Key takeaways
- ChatGPT served ads in about 32% of citation-enabled responses over the past 7 days as of mid-August 2026, up from zero before May 27, 2026 — and the rate swings between the low 20s and high 30s day to day, so a single spot-check tells you almost nothing.
- You cannot see this from a Plus, Pro, or Business ChatGPT account. OpenAI explicitly withholds ads from paying subscribers and from accounts flagged as under-18, which means most marketing teams checking from their own paid seats see a blank screen and wrongly conclude their category has no ad activity.
- Product feeds became mandatory for ChatGPT Shopping visibility in September 2026 — regardless of ad spend — and OpenAI's own ranking factors (availability, price, quality, primary-seller status) reshuffle products by an average of 5 positions compared to their Google Shopping rank.
- OpenAI's Ads Manager only reports on your own campaigns. It never shows you what competitors are running, so tracking rival ad creative, landing pages, and frequency requires a third-party tool that samples live ChatGPT responses the way a logged-out shopper would see them.
- Treat paid placement and organic AI citation as two separate, separately-tracked channels. Winning an ad slot does nothing for whether your product page gets cited organically in the same conversation, and vice versa.
Why this suddenly matters for DTC brands
For most of 2025, "AI search for e-commerce" meant one thing: trying to get your product mentioned organically when someone asked ChatGPT for a recommendation. That's still true, but it's no longer the whole picture. OpenAI opened ChatGPT to advertisers in the US in February 2026, and by the end of August its ad business was running at a $1 billion annualized revenue rate — reached in under 200 days. That's fast. Faster than most people in retail marketing expected, honestly.
What changed on the ground for a DTC brand is pretty concrete. Shoppers researching a purchase inside ChatGPT now see three different things stacked on top of each other: an organic answer that may or may not cite your product page, a shopping carousel with price-and-image cards pulled largely from Google Shopping feeds, and — on free-tier accounts — a labeled sponsored response underneath it all. Three layers, three separate things to monitor, three separate ways to be invisible.
How often do ads actually show up?
This is the part most teams get wrong, and it's worth sitting with for a second. According to Promptwatch's tracking of citation-enabled ChatGPT responses over the 90 days to August 17, 2026, ads averaged 20.1% of all responses that triggered a web search. But that average hides a lot of movement. Ads didn't exist at all until May 27, 2026. Then there's a visible jump past 30% on July 1 that stuck, and a single-day peak of 43.9%. The 7-day average sat at 32.4% as of the report date, up 12.5% from the prior week.
The breakdown by prompt type is the useful part for strategy. Of ad-bearing responses, 73.3% came from plain organic prompts (someone just asking a question, no brand name attached), 18.0% from brand-specific prompts, and 8.6% from competitor-comparison prompts. If your monitoring setup only runs your own brand name through ChatGPT, you're sampling the smallest slice of where ads actually show up. You'd see almost none of the real picture.
Separately, Promptwatch's data on shopping feature trigger rates shows product cards attach to a low single-digit share of all web-search responses, but that rate has moved in sudden steps — roughly doubling overnight in late May before dropping back weeks later. OpenAI is clearly tuning when shopping cards appear by category, which means a brand that wasn't eligible for cards in June might be eligible in October without changing anything, simply because OpenAI opened the window wider.
The blind spot almost everyone hits first
Here's a mistake I've seen repeated in half the blog posts written about this topic: a marketing team logs into ChatGPT on their work account, types a few prompts, sees no ads, and reports back that "ads aren't live in our category yet." That conclusion is almost always wrong.
OpenAI does not show ads to ChatGPT Plus, Pro, or Business subscribers, and it doesn't show them to accounts it has flagged as belonging to someone under 18. If your team is checking from paid work accounts (which, if you work in marketing, you probably are), you are structurally incapable of seeing the ads your customers see. You need to simulate the free-tier, logged-out experience — and ideally from the market and language your customers are actually in, since ads render per-region.
Add two more wrinkles. Ads attach per-response, not per-prompt, so the exact same question can return an ad on one run and nothing on the next. And ad delivery is auction-based with "context hints" rather than exact-keyword targeting, so advertisers themselves don't fully control which conversations their ad lands in. A one-off check, even from the right account type, is basically a coin flip.
What OpenAI's own Ads Manager will and won't show you
If you're running ChatGPT ads yourself, OpenAI's Ads Manager gives you real reporting on your own campaigns: impressions and clicks refresh roughly every 15 minutes, though spend data can lag 7-8 hours behind, so a $0 spend figure doesn't mean you haven't been charged yet. You can attach OpenAI's browser pixel or a server-side Conversions API (works with Shopify) to tie clicks to purchases, or just run UTM parameters through your existing analytics stack.
What Ads Manager will never show you is what anyone else is running. No auction insights, no competitor ad library, nothing. That's a real gap compared to Google Ads or Meta, where at least some competitive visibility exists natively. For ChatGPT, seeing a rival's ad copy, landing page, and how often they're winning the slot requires a separate tool that samples live responses from the free-tier perspective, across markets, repeatedly.
Getting your products feed-ready
Separate from the ad question, there's a harder requirement that kicked in during September 2026: product feeds are now mandatory for appearing in ChatGPT shopping results at all, regardless of whether you're spending a cent on ads. The feed goes in via SFTP, with credentials issued through Ads Manager, and OpenAI requires a minimum of 1,000 products and a maximum of 2 million per feed — the format resembles Google Merchant Center's spec but isn't identical, so don't assume you can just reuse the same file unchanged.
