Last updated September 2026.
On Aug. 11, 2026, OpenAI expanded ChatGPT Ads into the United Kingdom, Mexico, Brazil, Japan, and South Korea, according to Digiday and Modern Retail. The same week, Tech Insider reported a bigger number underneath that country list: Otterly.AI’s 2026 tracking found that 76.4% of ChatGPT shopping-intent prompts now surface a sponsored product result next to the organic answer.
That number raises a question a lot of marketing teams cannot answer cleanly: which parts of a ChatGPT answer are earned, and which parts are bought? This post draws the line.
The three slots in a ChatGPT answer
Every ChatGPT response that touches a commercial query is built from up to three distinct zones. They come from separate systems, and they are labeled differently on screen.
┌──────────────────────────────────────────┐
│ A. SYNTHESIZED ANSWER │
│ The generated paragraph. Model output, │
│ grounded in retrieved sources. │
├──────────────────────────────────────────┤
│ B. ORGANIC CITATIONS │
│ [1] example.com [2] retailer.com │
│ Source links the model picked itself. │
├──────────────────────────────────────────┤
│ C. SPONSORED · Ad │
│ Product name · $price · Buy now │
│ A paid block, labeled, sold by OpenAI. │
└──────────────────────────────────────────┘- A. Synthesized answer: the paragraph ChatGPT writes. This is model output, built by the retrieval and generation pipeline described in the AEO pillar guide.
- B. Organic citations: the numbered source links the model attaches to that paragraph. These come from the same retrieval process as the answer text, not from an ad system.
- C. Sponsored block: a separate, labeled unit, built from a headline, a short description, an image, and a link. It runs through OpenAI’s ad auction, not through retrieval.
Slots A and B are earned. Slot C is bought. They can all three appear in the same response, and increasingly they do.
What each slot actually is
The practical difference comes down to control. You control a sponsored slot the way you control any paid search ad: bid, budget, creative, and targeting. You do not control the synthesized answer or the organic citations directly. You influence them the same way you influence any AI citation, through content quality, structure, and the third-party sources that mention your brand.
Where the two slots stand, as of Sept. 2, 2026
ChatGPT Ads shipped as a single-market pilot and has expanded on a rolling basis since. This table reflects the rollout as of the date above and will be updated in place as OpenAI ships new markets or Ads Manager features.
| Date | What shipped | Slot affected |
|---|---|---|
| Feb. 9, 2026 | ChatGPT Ads launches, U.S. only, Free and Go tiers | Sponsored slot (new) |
| Jun. 2, 2026 | United Kingdom added | Sponsored slot |
| Aug. 6, 2026 | Product carousel ad format ships across live markets | Sponsored slot |
| Aug. 11, 2026 | Mexico, Brazil, Japan, and South Korea confirmed live | Sponsored slot |
| Aug. 24, 2026 | Expansion into 31 European Union countries, opt-in personalization only | Sponsored slot |
| Ongoing | Self-serve Ads Manager access still limited outside the U.S. | Sponsored slot |
Nothing on that table touches the synthesized answer or the organic citations. The rollout is entirely a story about where slot C is for sale. Slots A and B run on the same retrieval logic everywhere OpenAI operates, whether or not ads are live in that market.
Can you buy your way into the answer box?
No. This is the question the 76.4% figure raises, and the answer does not change once you separate the slots.
The sponsored block and the synthesized answer come from different pipelines inside ChatGPT. Buying the ad slot puts your product card in the response. It does not move the model’s own text, and it does not add or remove an organic citation. A competitor can win the organic citation in the same answer where you paid for the sponsored card, and neither placement affects the other.
That matters for how you read the 76.4% stat. It does not mean 76.4% of the answer box is now for sale. It means 76.4% of shopping answers now carry a paid block in addition to whatever the model decides to write and cite on its own.
The two systems do not talk to each other. Retrieval scores your page against the prompt and decides whether to cite it, the same way it always has. The ad auction scores a bid against the same prompt and decides whose product card to show. One pipeline runs on content and trust signals. The other runs on budget and targeting. A brand with a strong retrieval footprint and no ad spend can still win the organic citation. A brand with a large ad budget and thin content can still lose it.
What this means for your organic answer strategy
Before you build a plan around earning organic citations in ChatGPT shopping answers, check how much of your category already looks like the table above.
Run ten prompts a real buyer in your category would type into ChatGPT this week. For each one, note two things: whether a sponsored block appears, and whether your brand shows up in the organic citations sitting next to it. If your category’s sponsored rate tracks close to the 76.4% national average, you are optimizing for a slot that already shares the screen with a paid one in most answers.
That is not a reason to skip organic work. The organic citation is still the only slot in a ChatGPT answer you cannot simply buy, and it is the slot most other answer engines, including Perplexity, Gemini, Google AI Overviews, and Microsoft Copilot, run entirely on. It is a reason to know which slot you are actually competing for before you spend a quarter chasing it.
How to track which slot you are winning
Otterly.AI is the source of the shopping-prompt figure in this post, which makes it a reasonable starting point for watching that number move in your own category. It tracks prompt-level citations and reports on sponsored-placement density alongside them.
For deeper attribution across a wider engine set, Profound maps citations across nine or more AI engines with visual citation maps, useful for teams that need to separate organic performance from paid activity at scale.
Temso covers the organic side of this same question across eight AI engines, from $89/mo, and pairs the citation tracking with a prioritized fix queue so a lost organic citation turns into a content or schema fix rather than just a number on a dashboard. Unlimited projects, users, and recommendations run on every plan.
All three tools measure the organic slot. None of them can move the sponsored one for you, because that one runs through OpenAI’s ad system, not through content or citations.
For the broader framework these terms sit inside, the AEO pillar guide covers how each major engine decides what to cite. Unfamiliar terms are in the glossary, the ranked comparison of AEO tools is at /rankings/aeo-tools, and the scoring criteria behind that ranking are at /methodology.
Audit your category before you plan the quarter
Run five ChatGPT shopping prompts in your category today. Count how many already carry a sponsored block, and note whether your brand appears in the organic citations beside it.
Temso tracks that organic citation, and the competitors already winning it, across eight AI engines from $89/mo, with a guided setup that takes about five minutes.