentrates heavily on commercial-intent prompts and ChatGPT cites product pages as its single most-cited content type (22.4% of all classified citations, rising to 26% for commercial-intent prompts specifically, according to Promptwatch's content-type citation data), the practical move is the same one you'd make for AI Mode: write product titles and descriptions in natural, conversational language instead of pure keyword strings, fill out the Q&A and related-product fields where they exist, and keep pricing and availability data consistent across every surface that touches the Shopping Graph.
A practical rollout checklist for 2026
For ChatGPT:
- Submit a sample of 100 products from your existing Google Shopping feed to the ChatGPT Ads Manager rather than building a separate feed from scratch.
- Start with your top 50 products by revenue and a modest $500-$1,000 monthly budget; use purchase as the conversion event and skip a hard CPA target at first so oCPC bidding can build up conversion history.
- Install a measurement partner connector (Triple Whale, for example) before you spend a dollar, and enable dynamic URL parameters at launch so you're not retrofitting UTM tags later.
- A/B test the carousel ad format against single-product ads for at least 30 days before drawing conclusions; the format is new enough that early winners aren't necessarily durable winners.
For Google Shopping and AI Mode:
- Fill in the newer Merchant Center attributes (question and answer, related product, popularity rank) even if you think your category doesn't need them; AI Mode's selection logic rewards structured conversational data differently than the classic Shopping algorithm does.
- Segment your Google Search Console and Google Ads reporting specifically for AI Mode impressions; Google Search Console now supports this breakdown, and lumping AI Mode traffic in with standard search hides real performance differences.
- Audit Merchant Center disapprovals and data-quality issues separately for the SKUs that show up in AI Mode versus standard Shopping; the Productrise overlap data suggests these are functionally two different inventories even though they share one feed.
- If you're running Performance Max or Demand Gen, check whether AI Max auto-migration has already applied to your Search campaigns (rollout began September 1, 2026); Google reports an average 7% conversion lift, up to 27% for campaigns that previously leaned only on exact or phrase match.
Where AI visibility tracking fits in
None of this feed and bidding work matters if you don't know whether your products are actually showing up in the first place, and that's a harder question than it sounds, because ChatGPT and Google's AI Mode don't expose the same kind of reporting dashboards that classic Google Ads does. This is the gap that AI visibility platforms are built to close: tools like Promptwatch track whether your products and pages get cited or recommended across ChatGPT, Google AI Overviews, AI Mode, Perplexity, and other AI surfaces, including dedicated ChatGPT Shopping and Ads Radar reporting that shows which competitors are bidding on your product-related prompts.
That matters here specifically because, as the Productrise findings above show, visibility in standard Google Shopping tells you almost nothing about visibility inside AI Mode, let alone ChatGPT. You genuinely need separate monitoring for each surface, not an assumption that ranking well in one means you're showing up in the other.
For ecommerce brands running product catalogs at scale, a handful of tools focus specifically on product-level AI visibility rather than generic brand mentions.
If you want to compare more options across the GEO and AI visibility category before committing, the directory at bestgeosoftware.com is a reasonable starting point for browsing tools built specifically for this purpose.
The honest bottom line
ChatGPT Shopping and Google Shopping ads aren't competing products yet, not really. They're two monetization layers sitting on top of mostly the same underlying product data, moving at different speeds. Google's version has fifteen-plus years of accumulated tracking infrastructure and is now bolting AI-native ad formats onto that foundation. ChatGPT's version launched its first ad, period, in May 2026, and is building measurement tooling in near-real-time as advertisers start spending.
The practical implication for anyone managing a product feed right now: treat Merchant Center as the one piece of infrastructure that touches everything, keep conversational-friendly attributes filled in, and stop assuming that strong Shopping tab performance guarantees strong AI Mode or ChatGPT performance. The data so far says it mostly doesn't.