The October 2026 update added bulk campaign creation (upload a spreadsheet through Campaigns > Bulk uploads > Bulk sheet, pick the product feed template, and it spins up multiple campaigns and ad groups automatically) and virtual try-on for apparel and accessories across 65+ countries. The try-on detail matters more than it sounds: only brands with an active product feed campaign are eligible for the try-on button at all. If your catalog isn't feeding into Ads Manager, your products simply don't exist in that discovery moment, full stop.
Worth knowing how OpenAI ranks what shows up in the shopping carousel in the first place. Per OpenAI's own docs, ranking weighs availability, price, quality signals, and whether you're the primary seller of the item. Independent research cited by get-ryze.ai found around 83% of carousel products trace back to organic Google Shopping listings, with 60% coming from the top 10 Google Shopping positions — but ChatGPT reshuffles order by an average of 5 positions relative to Google Shopping, so a #5 Google Shopping rank can land at #1 in the ChatGPT carousel, or disappear entirely. Separately, Profound's analysis found top-ranked product cards were 123.58% more likely to carry high review counts (787 median vs. 352 for bottom-ranked cards). Review volume is doing real work here, not just price.
Building a monitoring setup that actually works
Given all of the above, a DTC brand's monitoring checklist for AI search ads needs to cover three distinct layers, tracked on the same prompt set over time rather than in one-off checks:
- Ad presence and competitive ad creative — who's advertising against your category's organic prompts, not just your brand name, since 73% of ad-bearing responses come from plain organic queries.
- Shopping carousel placement — whether your product cards appear, in what position, and whether that position tracks or diverges from your Google Shopping rank.
- AI crawler access to your product and landing pages — whether ChatGPT's and other bots' crawlers can actually reach and parse your catalog pages in the first place, since none of the above works if the content isn't crawlable.
| Tool | What it tracks for ads/shopping | Entry price | Good fit for |
|---|---|---|---|
| Promptwatch | Ads Radar (ad creative, advertiser, model, date) on every paid plan; Shopping Insights matches your catalog CSV to extracted product cards by ID | $95/mo (Essential) | DTC brands wanting both competitive ad tracking and product-card matching in one place |
| Otterly.AI | ChatGPT Ads and Shopping Cards included in every Brand Report at no extra charge | $29/mo (Lite) | Smaller teams wanting a cheap starting point |
| Semrush AI Visibility | Brand-level AI visibility tracking, less ad-specific depth | $99/mo | Teams already in the Semrush ecosystem |
| Peec AI | Multi-model prompt tracking with daily refresh | ~$75-245/mo | Agencies managing several brand accounts |
| Profound | Enterprise-grade monitoring, no self-serve pricing | Sales-led | Large catalogs, enterprise budgets |
One thing worth flagging clearly: none of these third-party tools replace OpenAI's own Ads Manager for reporting on your own campaign spend, clicks, or conversions. They exist specifically to show you what Ads Manager structurally can't — competitor activity and shopping-card placement across the prompts you don't own.
If you want to go deeper into product-level tracking, Promptwatch's Shopping Insights (available from the Business tier at $579/month) matches an uploaded catalog CSV against extracted product cards by exact product ID, refreshed daily, with a background job sweeping for new shopping blocks every 10 minutes. That matters because, per Promptwatch's research, the most common technical gap for DTC brands trying to track this themselves is a mismatched product ID between their Shopify catalog and what actually shows up in the platform's results. Get the ID mapping wrong and you'll think you're invisible when you're actually just mis-tracked.
For brands wanting to browse a wider set of options before committing, the GEO software directory at bestgeosoftware.com and the AI rank tracking listings at ai-rank-tools.com both cover tools in this space side by side.
Country eligibility, because this trips people up
ChatGPT Ads expanded fast through 2026: US, Canada, Australia, and New Zealand by February; self-serve opened to US businesses in May; UK in June; Brazil and Mexico in August; then a big jump to 31 European markets plus India and 11 MENA countries by late August, UAE/Saudi Arabia/Israel/Turkey in mid-September, and Southeast Asia plus Taiwan by late September, bringing the self-serve advertiser total to 63 countries. One rule catches brands off guard: the legal entity placing the ad and receiving invoices has to be registered in an eligible country. A US-registered DTC brand can't simply point ads at European shoppers without a billing entity set up in an eligible EU market, and the reverse is just as true.
Measuring whether any of it is working
Forbes contributor TerDawn DeBoe made a point in January 2026 that's held up well since: judging ChatGPT ad performance purely by last-click sales misses how these conversations actually function. A shopper might see your ad, keep chatting, research for another ten minutes, and buy a week later through an entirely different channel. McKinsey's guidance here lines up — accurate customer data and a regular tracking cadence, not a single campaign snapshot, is what tells you whether the spend is doing anything.
The practical version of that advice: treat paid placement and organic AI citation as two tracked channels that happen to live in the same app. Running an ad doesn't change whether your product page gets cited organically in the same ChatGPT conversation, and showing up organically doesn't mean you're getting ad impressions on adjacent queries. Measure both, separately, on a recurring schedule, from an account that actually sees ads.
If your team is building this out for the first time, start by confirming you're even checking from the right vantage point: a free-tier account, in the right market, run repeatedly rather than once. Everything else, product feeds, carousel position, competitor ad tracking, builds on getting that baseline right first.
For broader directory browsing of AI visibility and e-commerce-focused tools, surferstack.com's software directory is a reasonable starting point if you want to compare more options before settling on a stack.